add updated_at as an option

### Changes 🏗️

### Before

<img width="800" height="649" alt="Screenshot_2025-10-23_at_00 44 59"
src="https://github.com/user-attachments/assets/fd717d39-772a-4331-bc54-4db15a9a3107"
/>

### After

<img width="800" height="555" alt="Screenshot 2025-10-27 at 23 19 10"
src="https://github.com/user-attachments/assets/64878bd0-3a96-4b3a-8344-1a88c89de52e"
/>

### Checklist 📋

#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
  - [x] Try to signup with a non-approved email
  - [x] You see the modal with an updated copy

.github/copilot-instructions.md

@@ -12,6 +12,7 @@ This file provides comprehensive onboarding information for GitHub Copilot codin
 - **Infrastructure** - Docker configurations, CI/CD, and development tools
 
 **Primary Languages & Frameworks:**
+
 - **Backend**: Python 3.10-3.13, FastAPI, Prisma ORM, PostgreSQL, RabbitMQ
 - **Frontend**: TypeScript, Next.js 15, React, Tailwind CSS, Radix UI
 - **Development**: Docker, Poetry, pnpm, Playwright, Storybook
@@ -23,15 +24,17 @@ This file provides comprehensive onboarding information for GitHub Copilot codin
 **Always run these commands in the correct directory and in this order:**
 
 1. **Initial Setup** (required once):
+
    ```bash
    # Clone and enter repository
    git clone <repo> && cd AutoGPT
-   
+
    # Start all services (database, redis, rabbitmq, clamav)
    cd autogpt_platform && docker compose --profile local up deps --build --detach
    ```
 
 2. **Backend Setup** (always run before backend development):
+
    ```bash
    cd autogpt_platform/backend
    poetry install                    # Install dependencies
@@ -48,6 +51,7 @@ This file provides comprehensive onboarding information for GitHub Copilot codin
 ### Runtime Requirements
 
 **Critical:** Always ensure Docker services are running before starting development:
+
 ```bash
 cd autogpt_platform && docker compose --profile local up deps --build --detach
 ```
@@ -58,6 +62,7 @@ cd autogpt_platform && docker compose --profile local up deps --build --detach
 ### Development Commands
 
 **Backend Development:**
+
 ```bash
 cd autogpt_platform/backend
 poetry run serve                     # Start development server (port 8000)
@@ -68,6 +73,7 @@ poetry run lint                      # Lint code (ruff) - run after format
 ```
 
 **Frontend Development:**
+
 ```bash
 cd autogpt_platform/frontend
 pnpm dev                            # Start development server (port 3000) - use for active development
@@ -81,23 +87,27 @@ pnpm storybook                      # Start component development server
 ### Testing Strategy
 
 **Backend Tests:**
+
 - **Block Tests**: `poetry run pytest backend/blocks/test/test_block.py -xvs` (validates all blocks)
 - **Specific Block**: `poetry run pytest 'backend/blocks/test/test_block.py::test_available_blocks[BlockName]' -xvs`
 - **Snapshot Tests**: Use `--snapshot-update` when output changes, always review with `git diff`
 
 **Frontend Tests:**
+
 - **E2E Tests**: Always run `pnpm dev` before `pnpm test` (Playwright requires running instance)
 - **Component Tests**: Use Storybook for isolated component development
 
 ### Critical Validation Steps
 
 **Before committing changes:**
+
 1. Run `poetry run format` (backend) and `pnpm format` (frontend)
 2. Ensure all tests pass in modified areas
 3. Verify Docker services are still running
 4. Check that database migrations apply cleanly
 
 **Common Issues & Workarounds:**
+
 - **Prisma issues**: Run `poetry run prisma generate` after schema changes
 - **Permission errors**: Ensure Docker has proper permissions
 - **Port conflicts**: Check the `docker-compose.yml` file for the current list of exposed ports. You can list all mapped ports with:
@@ -108,6 +118,7 @@ pnpm storybook                      # Start component development server
 ### Core Architecture
 
 **AutoGPT Platform** (`autogpt_platform/`):
+
 - `backend/` - FastAPI server with async support
   - `backend/backend/` - Core API logic
   - `backend/blocks/` - Agent execution blocks
@@ -121,6 +132,7 @@ pnpm storybook                      # Start component development server
 - `docker-compose.yml` - Development stack orchestration
 
 **Key Configuration Files:**
+
 - `pyproject.toml` - Python dependencies and tooling
 - `package.json` - Node.js dependencies and scripts
 - `schema.prisma` - Database schema and migrations
@@ -136,6 +148,7 @@ pnpm storybook                      # Start component development server
 ### Development Workflow
 
 **GitHub Actions**: Multiple CI/CD workflows in `.github/workflows/`
+
 - `platform-backend-ci.yml` - Backend testing and validation
 - `platform-frontend-ci.yml` - Frontend testing and validation
 - `platform-fullstack-ci.yml` - End-to-end integration tests
@@ -146,11 +159,13 @@ pnpm storybook                      # Start component development server
 ### Key Source Files
 
 **Backend Entry Points:**
+
 - `backend/backend/server/server.py` - FastAPI application setup
 - `backend/backend/data/` - Database models and user management
 - `backend/blocks/` - Agent execution blocks and logic
 
 **Frontend Entry Points:**
+
 - `frontend/src/app/layout.tsx` - Root application layout
 - `frontend/src/app/page.tsx` - Home page
 - `frontend/src/lib/supabase/` - Authentication and database client
@@ -160,6 +175,7 @@ pnpm storybook                      # Start component development server
 ### Agent Block System
 
 Agents are built using a visual block-based system where each block performs a single action. Blocks are defined in `backend/blocks/` and must include:
+
 - Block definition with input/output schemas
 - Execution logic with proper error handling
 - Tests validating functionality
@@ -167,6 +183,7 @@ Agents are built using a visual block-based system where each block performs a s
 ### Database & ORM
 
 **Prisma ORM** with PostgreSQL backend including pgvector for embeddings:
+
 - Schema in `schema.prisma`
 - Migrations in `backend/migrations/`
 - Always run `prisma migrate dev` and `prisma generate` after schema changes
@@ -174,13 +191,15 @@ Agents are built using a visual block-based system where each block performs a s
 ## Environment Configuration
 
 ### Configuration Files Priority Order
+
 1. **Backend**: `/backend/.env.default` → `/backend/.env` (user overrides)
-2. **Frontend**: `/frontend/.env.default` → `/frontend/.env` (user overrides)  
+2. **Frontend**: `/frontend/.env.default` → `/frontend/.env` (user overrides)
 3. **Platform**: `/.env.default` (Supabase/shared) → `/.env` (user overrides)
 4. Docker Compose `environment:` sections override file-based config
 5. Shell environment variables have highest precedence
 
 ### Docker Environment Setup
+
 - All services use hardcoded defaults (no `${VARIABLE}` substitutions)
 - The `env_file` directive loads variables INTO containers at runtime
 - Backend/Frontend services use YAML anchors for consistent configuration
@@ -189,6 +208,7 @@ Agents are built using a visual block-based system where each block performs a s
 ## Advanced Development Patterns
 
 ### Adding New Blocks
+
 1. Create file in `/backend/backend/blocks/`
 2. Inherit from `Block` base class with input/output schemas
 3. Implement `run` method with proper error handling
@@ -198,6 +218,7 @@ Agents are built using a visual block-based system where each block performs a s
 7. Consider how inputs/outputs connect with other blocks in graph editor
 
 ### API Development
+
 1. Update routes in `/backend/backend/server/routers/`
 2. Add/update Pydantic models in same directory
 3. Write tests alongside route files
@@ -205,21 +226,76 @@ Agents are built using a visual block-based system where each block performs a s
 5. Run `poetry run test` to verify changes
 
 ### Frontend Development
-1. Components in `/frontend/src/components/`
-2. Use existing UI components from `/frontend/src/components/ui/`
-3. Add Storybook stories for component development
-4. Test user-facing features with Playwright E2E tests
-5. Update protected routes in middleware when needed
+
+**📖 Complete Frontend Guide**: See `autogpt_platform/frontend/CONTRIBUTING.md` and `autogpt_platform/frontend/.cursorrules` for comprehensive patterns and conventions.
+
+**Quick Reference:**
+
+**Component Structure:**
+
+- Separate render logic from data/behavior
+- Structure: `ComponentName/ComponentName.tsx` + `useComponentName.ts` + `helpers.ts`
+- Exception: Small components (3-4 lines of logic) can be inline
+- Render-only components can be direct files without folders
+
+**Data Fetching:**
+
+- Use generated API hooks from `@/app/api/__generated__/endpoints/`
+- Generated via Orval from backend OpenAPI spec
+- Pattern: `use{Method}{Version}{OperationName}`
+- Example: `useGetV2ListLibraryAgents`
+- Regenerate with: `pnpm generate:api`
+- **Never** use deprecated `BackendAPI` or `src/lib/autogpt-server-api/*`
+
+**Code Conventions:**
+
+- Use function declarations for components and handlers (not arrow functions)
+- Only arrow functions for small inline lambdas (map, filter, etc.)
+- Components: `PascalCase`, Hooks: `camelCase` with `use` prefix
+- No barrel files or `index.ts` re-exports
+- Minimal comments (code should be self-documenting)
+
+**Styling:**
+
+- Use Tailwind CSS utilities only
+- Use design system components from `src/components/` (atoms, molecules, organisms)
+- Never use `src/components/__legacy__/*`
+- Only use Phosphor Icons (`@phosphor-icons/react`)
+- Prefer design tokens over hardcoded values
+
+**Error Handling:**
+
+- Render errors: Use `<ErrorCard />` component
+- Mutation errors: Display with toast notifications
+- Manual exceptions: Use `Sentry.captureException()`
+- Global error boundaries already configured
+
+**Testing:**
+
+- Add/update Storybook stories for UI components (`pnpm storybook`)
+- Run Playwright E2E tests with `pnpm test`
+- Verify in Chromatic after PR
+
+**Architecture:**
+
+- Default to client components ("use client")
+- Server components only for SEO or extreme TTFB needs
+- Use React Query for server state (via generated hooks)
+- Co-locate UI state in components/hooks
 
 ### Security Guidelines
+
 **Cache Protection Middleware** (`/backend/backend/server/middleware/security.py`):
+
 - Default: Disables caching for ALL endpoints with `Cache-Control: no-store, no-cache, must-revalidate, private`
 - Uses allow list approach for cacheable paths (static assets, health checks, public pages)
 - Prevents sensitive data caching in browsers/proxies
 - Add new cacheable endpoints to `CACHEABLE_PATHS`
 
 ### CI/CD Alignment
+
 The repository has comprehensive CI workflows that test:
+
 - **Backend**: Python 3.11-3.13, services (Redis/RabbitMQ/ClamAV), Prisma migrations, Poetry lock validation
 - **Frontend**: Node.js 21, pnpm, Playwright with Docker Compose stack, API schema validation
 - **Integration**: Full-stack type checking and E2E testing
@@ -229,6 +305,7 @@ Match these patterns when developing locally - the copilot setup environment mir
 ## Collaboration with Other AI Assistants
 
 This repository is actively developed with assistance from Claude (via CLAUDE.md files). When working on this codebase:
+
 - Check for existing CLAUDE.md files that provide additional context
 - Follow established patterns and conventions already in the codebase
 - Maintain consistency with existing code style and architecture
@@ -237,8 +314,9 @@ This repository is actively developed with assistance from Claude (via CLAUDE.md
 ## Trust These Instructions
 
 These instructions are comprehensive and tested. Only perform additional searches if:
+
 1. Information here is incomplete for your specific task
 2. You encounter errors not covered by the workarounds
 3. You need to understand implementation details not covered above
 
-For detailed platform development patterns, refer to `autogpt_platform/CLAUDE.md` and `AGENTS.md` in the repository root.
+For detailed platform development patterns, refer to `autogpt_platform/CLAUDE.md` and `AGENTS.md` in the repository root.

.pr_agent.toml

@@ -1,6 +1,3 @@
-[pr_reviewer]
-num_code_suggestions=0
-
 [pr_code_suggestions]
 commitable_code_suggestions=false
 num_code_suggestions=0

autogpt_platform/CLAUDE.md

@@ -63,6 +63,9 @@ poetry run pytest path/to/test.py --snapshot-update
 # Install dependencies
 cd frontend && pnpm i
 
+# Generate API client from OpenAPI spec
+pnpm generate:api
+
 # Start development server
 pnpm dev
 
@@ -75,12 +78,23 @@ pnpm storybook
 # Build production
 pnpm build
 
+# Format and lint
+pnpm format
+
 # Type checking
 pnpm types
 ```
 
-We have a components library in autogpt_platform/frontend/src/components/atoms that should be used when adding new pages and components. 
+**📖 Complete Guide**: See `/frontend/CONTRIBUTING.md` and `/frontend/.cursorrules` for comprehensive frontend patterns.
 
+**Key Frontend Conventions:**
+
+- Separate render logic from data/behavior in components
+- Use generated API hooks from `@/app/api/__generated__/endpoints/`
+- Use function declarations (not arrow functions) for components/handlers
+- Use design system components from `src/components/` (atoms, molecules, organisms)
+- Only use Phosphor Icons
+- Never use `src/components/__legacy__/*` or deprecated `BackendAPI`
 
 ## Architecture Overview
 
@@ -95,11 +109,16 @@ We have a components library in autogpt_platform/frontend/src/components/atoms t
 
 ### Frontend Architecture
 
-- **Framework**: Next.js App Router with React Server Components
-- **State Management**: React hooks + Supabase client for real-time updates
+- **Framework**: Next.js 15 App Router (client-first approach)
+- **Data Fetching**: Type-safe generated API hooks via Orval + React Query
+- **State Management**: React Query for server state, co-located UI state in components/hooks
+- **Component Structure**: Separate render logic (`.tsx`) from business logic (`use*.ts` hooks)
 - **Workflow Builder**: Visual graph editor using @xyflow/react
-- **UI Components**: Radix UI primitives with Tailwind CSS styling
+- **UI Components**: shadcn/ui (Radix UI primitives) with Tailwind CSS styling
+- **Icons**: Phosphor Icons only
 - **Feature Flags**: LaunchDarkly integration
+- **Error Handling**: ErrorCard for render errors, toast for mutations, Sentry for exceptions
+- **Testing**: Playwright for E2E, Storybook for component development
 
 ### Key Concepts
 
@@ -153,6 +172,7 @@ Key models (defined in `/backend/schema.prisma`):
 **Adding a new block:**
 
 Follow the comprehensive [Block SDK Guide](../../../docs/content/platform/block-sdk-guide.md) which covers:
+
 - Provider configuration with `ProviderBuilder`
 - Block schema definition
 - Authentication (API keys, OAuth, webhooks)
@@ -160,6 +180,7 @@ Follow the comprehensive [Block SDK Guide](../../../docs/content/platform/block-
 - File organization
 
 Quick steps:
+
 1. Create new file in `/backend/backend/blocks/`
 2. Configure provider using `ProviderBuilder` in `_config.py`
 3. Inherit from `Block` base class
@@ -180,10 +201,20 @@ ex: do the inputs and outputs tie well together?
 
 **Frontend feature development:**
 
-1. Components go in `/frontend/src/components/`
-2. Use existing UI components from `/frontend/src/components/ui/`
-3. Add Storybook stories for new components
-4. Test with Playwright if user-facing
+See `/frontend/CONTRIBUTING.md` for complete patterns. Quick reference:
+
+1. **Pages**: Create in `src/app/(platform)/feature-name/page.tsx`
+   - Add `usePageName.ts` hook for logic
+   - Put sub-components in local `components/` folder
+2. **Components**: Structure as `ComponentName/ComponentName.tsx` + `useComponentName.ts` + `helpers.ts`
+   - Use design system components from `src/components/` (atoms, molecules, organisms)
+   - Never use `src/components/__legacy__/*`
+3. **Data fetching**: Use generated API hooks from `@/app/api/__generated__/endpoints/`
+   - Regenerate with `pnpm generate:api`
+   - Pattern: `use{Method}{Version}{OperationName}`
+4. **Styling**: Tailwind CSS only, use design tokens, Phosphor Icons only
+5. **Testing**: Add Storybook stories for new components, Playwright for E2E
+6. **Code conventions**: Function declarations (not arrow functions) for components/handlers
 
 ### Security Implementation
 

autogpt_platform/Makefile

@@ -0,0 +1,57 @@
+.PHONY: start-core stop-core logs-core format lint migrate run-backend run-frontend
+
+# Run just Supabase + Redis + RabbitMQ
+start-core:
+	docker compose up -d deps
+
+# Stop core services
+stop-core:
+	docker compose stop deps
+
+reset-db:
+	rm -rf db/docker/volumes/db/data
+	cd backend && poetry run prisma migrate deploy
+	cd backend && poetry run prisma generate
+	
+# View logs for core services
+logs-core:
+	docker compose logs -f deps
+
+# Run formatting and linting for backend and frontend
+format:
+	cd backend && poetry run format
+	cd frontend && pnpm format
+	cd frontend && pnpm lint
+
+init-env:
+	cp -n .env.default .env || true
+	cd backend && cp -n .env.default .env || true
+	cd frontend && cp -n .env.default .env || true
+
+
+# Run migrations for backend
+migrate:
+	cd backend && poetry run prisma migrate deploy
+	cd backend && poetry run prisma generate
+
+run-backend:
+	cd backend && poetry run app
+
+run-frontend:
+	cd frontend && pnpm dev
+
+test-data:
+	cd backend && poetry run python test/test_data_creator.py
+	
+help:
+	@echo "Usage: make <target>"
+	@echo "Targets:"
+	@echo "  start-core - Start just the core services (Supabase, Redis, RabbitMQ) in background"
+	@echo "  stop-core - Stop the core services"
+	@echo "  reset-db - Reset the database by deleting the volume"
+	@echo "  logs-core - Tail the logs for core services"
+	@echo "  format - Format & lint backend (Python) and frontend (TypeScript) code"
+	@echo "  migrate - Run backend database migrations"
+	@echo "  run-backend - Run the backend FastAPI server"
+	@echo "  run-frontend - Run the frontend Next.js development server"
+	@echo "  test-data - Run the test data creator"

autogpt_platform/README.md

@@ -38,6 +38,37 @@ To run the AutoGPT Platform, follow these steps:
 
 4. After all the services are in ready state, open your browser and navigate to `http://localhost:3000` to access the AutoGPT Platform frontend.
 
+### Running Just Core services
+
+You can now run the following to enable just the core services.
+
+```
+# For help
+make help
+
+# Run just Supabase + Redis + RabbitMQ
+make start-core
+
+# Stop core services
+make stop-core
+
+# View logs from core services 
+make logs-core
+
+# Run formatting and linting for backend and frontend
+make format
+
+# Run migrations for backend database
+make migrate
+
+# Run backend server
+make run-backend
+
+# Run frontend development server
+make run-frontend
+
+```
+
 ### Docker Compose Commands
 
 Here are some useful Docker Compose commands for managing your AutoGPT Platform:

autogpt_platform/autogpt_libs/autogpt_libs/auth/dependencies.py

@@ -10,7 +10,7 @@ from .jwt_utils import get_jwt_payload, verify_user
 from .models import User
 
 
-def requires_user(jwt_payload: dict = fastapi.Security(get_jwt_payload)) -> User:
+async def requires_user(jwt_payload: dict = fastapi.Security(get_jwt_payload)) -> User:
     """
     FastAPI dependency that requires a valid authenticated user.
 
@@ -20,7 +20,9 @@ def requires_user(jwt_payload: dict = fastapi.Security(get_jwt_payload)) -> User
     return verify_user(jwt_payload, admin_only=False)
 
 
-def requires_admin_user(jwt_payload: dict = fastapi.Security(get_jwt_payload)) -> User:
+async def requires_admin_user(
+    jwt_payload: dict = fastapi.Security(get_jwt_payload),
+) -> User:
     """
     FastAPI dependency that requires a valid admin user.
 
@@ -30,7 +32,7 @@ def requires_admin_user(jwt_payload: dict = fastapi.Security(get_jwt_payload)) -
     return verify_user(jwt_payload, admin_only=True)
 
 
-def get_user_id(jwt_payload: dict = fastapi.Security(get_jwt_payload)) -> str:
+async def get_user_id(jwt_payload: dict = fastapi.Security(get_jwt_payload)) -> str:
     """
     FastAPI dependency that returns the ID of the authenticated user.
 

autogpt_platform/autogpt_libs/autogpt_libs/auth/dependencies_test.py

@@ -45,7 +45,7 @@ class TestAuthDependencies:
         """Create a test client."""
         return TestClient(app)
 
-    def test_requires_user_with_valid_jwt_payload(self, mocker: MockerFixture):
+    async def test_requires_user_with_valid_jwt_payload(self, mocker: MockerFixture):
         """Test requires_user with valid JWT payload."""
         jwt_payload = {"sub": "user-123", "role": "user", "email": "user@example.com"}
 
@@ -53,12 +53,12 @@ class TestAuthDependencies:
         mocker.patch(
             "autogpt_libs.auth.dependencies.get_jwt_payload", return_value=jwt_payload
         )
-        user = requires_user(jwt_payload)
+        user = await requires_user(jwt_payload)
         assert isinstance(user, User)
         assert user.user_id == "user-123"
         assert user.role == "user"
 
-    def test_requires_user_with_admin_jwt_payload(self, mocker: MockerFixture):
+    async def test_requires_user_with_admin_jwt_payload(self, mocker: MockerFixture):
         """Test requires_user accepts admin users."""
         jwt_payload = {
             "sub": "admin-456",
@@ -69,28 +69,28 @@ class TestAuthDependencies:
         mocker.patch(
             "autogpt_libs.auth.dependencies.get_jwt_payload", return_value=jwt_payload
         )
-        user = requires_user(jwt_payload)
+        user = await requires_user(jwt_payload)
         assert user.user_id == "admin-456"
         assert user.role == "admin"
 
-    def test_requires_user_missing_sub(self):
+    async def test_requires_user_missing_sub(self):
         """Test requires_user with missing user ID."""
         jwt_payload = {"role": "user", "email": "user@example.com"}
 
         with pytest.raises(HTTPException) as exc_info:
-            requires_user(jwt_payload)
+            await requires_user(jwt_payload)
         assert exc_info.value.status_code == 401
         assert "User ID not found" in exc_info.value.detail
 
-    def test_requires_user_empty_sub(self):
+    async def test_requires_user_empty_sub(self):
         """Test requires_user with empty user ID."""
         jwt_payload = {"sub": "", "role": "user"}
 
         with pytest.raises(HTTPException) as exc_info:
-            requires_user(jwt_payload)
+            await requires_user(jwt_payload)
         assert exc_info.value.status_code == 401
 
-    def test_requires_admin_user_with_admin(self, mocker: MockerFixture):
+    async def test_requires_admin_user_with_admin(self, mocker: MockerFixture):
         """Test requires_admin_user with admin role."""
         jwt_payload = {
             "sub": "admin-789",
@@ -101,51 +101,51 @@ class TestAuthDependencies:
         mocker.patch(
             "autogpt_libs.auth.dependencies.get_jwt_payload", return_value=jwt_payload
         )
-        user = requires_admin_user(jwt_payload)
+        user = await requires_admin_user(jwt_payload)
         assert user.user_id == "admin-789"
         assert user.role == "admin"
 
-    def test_requires_admin_user_with_regular_user(self):
+    async def test_requires_admin_user_with_regular_user(self):
         """Test requires_admin_user rejects regular users."""
         jwt_payload = {"sub": "user-123", "role": "user", "email": "user@example.com"}
 
         with pytest.raises(HTTPException) as exc_info:
-            requires_admin_user(jwt_payload)
+            await requires_admin_user(jwt_payload)
         assert exc_info.value.status_code == 403
         assert "Admin access required" in exc_info.value.detail
 
-    def test_requires_admin_user_missing_role(self):
+    async def test_requires_admin_user_missing_role(self):
         """Test requires_admin_user with missing role."""
         jwt_payload = {"sub": "user-123", "email": "user@example.com"}
 
         with pytest.raises(KeyError):
-            requires_admin_user(jwt_payload)
+            await requires_admin_user(jwt_payload)
 
-    def test_get_user_id_with_valid_payload(self, mocker: MockerFixture):
+    async def test_get_user_id_with_valid_payload(self, mocker: MockerFixture):
         """Test get_user_id extracts user ID correctly."""
         jwt_payload = {"sub": "user-id-xyz", "role": "user"}
 
         mocker.patch(
             "autogpt_libs.auth.dependencies.get_jwt_payload", return_value=jwt_payload
         )
-        user_id = get_user_id(jwt_payload)
+        user_id = await get_user_id(jwt_payload)
         assert user_id == "user-id-xyz"
 
-    def test_get_user_id_missing_sub(self):
+    async def test_get_user_id_missing_sub(self):
         """Test get_user_id with missing user ID."""
         jwt_payload = {"role": "user"}
 
         with pytest.raises(HTTPException) as exc_info:
-            get_user_id(jwt_payload)
+            await get_user_id(jwt_payload)
         assert exc_info.value.status_code == 401
         assert "User ID not found" in exc_info.value.detail
 
-    def test_get_user_id_none_sub(self):
+    async def test_get_user_id_none_sub(self):
         """Test get_user_id with None user ID."""
         jwt_payload = {"sub": None, "role": "user"}
 
         with pytest.raises(HTTPException) as exc_info:
-            get_user_id(jwt_payload)
+            await get_user_id(jwt_payload)
         assert exc_info.value.status_code == 401
 
 
@@ -170,7 +170,7 @@ class TestAuthDependenciesIntegration:
 
         return _create_token
 
-    def test_endpoint_auth_enabled_no_token(self):
+    async def test_endpoint_auth_enabled_no_token(self):
         """Test endpoints require token when auth is enabled."""
         app = FastAPI()
 
@@ -184,7 +184,7 @@ class TestAuthDependenciesIntegration:
         response = client.get("/test")
         assert response.status_code == 401
 
-    def test_endpoint_with_valid_token(self, create_token):
+    async def test_endpoint_with_valid_token(self, create_token):
         """Test endpoint with valid JWT token."""
         app = FastAPI()
 
@@ -203,7 +203,7 @@ class TestAuthDependenciesIntegration:
         assert response.status_code == 200
         assert response.json()["user_id"] == "test-user"
 
-    def test_admin_endpoint_requires_admin_role(self, create_token):
+    async def test_admin_endpoint_requires_admin_role(self, create_token):
         """Test admin endpoint rejects non-admin users."""
         app = FastAPI()
 
@@ -240,7 +240,7 @@ class TestAuthDependenciesIntegration:
 class TestAuthDependenciesEdgeCases:
     """Edge case tests for authentication dependencies."""
 
-    def test_dependency_with_complex_payload(self):
+    async def test_dependency_with_complex_payload(self):
         """Test dependencies handle complex JWT payloads."""
         complex_payload = {
             "sub": "user-123",
@@ -256,14 +256,14 @@ class TestAuthDependenciesEdgeCases:
             "exp": 9999999999,
         }
 
-        user = requires_user(complex_payload)
+        user = await requires_user(complex_payload)
         assert user.user_id == "user-123"
         assert user.email == "test@example.com"
 
-        admin = requires_admin_user(complex_payload)
+        admin = await requires_admin_user(complex_payload)
         assert admin.role == "admin"
 
-    def test_dependency_with_unicode_in_payload(self):
+    async def test_dependency_with_unicode_in_payload(self):
         """Test dependencies handle unicode in JWT payloads."""
         unicode_payload = {
             "sub": "user-😀-123",
@@ -272,11 +272,11 @@ class TestAuthDependenciesEdgeCases:
             "name": "日本語",
         }
 
-        user = requires_user(unicode_payload)
+        user = await requires_user(unicode_payload)
         assert "😀" in user.user_id
         assert user.email == "测试@example.com"
 
-    def test_dependency_with_null_values(self):
+    async def test_dependency_with_null_values(self):
         """Test dependencies handle null values in payload."""
         null_payload = {
             "sub": "user-123",
@@ -286,18 +286,18 @@ class TestAuthDependenciesEdgeCases:
             "metadata": None,
         }
 
-        user = requires_user(null_payload)
+        user = await requires_user(null_payload)
         assert user.user_id == "user-123"
         assert user.email is None
 
-    def test_concurrent_requests_isolation(self):
+    async def test_concurrent_requests_isolation(self):
         """Test that concurrent requests don't interfere with each other."""
         payload1 = {"sub": "user-1", "role": "user"}
         payload2 = {"sub": "user-2", "role": "admin"}
 
         # Simulate concurrent processing
-        user1 = requires_user(payload1)
-        user2 = requires_admin_user(payload2)
+        user1 = await requires_user(payload1)
+        user2 = await requires_admin_user(payload2)
 
         assert user1.user_id == "user-1"
         assert user2.user_id == "user-2"
@@ -314,7 +314,7 @@ class TestAuthDependenciesEdgeCases:
             ({"sub": "user", "role": "user"}, "Admin access required", True),
         ],
     )
-    def test_dependency_error_cases(
+    async def test_dependency_error_cases(
         self, payload, expected_error: str, admin_only: bool
     ):
         """Test that errors propagate correctly through dependencies."""
@@ -325,7 +325,7 @@ class TestAuthDependenciesEdgeCases:
             verify_user(payload, admin_only=admin_only)
         assert expected_error in exc_info.value.detail
 
-    def test_dependency_valid_user(self):
+    async def test_dependency_valid_user(self):
         """Test valid user case for dependency."""
         # Import verify_user to test it directly since dependencies use FastAPI Security
         from autogpt_libs.auth.jwt_utils import verify_user

autogpt_platform/autogpt_libs/autogpt_libs/auth/jwt_utils.py

@@ -16,7 +16,7 @@ bearer_jwt_auth = HTTPBearer(
 )
 
 
-def get_jwt_payload(
+async def get_jwt_payload(
     credentials: HTTPAuthorizationCredentials | None = Security(bearer_jwt_auth),
 ) -> dict[str, Any]:
     """

autogpt_platform/autogpt_libs/autogpt_libs/auth/jwt_utils_test.py

@@ -116,32 +116,32 @@ def test_parse_jwt_token_missing_audience():
     assert "Invalid token" in str(exc_info.value)
 
 
-def test_get_jwt_payload_with_valid_token():
+async def test_get_jwt_payload_with_valid_token():
     """Test extracting JWT payload with valid bearer token."""
     token = create_token(TEST_USER_PAYLOAD)
     credentials = HTTPAuthorizationCredentials(scheme="Bearer", credentials=token)
 
-    result = jwt_utils.get_jwt_payload(credentials)
+    result = await jwt_utils.get_jwt_payload(credentials)
     assert result["sub"] == "test-user-id"
     assert result["role"] == "user"
 
 
-def test_get_jwt_payload_no_credentials():
+async def test_get_jwt_payload_no_credentials():
     """Test JWT payload when no credentials provided."""
     with pytest.raises(HTTPException) as exc_info:
-        jwt_utils.get_jwt_payload(None)
+        await jwt_utils.get_jwt_payload(None)
     assert exc_info.value.status_code == 401
     assert "Authorization header is missing" in exc_info.value.detail
 
 
-def test_get_jwt_payload_invalid_token():
+async def test_get_jwt_payload_invalid_token():
     """Test JWT payload extraction with invalid token."""
     credentials = HTTPAuthorizationCredentials(
         scheme="Bearer", credentials="invalid.token.here"
     )
 
     with pytest.raises(HTTPException) as exc_info:
-        jwt_utils.get_jwt_payload(credentials)
+        await jwt_utils.get_jwt_payload(credentials)
     assert exc_info.value.status_code == 401
     assert "Invalid token" in exc_info.value.detail
 

autogpt_platform/autogpt_libs/autogpt_libs/logging/config.py

@@ -4,6 +4,7 @@ import logging
 import os
 import socket
 import sys
+from logging.handlers import RotatingFileHandler
 from pathlib import Path
 
 from pydantic import Field, field_validator
@@ -93,42 +94,36 @@ def configure_logging(force_cloud_logging: bool = False) -> None:
     config = LoggingConfig()
     log_handlers: list[logging.Handler] = []
 
+    structured_logging = config.enable_cloud_logging or force_cloud_logging
+
     # Console output handlers
-    stdout = logging.StreamHandler(stream=sys.stdout)
-    stdout.setLevel(config.level)
-    stdout.addFilter(BelowLevelFilter(logging.WARNING))
-    if config.level == logging.DEBUG:
-        stdout.setFormatter(AGPTFormatter(DEBUG_LOG_FORMAT))
-    else:
-        stdout.setFormatter(AGPTFormatter(SIMPLE_LOG_FORMAT))
+    if not structured_logging:
+        stdout = logging.StreamHandler(stream=sys.stdout)
+        stdout.setLevel(config.level)
+        stdout.addFilter(BelowLevelFilter(logging.WARNING))
+        if config.level == logging.DEBUG:
+            stdout.setFormatter(AGPTFormatter(DEBUG_LOG_FORMAT))
+        else:
+            stdout.setFormatter(AGPTFormatter(SIMPLE_LOG_FORMAT))
 
-    stderr = logging.StreamHandler()
-    stderr.setLevel(logging.WARNING)
-    if config.level == logging.DEBUG:
-        stderr.setFormatter(AGPTFormatter(DEBUG_LOG_FORMAT))
-    else:
-        stderr.setFormatter(AGPTFormatter(SIMPLE_LOG_FORMAT))
+        stderr = logging.StreamHandler()
+        stderr.setLevel(logging.WARNING)
+        if config.level == logging.DEBUG:
+            stderr.setFormatter(AGPTFormatter(DEBUG_LOG_FORMAT))
+        else:
+            stderr.setFormatter(AGPTFormatter(SIMPLE_LOG_FORMAT))
 
-    log_handlers += [stdout, stderr]
+        log_handlers += [stdout, stderr]
 
     # Cloud logging setup
-    if config.enable_cloud_logging or force_cloud_logging:
-        import google.cloud.logging
-        from google.cloud.logging.handlers import CloudLoggingHandler
-        from google.cloud.logging_v2.handlers.transports import (
-            BackgroundThreadTransport,
-        )
+    else:
+        # Use Google Cloud Structured Log Handler. Log entries are printed to stdout
+        # in a JSON format which is automatically picked up by Google Cloud Logging.
+        from google.cloud.logging.handlers import StructuredLogHandler
 
-        client = google.cloud.logging.Client()
-        # Use BackgroundThreadTransport to prevent blocking the main thread
-        # and deadlocks when gRPC calls to Google Cloud Logging hang
-        cloud_handler = CloudLoggingHandler(
-            client,
-            name="autogpt_logs",
-            transport=BackgroundThreadTransport,
-        )
-        cloud_handler.setLevel(config.level)
-        log_handlers.append(cloud_handler)
+        structured_log_handler = StructuredLogHandler(stream=sys.stdout)
+        structured_log_handler.setLevel(config.level)
+        log_handlers.append(structured_log_handler)
 
     # File logging setup
     if config.enable_file_logging:
@@ -139,8 +134,13 @@ def configure_logging(force_cloud_logging: bool = False) -> None:
         print(f"Log directory: {config.log_dir}")
 
         # Activity log handler (INFO and above)
-        activity_log_handler = logging.FileHandler(
-            config.log_dir / LOG_FILE, "a", "utf-8"
+        # Security fix: Use RotatingFileHandler with size limits to prevent disk exhaustion
+        activity_log_handler = RotatingFileHandler(
+            config.log_dir / LOG_FILE,
+            mode="a",
+            encoding="utf-8",
+            maxBytes=10 * 1024 * 1024,  # 10MB per file
+            backupCount=3,  # Keep 3 backup files (40MB total)
         )
         activity_log_handler.setLevel(config.level)
         activity_log_handler.setFormatter(
@@ -150,8 +150,13 @@ def configure_logging(force_cloud_logging: bool = False) -> None:
 
         if config.level == logging.DEBUG:
             # Debug log handler (all levels)
-            debug_log_handler = logging.FileHandler(
-                config.log_dir / DEBUG_LOG_FILE, "a", "utf-8"
+            # Security fix: Use RotatingFileHandler with size limits
+            debug_log_handler = RotatingFileHandler(
+                config.log_dir / DEBUG_LOG_FILE,
+                mode="a",
+                encoding="utf-8",
+                maxBytes=10 * 1024 * 1024,  # 10MB per file
+                backupCount=3,  # Keep 3 backup files (40MB total)
             )
             debug_log_handler.setLevel(logging.DEBUG)
             debug_log_handler.setFormatter(
@@ -160,8 +165,13 @@ def configure_logging(force_cloud_logging: bool = False) -> None:
             log_handlers.append(debug_log_handler)
 
         # Error log handler (ERROR and above)
-        error_log_handler = logging.FileHandler(
-            config.log_dir / ERROR_LOG_FILE, "a", "utf-8"
+        # Security fix: Use RotatingFileHandler with size limits
+        error_log_handler = RotatingFileHandler(
+            config.log_dir / ERROR_LOG_FILE,
+            mode="a",
+            encoding="utf-8",
+            maxBytes=10 * 1024 * 1024,  # 10MB per file
+            backupCount=3,  # Keep 3 backup files (40MB total)
         )
         error_log_handler.setLevel(logging.ERROR)
         error_log_handler.setFormatter(AGPTFormatter(DEBUG_LOG_FORMAT, no_color=True))
@@ -169,7 +179,13 @@ def configure_logging(force_cloud_logging: bool = False) -> None:
 
     # Configure the root logger
     logging.basicConfig(
-        format=DEBUG_LOG_FORMAT if config.level == logging.DEBUG else SIMPLE_LOG_FORMAT,
+        format=(
+            "%(levelname)s  %(message)s"
+            if structured_logging
+            else (
+                DEBUG_LOG_FORMAT if config.level == logging.DEBUG else SIMPLE_LOG_FORMAT
+            )
+        ),
         level=config.level,
         handlers=log_handlers,
     )

autogpt_platform/autogpt_libs/autogpt_libs/utils/cache.py

@@ -1,328 +0,0 @@
-import asyncio
-import inspect
-import logging
-import threading
-import time
-from functools import wraps
-from typing import (
-    Any,
-    Callable,
-    ParamSpec,
-    Protocol,
-    TypeVar,
-    cast,
-    runtime_checkable,
-)
-
-P = ParamSpec("P")
-R = TypeVar("R")
-R_co = TypeVar("R_co", covariant=True)
-
-logger = logging.getLogger(__name__)
-
-
-def _make_hashable_key(
-    args: tuple[Any, ...], kwargs: dict[str, Any]
-) -> tuple[Any, ...]:
-    """
-    Convert args and kwargs into a hashable cache key.
-
-    Handles unhashable types like dict, list, set by converting them to
-    their sorted string representations.
-    """
-
-    def make_hashable(obj: Any) -> Any:
-        """Recursively convert an object to a hashable representation."""
-        if isinstance(obj, dict):
-            # Sort dict items to ensure consistent ordering
-            return (
-                "__dict__",
-                tuple(sorted((k, make_hashable(v)) for k, v in obj.items())),
-            )
-        elif isinstance(obj, (list, tuple)):
-            return ("__list__", tuple(make_hashable(item) for item in obj))
-        elif isinstance(obj, set):
-            return ("__set__", tuple(sorted(make_hashable(item) for item in obj)))
-        elif hasattr(obj, "__dict__"):
-            # Handle objects with __dict__ attribute
-            return ("__obj__", obj.__class__.__name__, make_hashable(obj.__dict__))
-        else:
-            # For basic hashable types (str, int, bool, None, etc.)
-            try:
-                hash(obj)
-                return obj
-            except TypeError:
-                # Fallback: convert to string representation
-                return ("__str__", str(obj))
-
-    hashable_args = tuple(make_hashable(arg) for arg in args)
-    hashable_kwargs = tuple(sorted((k, make_hashable(v)) for k, v in kwargs.items()))
-    return (hashable_args, hashable_kwargs)
-
-
-@runtime_checkable
-class CachedFunction(Protocol[P, R_co]):
-    """Protocol for cached functions with cache management methods."""
-
-    def cache_clear(self) -> None:
-        """Clear all cached entries."""
-        return None
-
-    def cache_info(self) -> dict[str, int | None]:
-        """Get cache statistics."""
-        return {}
-
-    def cache_delete(self, *args: P.args, **kwargs: P.kwargs) -> bool:
-        """Delete a specific cache entry by its arguments. Returns True if entry existed."""
-        return False
-
-    def __call__(self, *args: P.args, **kwargs: P.kwargs) -> R_co:
-        """Call the cached function."""
-        return None  # type: ignore
-
-
-def cached(
-    *,
-    maxsize: int = 128,
-    ttl_seconds: int | None = None,
-) -> Callable[[Callable], CachedFunction]:
-    """
-    Thundering herd safe cache decorator for both sync and async functions.
-
-    Uses double-checked locking to prevent multiple threads/coroutines from
-    executing the expensive operation simultaneously during cache misses.
-
-    Args:
-        func: The function to cache (when used without parentheses)
-        maxsize: Maximum number of cached entries
-        ttl_seconds: Time to live in seconds. If None, entries never expire
-
-    Returns:
-        Decorated function or decorator
-
-    Example:
-        @cache()  # Default: maxsize=128, no TTL
-        def expensive_sync_operation(param: str) -> dict:
-            return {"result": param}
-
-        @cache()  # Works with async too
-        async def expensive_async_operation(param: str) -> dict:
-            return {"result": param}
-
-        @cache(maxsize=1000, ttl_seconds=300)  # Custom maxsize and TTL
-        def another_operation(param: str) -> dict:
-            return {"result": param}
-    """
-
-    def decorator(target_func):
-        # Cache storage and locks
-        cache_storage = {}
-
-        if inspect.iscoroutinefunction(target_func):
-            # Async function with asyncio.Lock
-            cache_lock = asyncio.Lock()
-
-            @wraps(target_func)
-            async def async_wrapper(*args: P.args, **kwargs: P.kwargs):
-                key = _make_hashable_key(args, kwargs)
-                current_time = time.time()
-
-                # Fast path: check cache without lock
-                if key in cache_storage:
-                    if ttl_seconds is None:
-                        logger.debug(f"Cache hit for {target_func.__name__}")
-                        return cache_storage[key]
-                    else:
-                        cached_data = cache_storage[key]
-                        if isinstance(cached_data, tuple):
-                            result, timestamp = cached_data
-                            if current_time - timestamp < ttl_seconds:
-                                logger.debug(f"Cache hit for {target_func.__name__}")
-                                return result
-
-                # Slow path: acquire lock for cache miss/expiry
-                async with cache_lock:
-                    # Double-check: another coroutine might have populated cache
-                    if key in cache_storage:
-                        if ttl_seconds is None:
-                            return cache_storage[key]
-                        else:
-                            cached_data = cache_storage[key]
-                            if isinstance(cached_data, tuple):
-                                result, timestamp = cached_data
-                                if current_time - timestamp < ttl_seconds:
-                                    return result
-
-                    # Cache miss - execute function
-                    logger.debug(f"Cache miss for {target_func.__name__}")
-                    result = await target_func(*args, **kwargs)
-
-                    # Store result
-                    if ttl_seconds is None:
-                        cache_storage[key] = result
-                    else:
-                        cache_storage[key] = (result, current_time)
-
-                    # Cleanup if needed
-                    if len(cache_storage) > maxsize:
-                        cutoff = maxsize // 2
-                        oldest_keys = (
-                            list(cache_storage.keys())[:-cutoff] if cutoff > 0 else []
-                        )
-                        for old_key in oldest_keys:
-                            cache_storage.pop(old_key, None)
-
-                    return result
-
-            wrapper = async_wrapper
-
-        else:
-            # Sync function with threading.Lock
-            cache_lock = threading.Lock()
-
-            @wraps(target_func)
-            def sync_wrapper(*args: P.args, **kwargs: P.kwargs):
-                key = _make_hashable_key(args, kwargs)
-                current_time = time.time()
-
-                # Fast path: check cache without lock
-                if key in cache_storage:
-                    if ttl_seconds is None:
-                        logger.debug(f"Cache hit for {target_func.__name__}")
-                        return cache_storage[key]
-                    else:
-                        cached_data = cache_storage[key]
-                        if isinstance(cached_data, tuple):
-                            result, timestamp = cached_data
-                            if current_time - timestamp < ttl_seconds:
-                                logger.debug(f"Cache hit for {target_func.__name__}")
-                                return result
-
-                # Slow path: acquire lock for cache miss/expiry
-                with cache_lock:
-                    # Double-check: another thread might have populated cache
-                    if key in cache_storage:
-                        if ttl_seconds is None:
-                            return cache_storage[key]
-                        else:
-                            cached_data = cache_storage[key]
-                            if isinstance(cached_data, tuple):
-                                result, timestamp = cached_data
-                                if current_time - timestamp < ttl_seconds:
-                                    return result
-
-                    # Cache miss - execute function
-                    logger.debug(f"Cache miss for {target_func.__name__}")
-                    result = target_func(*args, **kwargs)
-
-                    # Store result
-                    if ttl_seconds is None:
-                        cache_storage[key] = result
-                    else:
-                        cache_storage[key] = (result, current_time)
-
-                    # Cleanup if needed
-                    if len(cache_storage) > maxsize:
-                        cutoff = maxsize // 2
-                        oldest_keys = (
-                            list(cache_storage.keys())[:-cutoff] if cutoff > 0 else []
-                        )
-                        for old_key in oldest_keys:
-                            cache_storage.pop(old_key, None)
-
-                    return result
-
-            wrapper = sync_wrapper
-
-        # Add cache management methods
-        def cache_clear() -> None:
-            cache_storage.clear()
-
-        def cache_info() -> dict[str, int | None]:
-            return {
-                "size": len(cache_storage),
-                "maxsize": maxsize,
-                "ttl_seconds": ttl_seconds,
-            }
-
-        def cache_delete(*args, **kwargs) -> bool:
-            """Delete a specific cache entry. Returns True if entry existed."""
-            key = _make_hashable_key(args, kwargs)
-            if key in cache_storage:
-                del cache_storage[key]
-                return True
-            return False
-
-        setattr(wrapper, "cache_clear", cache_clear)
-        setattr(wrapper, "cache_info", cache_info)
-        setattr(wrapper, "cache_delete", cache_delete)
-
-        return cast(CachedFunction, wrapper)
-
-    return decorator
-
-
-def thread_cached(func):
-    """
-    Thread-local cache decorator for both sync and async functions.
-
-    Each thread gets its own cache, which is useful for request-scoped caching
-    in web applications where you want to cache within a single request but
-    not across requests.
-
-    Args:
-        func: The function to cache
-
-    Returns:
-        Decorated function with thread-local caching
-
-    Example:
-        @thread_cached
-        def expensive_operation(param: str) -> dict:
-            return {"result": param}
-
-        @thread_cached  # Works with async too
-        async def expensive_async_operation(param: str) -> dict:
-            return {"result": param}
-    """
-    thread_local = threading.local()
-
-    def _clear():
-        if hasattr(thread_local, "cache"):
-            del thread_local.cache
-
-    if inspect.iscoroutinefunction(func):
-
-        @wraps(func)
-        async def async_wrapper(*args, **kwargs):
-            cache = getattr(thread_local, "cache", None)
-            if cache is None:
-                cache = thread_local.cache = {}
-            key = _make_hashable_key(args, kwargs)
-            if key not in cache:
-                cache[key] = await func(*args, **kwargs)
-            return cache[key]
-
-        setattr(async_wrapper, "clear_cache", _clear)
-        return async_wrapper
-
-    else:
-
-        @wraps(func)
-        def sync_wrapper(*args, **kwargs):
-            cache = getattr(thread_local, "cache", None)
-            if cache is None:
-                cache = thread_local.cache = {}
-            key = _make_hashable_key(args, kwargs)
-            if key not in cache:
-                cache[key] = func(*args, **kwargs)
-            return cache[key]
-
-        setattr(sync_wrapper, "clear_cache", _clear)
-        return sync_wrapper
-
-
-def clear_thread_cache(func: Callable) -> None:
-    """Clear thread-local cache for a function."""
-    if clear := getattr(func, "clear_cache", None):
-        clear()

autogpt_platform/backend/Dockerfile

@@ -47,6 +47,7 @@ RUN poetry install --no-ansi --no-root
 
 # Generate Prisma client
 COPY autogpt_platform/backend/schema.prisma ./
+COPY autogpt_platform/backend/backend/data/partial_types.py ./backend/data/partial_types.py
 RUN poetry run prisma generate
 
 FROM debian:13-slim AS server_dependencies
@@ -92,6 +93,7 @@ FROM server_dependencies AS migrate
 
 # Migration stage only needs schema and migrations - much lighter than full backend
 COPY autogpt_platform/backend/schema.prisma /app/autogpt_platform/backend/
+COPY autogpt_platform/backend/backend/data/partial_types.py /app/autogpt_platform/backend/backend/data/partial_types.py
 COPY autogpt_platform/backend/migrations /app/autogpt_platform/backend/migrations
 
 FROM server_dependencies AS server

autogpt_platform/backend/backend/blocks/__init__.py

@@ -5,7 +5,7 @@ import re
 from pathlib import Path
 from typing import TYPE_CHECKING, TypeVar
 
-from autogpt_libs.utils.cache import cached
+from backend.util.cache import cached
 
 logger = logging.getLogger(__name__)
 
@@ -16,7 +16,7 @@ if TYPE_CHECKING:
 T = TypeVar("T")
 
 
-@cached()
+@cached(ttl_seconds=3600)
 def load_all_blocks() -> dict[str, type["Block"]]:
     from backend.data.block import Block
     from backend.util.settings import Config

autogpt_platform/backend/backend/blocks/ai_condition.py

@@ -0,0 +1,214 @@
+from typing import Any
+
+from backend.blocks.llm import (
+    TEST_CREDENTIALS,
+    TEST_CREDENTIALS_INPUT,
+    AIBlockBase,
+    AICredentials,
+    AICredentialsField,
+    LlmModel,
+    LLMResponse,
+    llm_call,
+)
+from backend.data.block import BlockCategory, BlockOutput, BlockSchema
+from backend.data.model import APIKeyCredentials, NodeExecutionStats, SchemaField
+
+
+class AIConditionBlock(AIBlockBase):
+    """
+    An AI-powered condition block that uses natural language to evaluate conditions.
+
+    This block allows users to define conditions in plain English (e.g., "the input is an email address",
+    "the input is a city in the USA") and uses AI to determine if the input satisfies the condition.
+    It provides the same yes/no data pass-through functionality as the standard ConditionBlock.
+    """
+
+    class Input(BlockSchema):
+        input_value: Any = SchemaField(
+            description="The input value to evaluate with the AI condition",
+            placeholder="Enter the value to be evaluated (text, number, or any data)",
+        )
+        condition: str = SchemaField(
+            description="A plaintext English description of the condition to evaluate",
+            placeholder="E.g., 'the input is the body of an email', 'the input is a City in the USA', 'the input is an error or a refusal'",
+        )
+        yes_value: Any = SchemaField(
+            description="(Optional) Value to output if the condition is true. If not provided, input_value will be used.",
+            placeholder="Leave empty to use input_value, or enter a specific value",
+            default=None,
+        )
+        no_value: Any = SchemaField(
+            description="(Optional) Value to output if the condition is false. If not provided, input_value will be used.",
+            placeholder="Leave empty to use input_value, or enter a specific value",
+            default=None,
+        )
+        model: LlmModel = SchemaField(
+            title="LLM Model",
+            default=LlmModel.GPT4O,
+            description="The language model to use for evaluating the condition.",
+            advanced=False,
+        )
+        credentials: AICredentials = AICredentialsField()
+
+    class Output(BlockSchema):
+        result: bool = SchemaField(
+            description="The result of the AI condition evaluation (True or False)"
+        )
+        yes_output: Any = SchemaField(
+            description="The output value if the condition is true"
+        )
+        no_output: Any = SchemaField(
+            description="The output value if the condition is false"
+        )
+        error: str = SchemaField(
+            description="Error message if the AI evaluation is uncertain or fails"
+        )
+
+    def __init__(self):
+        super().__init__(
+            id="553ec5b8-6c45-4299-8d75-b394d05f72ff",
+            input_schema=AIConditionBlock.Input,
+            output_schema=AIConditionBlock.Output,
+            description="Uses AI to evaluate natural language conditions and provide conditional outputs",
+            categories={BlockCategory.AI, BlockCategory.LOGIC},
+            test_input={
+                "input_value": "john@example.com",
+                "condition": "the input is an email address",
+                "yes_value": "Valid email",
+                "no_value": "Not an email",
+                "model": LlmModel.GPT4O,
+                "credentials": TEST_CREDENTIALS_INPUT,
+            },
+            test_credentials=TEST_CREDENTIALS,
+            test_output=[
+                ("result", True),
+                ("yes_output", "Valid email"),
+            ],
+            test_mock={
+                "llm_call": lambda *args, **kwargs: LLMResponse(
+                    raw_response="",
+                    prompt=[],
+                    response="true",
+                    tool_calls=None,
+                    prompt_tokens=50,
+                    completion_tokens=10,
+                    reasoning=None,
+                )
+            },
+        )
+
+    async def llm_call(
+        self,
+        credentials: APIKeyCredentials,
+        llm_model: LlmModel,
+        prompt: list,
+        max_tokens: int,
+    ) -> LLMResponse:
+        """Wrapper method for llm_call to enable mocking in tests."""
+        return await llm_call(
+            credentials=credentials,
+            llm_model=llm_model,
+            prompt=prompt,
+            force_json_output=False,
+            max_tokens=max_tokens,
+        )
+
+    async def run(
+        self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs
+    ) -> BlockOutput:
+        """
+        Evaluate the AI condition and return appropriate outputs.
+        """
+        # Prepare the yes and no values, using input_value as default
+        yes_value = (
+            input_data.yes_value
+            if input_data.yes_value is not None
+            else input_data.input_value
+        )
+        no_value = (
+            input_data.no_value
+            if input_data.no_value is not None
+            else input_data.input_value
+        )
+
+        # Convert input_value to string for AI evaluation
+        input_str = str(input_data.input_value)
+
+        # Create the prompt for AI evaluation
+        prompt = [
+            {
+                "role": "system",
+                "content": (
+                    "You are an AI assistant that evaluates conditions based on input data. "
+                    "You must respond with only 'true' or 'false' (lowercase) to indicate whether "
+                    "the given condition is met by the input value. Be accurate and consider the "
+                    "context and meaning of both the input and the condition."
+                ),
+            },
+            {
+                "role": "user",
+                "content": (
+                    f"Input value: {input_str}\n"
+                    f"Condition to evaluate: {input_data.condition}\n\n"
+                    f"Does the input value satisfy the condition? Respond with only 'true' or 'false'."
+                ),
+            },
+        ]
+
+        # Call the LLM
+        try:
+            response = await self.llm_call(
+                credentials=credentials,
+                llm_model=input_data.model,
+                prompt=prompt,
+                max_tokens=10,  # We only expect a true/false response
+            )
+
+            # Extract the boolean result from the response
+            response_text = response.response.strip().lower()
+            if response_text == "true":
+                result = True
+            elif response_text == "false":
+                result = False
+            else:
+                # If the response is not clear, try to interpret it using word boundaries
+                import re
+
+                # Use word boundaries to avoid false positives like 'untrue' or '10'
+                tokens = set(re.findall(r"\b(true|false|yes|no|1|0)\b", response_text))
+
+                if tokens == {"true"} or tokens == {"yes"} or tokens == {"1"}:
+                    result = True
+                elif tokens == {"false"} or tokens == {"no"} or tokens == {"0"}:
+                    result = False
+                else:
+                    # Unclear or conflicting response - default to False and yield error
+                    result = False
+                    yield "error", f"Unclear AI response: '{response.response}'"
+
+            # Update internal stats
+            self.merge_stats(
+                NodeExecutionStats(
+                    input_token_count=response.prompt_tokens,
+                    output_token_count=response.completion_tokens,
+                )
+            )
+            self.prompt = response.prompt
+
+        except Exception as e:
+            # In case of any error, default to False to be safe
+            result = False
+            # Log the error but don't fail the block execution
+            import logging
+
+            logger = logging.getLogger(__name__)
+            logger.error(f"AI condition evaluation failed: {str(e)}")
+            yield "error", f"AI evaluation failed: {str(e)}"
+
+        # Yield results
+        yield "result", result
+
+        if result:
+            yield "yes_output", yes_value
+        else:
+            yield "no_output", no_value

autogpt_platform/backend/backend/blocks/code_executor.py

@@ -1,8 +1,10 @@
 from enum import Enum
-from typing import Literal
+from typing import Any, Literal, Optional
 
 from e2b_code_interpreter import AsyncSandbox
-from pydantic import SecretStr
+from e2b_code_interpreter import Result as E2BExecutionResult
+from e2b_code_interpreter.charts import Chart as E2BExecutionResultChart
+from pydantic import BaseModel, JsonValue, SecretStr
 
 from backend.data.block import Block, BlockCategory, BlockOutput, BlockSchema
 from backend.data.model import (
@@ -36,14 +38,135 @@ class ProgrammingLanguage(Enum):
     JAVA = "java"
 
 
-class CodeExecutionBlock(Block):
+class MainCodeExecutionResult(BaseModel):
+    """
+    *Pydantic model mirroring `e2b_code_interpreter.Result`*
+
+    Represents the data to be displayed as a result of executing a cell in a Jupyter notebook.
+    The result is similar to the structure returned by ipython kernel: https://ipython.readthedocs.io/en/stable/development/execution.html#execution-semantics
+
+    The result can contain multiple types of data, such as text, images, plots, etc. Each type of data is represented
+    as a string, and the result can contain multiple types of data. The display calls don't have to have text representation,
+    for the actual result the representation is always present for the result, the other representations are always optional.
+    """  # noqa
+
+    class Chart(BaseModel, E2BExecutionResultChart):
+        pass
+
+    text: Optional[str] = None
+    html: Optional[str] = None
+    markdown: Optional[str] = None
+    svg: Optional[str] = None
+    png: Optional[str] = None
+    jpeg: Optional[str] = None
+    pdf: Optional[str] = None
+    latex: Optional[str] = None
+    json: Optional[JsonValue] = None  # type: ignore (reportIncompatibleMethodOverride)
+    javascript: Optional[str] = None
+    data: Optional[dict] = None
+    chart: Optional[Chart] = None
+    extra: Optional[dict] = None
+    """Extra data that can be included. Not part of the standard types."""
+
+
+class CodeExecutionResult(MainCodeExecutionResult):
+    __doc__ = MainCodeExecutionResult.__doc__
+
+    is_main_result: bool = False
+    """Whether this data is the main result of the cell. Data can be produced by display calls of which can be multiple in a cell."""  # noqa
+
+
+class BaseE2BExecutorMixin:
+    """Shared implementation methods for E2B executor blocks."""
+
+    async def execute_code(
+        self,
+        api_key: str,
+        code: str,
+        language: ProgrammingLanguage,
+        template_id: str = "",
+        setup_commands: Optional[list[str]] = None,
+        timeout: Optional[int] = None,
+        sandbox_id: Optional[str] = None,
+        dispose_sandbox: bool = False,
+    ):
+        """
+        Unified code execution method that handles all three use cases:
+        1. Create new sandbox and execute (ExecuteCodeBlock)
+        2. Create new sandbox, execute, and return sandbox_id (InstantiateCodeSandboxBlock)
+        3. Connect to existing sandbox and execute (ExecuteCodeStepBlock)
+        """  # noqa
+        sandbox = None
+        try:
+            if sandbox_id:
+                # Connect to existing sandbox (ExecuteCodeStepBlock case)
+                sandbox = await AsyncSandbox.connect(
+                    sandbox_id=sandbox_id, api_key=api_key
+                )
+            else:
+                # Create new sandbox (ExecuteCodeBlock/InstantiateCodeSandboxBlock case)
+                sandbox = await AsyncSandbox.create(
+                    api_key=api_key, template=template_id, timeout=timeout
+                )
+                if setup_commands:
+                    for cmd in setup_commands:
+                        await sandbox.commands.run(cmd)
+
+            # Execute the code
+            execution = await sandbox.run_code(
+                code,
+                language=language.value,
+                on_error=lambda e: sandbox.kill(),  # Kill the sandbox on error
+            )
+
+            if execution.error:
+                raise Exception(execution.error)
+
+            results = execution.results
+            text_output = execution.text
+            stdout_logs = "".join(execution.logs.stdout)
+            stderr_logs = "".join(execution.logs.stderr)
+
+            return results, text_output, stdout_logs, stderr_logs, sandbox.sandbox_id
+        finally:
+            # Dispose of sandbox if requested to reduce usage costs
+            if dispose_sandbox and sandbox:
+                await sandbox.kill()
+
+    def process_execution_results(
+        self, results: list[E2BExecutionResult]
+    ) -> tuple[dict[str, Any] | None, list[dict[str, Any]]]:
+        """Process and filter execution results."""
+        # Filter out empty formats and convert to dicts
+        processed_results = [
+            {
+                f: value
+                for f in [*r.formats(), "extra", "is_main_result"]
+                if (value := getattr(r, f, None)) is not None
+            }
+            for r in results
+        ]
+        if main_result := next(
+            (r for r in processed_results if r.get("is_main_result")), None
+        ):
+            # Make main_result a copy we can modify & remove is_main_result
+            (main_result := {**main_result}).pop("is_main_result")
+
+        return main_result, processed_results
+
+
+class ExecuteCodeBlock(Block, BaseE2BExecutorMixin):
     # TODO : Add support to upload and download files
-    # Currently, You can customized the CPU and Memory, only by creating a pre customized sandbox template
+    # NOTE: Currently, you can only customize the CPU and Memory
+    #       by creating a pre customized sandbox template
     class Input(BlockSchema):
         credentials: CredentialsMetaInput[
             Literal[ProviderName.E2B], Literal["api_key"]
         ] = CredentialsField(
-            description="Enter your api key for the E2B Sandbox. You can get it in here - https://e2b.dev/docs",
+            description=(
+                "Enter your API key for the E2B platform. "
+                "You can get it in here - https://e2b.dev/docs"
+            ),
         )
 
         # Todo : Option to run commond in background
@@ -76,6 +199,14 @@ class CodeExecutionBlock(Block):
             description="Execution timeout in seconds", default=300
         )
 
+        dispose_sandbox: bool = SchemaField(
+            description=(
+                "Whether to dispose of the sandbox immediately after execution. "
+                "If disabled, the sandbox will run until its timeout expires."
+            ),
+            default=True,
+        )
+
         template_id: str = SchemaField(
             description=(
                 "You can use an E2B sandbox template by entering its ID here. "
@@ -87,7 +218,16 @@ class CodeExecutionBlock(Block):
         )
 
     class Output(BlockSchema):
-        response: str = SchemaField(description="Response from code execution")
+        main_result: MainCodeExecutionResult = SchemaField(
+            title="Main Result", description="The main result from the code execution"
+        )
+        results: list[CodeExecutionResult] = SchemaField(
+            description="List of results from the code execution"
+        )
+        response: str = SchemaField(
+            title="Main Text Output",
+            description="Text output (if any) of the main execution result",
+        )
         stdout_logs: str = SchemaField(
             description="Standard output logs from execution"
         )
@@ -97,10 +237,10 @@ class CodeExecutionBlock(Block):
     def __init__(self):
         super().__init__(
             id="0b02b072-abe7-11ef-8372-fb5d162dd712",
-            description="Executes code in an isolated sandbox environment with internet access.",
+            description="Executes code in a sandbox environment with internet access.",
             categories={BlockCategory.DEVELOPER_TOOLS},
-            input_schema=CodeExecutionBlock.Input,
-            output_schema=CodeExecutionBlock.Output,
+            input_schema=ExecuteCodeBlock.Input,
+            output_schema=ExecuteCodeBlock.Output,
             test_credentials=TEST_CREDENTIALS,
             test_input={
                 "credentials": TEST_CREDENTIALS_INPUT,
@@ -111,91 +251,59 @@ class CodeExecutionBlock(Block):
                 "template_id": "",
             },
             test_output=[
+                ("results", []),
                 ("response", "Hello World"),
                 ("stdout_logs", "Hello World\n"),
             ],
             test_mock={
-                "execute_code": lambda code, language, setup_commands, timeout, api_key, template_id: (
-                    "Hello World",
-                    "Hello World\n",
-                    "",
+                "execute_code": lambda api_key, code, language, template_id, setup_commands, timeout, dispose_sandbox: (  # noqa
+                    [],  # results
+                    "Hello World",  # text_output
+                    "Hello World\n",  # stdout_logs
+                    "",  # stderr_logs
+                    "sandbox_id",  # sandbox_id
                 ),
             },
         )
 
-    async def execute_code(
-        self,
-        code: str,
-        language: ProgrammingLanguage,
-        setup_commands: list[str],
-        timeout: int,
-        api_key: str,
-        template_id: str,
-    ):
-        try:
-            sandbox = None
-            if template_id:
-                sandbox = await AsyncSandbox.create(
-                    template=template_id, api_key=api_key, timeout=timeout
-                )
-            else:
-                sandbox = await AsyncSandbox.create(api_key=api_key, timeout=timeout)
-
-            if not sandbox:
-                raise Exception("Sandbox not created")
-
-            # Running setup commands
-            for cmd in setup_commands:
-                await sandbox.commands.run(cmd)
-
-            # Executing the code
-            execution = await sandbox.run_code(
-                code,
-                language=language.value,
-                on_error=lambda e: sandbox.kill(),  # Kill the sandbox if there is an error
-            )
-
-            if execution.error:
-                raise Exception(execution.error)
-
-            response = execution.text
-            stdout_logs = "".join(execution.logs.stdout)
-            stderr_logs = "".join(execution.logs.stderr)
-
-            return response, stdout_logs, stderr_logs
-
-        except Exception as e:
-            raise e
-
     async def run(
         self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs
     ) -> BlockOutput:
         try:
-            response, stdout_logs, stderr_logs = await self.execute_code(
-                input_data.code,
-                input_data.language,
-                input_data.setup_commands,
-                input_data.timeout,
-                credentials.api_key.get_secret_value(),
-                input_data.template_id,
+            results, text_output, stdout, stderr, _ = await self.execute_code(
+                api_key=credentials.api_key.get_secret_value(),
+                code=input_data.code,
+                language=input_data.language,
+                template_id=input_data.template_id,
+                setup_commands=input_data.setup_commands,
+                timeout=input_data.timeout,
+                dispose_sandbox=input_data.dispose_sandbox,
             )
 
-            if response:
-                yield "response", response
-            if stdout_logs:
-                yield "stdout_logs", stdout_logs
-            if stderr_logs:
-                yield "stderr_logs", stderr_logs
+            # Determine result object shape & filter out empty formats
+            main_result, results = self.process_execution_results(results)
+            if main_result:
+                yield "main_result", main_result
+            yield "results", results
+            if text_output:
+                yield "response", text_output
+            if stdout:
+                yield "stdout_logs", stdout
+            if stderr:
+                yield "stderr_logs", stderr
         except Exception as e:
             yield "error", str(e)
 
 
-class InstantiationBlock(Block):
+class InstantiateCodeSandboxBlock(Block, BaseE2BExecutorMixin):
     class Input(BlockSchema):
         credentials: CredentialsMetaInput[
             Literal[ProviderName.E2B], Literal["api_key"]
         ] = CredentialsField(
-            description="Enter your api key for the E2B Sandbox. You can get it in here - https://e2b.dev/docs",
+            description=(
+                "Enter your API key for the E2B platform. "
+                "You can get it in here - https://e2b.dev/docs"
+            )
         )
 
         # Todo : Option to run commond in background
@@ -240,7 +348,10 @@ class InstantiationBlock(Block):
 
     class Output(BlockSchema):
         sandbox_id: str = SchemaField(description="ID of the sandbox instance")
-        response: str = SchemaField(description="Response from code execution")
+        response: str = SchemaField(
+            title="Text Result",
+            description="Text result (if any) of the setup code execution",
+        )
         stdout_logs: str = SchemaField(
             description="Standard output logs from execution"
         )
@@ -250,10 +361,13 @@ class InstantiationBlock(Block):
     def __init__(self):
         super().__init__(
             id="ff0861c9-1726-4aec-9e5b-bf53f3622112",
-            description="Instantiate an isolated sandbox environment with internet access where to execute code in.",
+            description=(
+                "Instantiate a sandbox environment with internet access "
+                "in which you can execute code with the Execute Code Step block."
+            ),
             categories={BlockCategory.DEVELOPER_TOOLS},
-            input_schema=InstantiationBlock.Input,
-            output_schema=InstantiationBlock.Output,
+            input_schema=InstantiateCodeSandboxBlock.Input,
+            output_schema=InstantiateCodeSandboxBlock.Output,
             test_credentials=TEST_CREDENTIALS,
             test_input={
                 "credentials": TEST_CREDENTIALS_INPUT,
@@ -269,11 +383,12 @@ class InstantiationBlock(Block):
                 ("stdout_logs", "Hello World\n"),
             ],
             test_mock={
-                "execute_code": lambda setup_code, language, setup_commands, timeout, api_key, template_id: (
-                    "sandbox_id",
-                    "Hello World",
-                    "Hello World\n",
-                    "",
+                "execute_code": lambda api_key, code, language, template_id, setup_commands, timeout: (  # noqa
+                    [],  # results
+                    "Hello World",  # text_output
+                    "Hello World\n",  # stdout_logs
+                    "",  # stderr_logs
+                    "sandbox_id",  # sandbox_id
                 ),
             },
         )
@@ -282,78 +397,38 @@ class InstantiationBlock(Block):
         self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs
     ) -> BlockOutput:
         try:
-            sandbox_id, response, stdout_logs, stderr_logs = await self.execute_code(
-                input_data.setup_code,
-                input_data.language,
-                input_data.setup_commands,
-                input_data.timeout,
-                credentials.api_key.get_secret_value(),
-                input_data.template_id,
+            _, text_output, stdout, stderr, sandbox_id = await self.execute_code(
+                api_key=credentials.api_key.get_secret_value(),
+                code=input_data.setup_code,
+                language=input_data.language,
+                template_id=input_data.template_id,
+                setup_commands=input_data.setup_commands,
+                timeout=input_data.timeout,
             )
             if sandbox_id:
                 yield "sandbox_id", sandbox_id
             else:
                 yield "error", "Sandbox ID not found"
-            if response:
-                yield "response", response
-            if stdout_logs:
-                yield "stdout_logs", stdout_logs
-            if stderr_logs:
-                yield "stderr_logs", stderr_logs
+
+            if text_output:
+                yield "response", text_output
+            if stdout:
+                yield "stdout_logs", stdout
+            if stderr:
+                yield "stderr_logs", stderr
         except Exception as e:
             yield "error", str(e)
 
-    async def execute_code(
-        self,
-        code: str,
-        language: ProgrammingLanguage,
-        setup_commands: list[str],
-        timeout: int,
-        api_key: str,
-        template_id: str,
-    ):
-        try:
-            sandbox = None
-            if template_id:
-                sandbox = await AsyncSandbox.create(
-                    template=template_id, api_key=api_key, timeout=timeout
-                )
-            else:
-                sandbox = await AsyncSandbox.create(api_key=api_key, timeout=timeout)
 
-            if not sandbox:
-                raise Exception("Sandbox not created")
-
-            # Running setup commands
-            for cmd in setup_commands:
-                await sandbox.commands.run(cmd)
-
-            # Executing the code
-            execution = await sandbox.run_code(
-                code,
-                language=language.value,
-                on_error=lambda e: sandbox.kill(),  # Kill the sandbox if there is an error
-            )
-
-            if execution.error:
-                raise Exception(execution.error)
-
-            response = execution.text
-            stdout_logs = "".join(execution.logs.stdout)
-            stderr_logs = "".join(execution.logs.stderr)
-
-            return sandbox.sandbox_id, response, stdout_logs, stderr_logs
-
-        except Exception as e:
-            raise e
-
-
-class StepExecutionBlock(Block):
+class ExecuteCodeStepBlock(Block, BaseE2BExecutorMixin):
     class Input(BlockSchema):
         credentials: CredentialsMetaInput[
             Literal[ProviderName.E2B], Literal["api_key"]
         ] = CredentialsField(
-            description="Enter your api key for the E2B Sandbox. You can get it in here - https://e2b.dev/docs",
+            description=(
+                "Enter your API key for the E2B platform. "
+                "You can get it in here - https://e2b.dev/docs"
+            ),
         )
 
         sandbox_id: str = SchemaField(
@@ -374,8 +449,22 @@ class StepExecutionBlock(Block):
             advanced=False,
         )
 
+        dispose_sandbox: bool = SchemaField(
+            description="Whether to dispose of the sandbox after executing this code.",
+            default=False,
+        )
+
     class Output(BlockSchema):
-        response: str = SchemaField(description="Response from code execution")
+        main_result: MainCodeExecutionResult = SchemaField(
+            title="Main Result", description="The main result from the code execution"
+        )
+        results: list[CodeExecutionResult] = SchemaField(
+            description="List of results from the code execution"
+        )
+        response: str = SchemaField(
+            title="Main Text Output",
+            description="Text output (if any) of the main execution result",
+        )
         stdout_logs: str = SchemaField(
             description="Standard output logs from execution"
         )
@@ -385,10 +474,10 @@ class StepExecutionBlock(Block):
     def __init__(self):
         super().__init__(
             id="82b59b8e-ea10-4d57-9161-8b169b0adba6",
-            description="Execute code in a previously instantiated sandbox environment.",
+            description="Execute code in a previously instantiated sandbox.",
             categories={BlockCategory.DEVELOPER_TOOLS},
-            input_schema=StepExecutionBlock.Input,
-            output_schema=StepExecutionBlock.Output,
+            input_schema=ExecuteCodeStepBlock.Input,
+            output_schema=ExecuteCodeStepBlock.Output,
             test_credentials=TEST_CREDENTIALS,
             test_input={
                 "credentials": TEST_CREDENTIALS_INPUT,
@@ -397,61 +486,43 @@ class StepExecutionBlock(Block):
                 "language": ProgrammingLanguage.PYTHON.value,
             },
             test_output=[
+                ("results", []),
                 ("response", "Hello World"),
                 ("stdout_logs", "Hello World\n"),
             ],
             test_mock={
-                "execute_step_code": lambda sandbox_id, step_code, language, api_key: (
-                    "Hello World",
-                    "Hello World\n",
-                    "",
+                "execute_code": lambda api_key, code, language, sandbox_id, dispose_sandbox: (  # noqa
+                    [],  # results
+                    "Hello World",  # text_output
+                    "Hello World\n",  # stdout_logs
+                    "",  # stderr_logs
+                    sandbox_id,  # sandbox_id
                 ),
             },
         )
 
-    async def execute_step_code(
-        self,
-        sandbox_id: str,
-        code: str,
-        language: ProgrammingLanguage,
-        api_key: str,
-    ):
-        try:
-            sandbox = await AsyncSandbox.connect(sandbox_id=sandbox_id, api_key=api_key)
-            if not sandbox:
-                raise Exception("Sandbox not found")
-
-            # Executing the code
-            execution = await sandbox.run_code(code, language=language.value)
-
-            if execution.error:
-                raise Exception(execution.error)
-
-            response = execution.text
-            stdout_logs = "".join(execution.logs.stdout)
-            stderr_logs = "".join(execution.logs.stderr)
-
-            return response, stdout_logs, stderr_logs
-
-        except Exception as e:
-            raise e
-
     async def run(
         self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs
     ) -> BlockOutput:
         try:
-            response, stdout_logs, stderr_logs = await self.execute_step_code(
-                input_data.sandbox_id,
-                input_data.step_code,
-                input_data.language,
-                credentials.api_key.get_secret_value(),
+            results, text_output, stdout, stderr, _ = await self.execute_code(
+                api_key=credentials.api_key.get_secret_value(),
+                code=input_data.step_code,
+                language=input_data.language,
+                sandbox_id=input_data.sandbox_id,
+                dispose_sandbox=input_data.dispose_sandbox,
             )
 
-            if response:
-                yield "response", response
-            if stdout_logs:
-                yield "stdout_logs", stdout_logs
-            if stderr_logs:
-                yield "stderr_logs", stderr_logs
+            # Determine result object shape & filter out empty formats
+            main_result, results = self.process_execution_results(results)
+            if main_result:
+                yield "main_result", main_result
+            yield "results", results
+            if text_output:
+                yield "response", text_output
+            if stdout:
+                yield "stdout_logs", stdout
+            if stderr:
+                yield "stderr_logs", stderr
         except Exception as e:
             yield "error", str(e)

autogpt_platform/backend/backend/blocks/code_extraction_block.py

@@ -90,7 +90,7 @@ class CodeExtractionBlock(Block):
                 for aliases in language_aliases.values()
                 for alias in aliases
             )
-            + r")\s+[\s\S]*?```"
+            + r")[ \t]*\n[\s\S]*?```"
         )
 
         remaining_text = re.sub(pattern, "", input_data.text).strip()
@@ -103,7 +103,9 @@ class CodeExtractionBlock(Block):
         # Escape special regex characters in the language string
         language = re.escape(language)
         # Extract all code blocks enclosed in ```language``` blocks
-        pattern = re.compile(rf"```{language}\s+(.*?)```", re.DOTALL | re.IGNORECASE)
+        pattern = re.compile(
+            rf"```{language}[ \t]*\n(.*?)\n```", re.DOTALL | re.IGNORECASE
+        )
         matches = pattern.finditer(text)
         # Combine all code blocks for this language with newlines between them
         code_blocks = [match.group(1).strip() for match in matches]

autogpt_platform/backend/backend/blocks/data_manipulation.py

@@ -66,6 +66,7 @@ class AddToDictionaryBlock(Block):
         dictionary: dict[Any, Any] = SchemaField(
             default_factory=dict,
             description="The dictionary to add the entry to. If not provided, a new dictionary will be created.",
+            advanced=False,
         )
         key: str = SchemaField(
             default="",

autogpt_platform/backend/backend/blocks/dataforseo/keyword_suggestions.py

@@ -90,6 +90,7 @@ class DataForSeoKeywordSuggestionsBlock(Block):
         seed_keyword: str = SchemaField(
             description="The seed keyword used for the query"
         )
+        error: str = SchemaField(description="Error message if the API call failed")
 
     def __init__(self):
         super().__init__(
@@ -161,43 +162,52 @@ class DataForSeoKeywordSuggestionsBlock(Block):
         **kwargs,
     ) -> BlockOutput:
         """Execute the keyword suggestions query."""
-        client = DataForSeoClient(credentials)
+        try:
+            client = DataForSeoClient(credentials)
 
-        results = await self._fetch_keyword_suggestions(client, input_data)
+            results = await self._fetch_keyword_suggestions(client, input_data)
 
-        # Process and format the results
-        suggestions = []
-        if results and len(results) > 0:
-            # results is a list, get the first element
-            first_result = results[0] if isinstance(results, list) else results
-            items = (
-                first_result.get("items", []) if isinstance(first_result, dict) else []
-            )
-            for item in items:
-                # Create the KeywordSuggestion object
-                suggestion = KeywordSuggestion(
-                    keyword=item.get("keyword", ""),
-                    search_volume=item.get("keyword_info", {}).get("search_volume"),
-                    competition=item.get("keyword_info", {}).get("competition"),
-                    cpc=item.get("keyword_info", {}).get("cpc"),
-                    keyword_difficulty=item.get("keyword_properties", {}).get(
-                        "keyword_difficulty"
-                    ),
-                    serp_info=(
-                        item.get("serp_info") if input_data.include_serp_info else None
-                    ),
-                    clickstream_data=(
-                        item.get("clickstream_keyword_info")
-                        if input_data.include_clickstream_data
-                        else None
-                    ),
+            # Process and format the results
+            suggestions = []
+            if results and len(results) > 0:
+                # results is a list, get the first element
+                first_result = results[0] if isinstance(results, list) else results
+                items = (
+                    first_result.get("items", [])
+                    if isinstance(first_result, dict)
+                    else []
                 )
-                yield "suggestion", suggestion
-                suggestions.append(suggestion)
+                if items is None:
+                    items = []
+                for item in items:
+                    # Create the KeywordSuggestion object
+                    suggestion = KeywordSuggestion(
+                        keyword=item.get("keyword", ""),
+                        search_volume=item.get("keyword_info", {}).get("search_volume"),
+                        competition=item.get("keyword_info", {}).get("competition"),
+                        cpc=item.get("keyword_info", {}).get("cpc"),
+                        keyword_difficulty=item.get("keyword_properties", {}).get(
+                            "keyword_difficulty"
+                        ),
+                        serp_info=(
+                            item.get("serp_info")
+                            if input_data.include_serp_info
+                            else None
+                        ),
+                        clickstream_data=(
+                            item.get("clickstream_keyword_info")
+                            if input_data.include_clickstream_data
+                            else None
+                        ),
+                    )
+                    yield "suggestion", suggestion
+                    suggestions.append(suggestion)
 
-        yield "suggestions", suggestions
-        yield "total_count", len(suggestions)
-        yield "seed_keyword", input_data.keyword
+            yield "suggestions", suggestions
+            yield "total_count", len(suggestions)
+            yield "seed_keyword", input_data.keyword
+        except Exception as e:
+            yield "error", f"Failed to fetch keyword suggestions: {str(e)}"
 
 
 class KeywordSuggestionExtractorBlock(Block):

autogpt_platform/backend/backend/blocks/dataforseo/related_keywords.py

@@ -98,6 +98,7 @@ class DataForSeoRelatedKeywordsBlock(Block):
         seed_keyword: str = SchemaField(
             description="The seed keyword used for the query"
         )
+        error: str = SchemaField(description="Error message if the API call failed")
 
     def __init__(self):
         super().__init__(
@@ -171,50 +172,60 @@ class DataForSeoRelatedKeywordsBlock(Block):
         **kwargs,
     ) -> BlockOutput:
         """Execute the related keywords query."""
-        client = DataForSeoClient(credentials)
+        try:
+            client = DataForSeoClient(credentials)
 
-        results = await self._fetch_related_keywords(client, input_data)
+            results = await self._fetch_related_keywords(client, input_data)
 
-        # Process and format the results
-        related_keywords = []
-        if results and len(results) > 0:
-            # results is a list, get the first element
-            first_result = results[0] if isinstance(results, list) else results
-            items = (
-                first_result.get("items", []) if isinstance(first_result, dict) else []
-            )
-            for item in items:
-                # Extract keyword_data from the item
-                keyword_data = item.get("keyword_data", {})
-
-                # Create the RelatedKeyword object
-                keyword = RelatedKeyword(
-                    keyword=keyword_data.get("keyword", ""),
-                    search_volume=keyword_data.get("keyword_info", {}).get(
-                        "search_volume"
-                    ),
-                    competition=keyword_data.get("keyword_info", {}).get("competition"),
-                    cpc=keyword_data.get("keyword_info", {}).get("cpc"),
-                    keyword_difficulty=keyword_data.get("keyword_properties", {}).get(
-                        "keyword_difficulty"
-                    ),
-                    serp_info=(
-                        keyword_data.get("serp_info")
-                        if input_data.include_serp_info
-                        else None
-                    ),
-                    clickstream_data=(
-                        keyword_data.get("clickstream_keyword_info")
-                        if input_data.include_clickstream_data
-                        else None
-                    ),
+            # Process and format the results
+            related_keywords = []
+            if results and len(results) > 0:
+                # results is a list, get the first element
+                first_result = results[0] if isinstance(results, list) else results
+                items = (
+                    first_result.get("items", [])
+                    if isinstance(first_result, dict)
+                    else []
                 )
-                yield "related_keyword", keyword
-                related_keywords.append(keyword)
+                # Ensure items is never None
+                if items is None:
+                    items = []
+                for item in items:
+                    # Extract keyword_data from the item
+                    keyword_data = item.get("keyword_data", {})
 
-        yield "related_keywords", related_keywords
-        yield "total_count", len(related_keywords)
-        yield "seed_keyword", input_data.keyword
+                    # Create the RelatedKeyword object
+                    keyword = RelatedKeyword(
+                        keyword=keyword_data.get("keyword", ""),
+                        search_volume=keyword_data.get("keyword_info", {}).get(
+                            "search_volume"
+                        ),
+                        competition=keyword_data.get("keyword_info", {}).get(
+                            "competition"
+                        ),
+                        cpc=keyword_data.get("keyword_info", {}).get("cpc"),
+                        keyword_difficulty=keyword_data.get(
+                            "keyword_properties", {}
+                        ).get("keyword_difficulty"),
+                        serp_info=(
+                            keyword_data.get("serp_info")
+                            if input_data.include_serp_info
+                            else None
+                        ),
+                        clickstream_data=(
+                            keyword_data.get("clickstream_keyword_info")
+                            if input_data.include_clickstream_data
+                            else None
+                        ),
+                    )
+                    yield "related_keyword", keyword
+                    related_keywords.append(keyword)
+
+            yield "related_keywords", related_keywords
+            yield "total_count", len(related_keywords)
+            yield "seed_keyword", input_data.keyword
+        except Exception as e:
+            yield "error", f"Failed to fetch related keywords: {str(e)}"
 
 
 class RelatedKeywordExtractorBlock(Block):

autogpt_platform/backend/backend/blocks/discord/bot_blocks.py

@@ -4,13 +4,13 @@ import mimetypes
 from pathlib import Path
 from typing import Any
 
-import aiohttp
 import discord
 from pydantic import SecretStr
 
 from backend.data.block import Block, BlockCategory, BlockOutput, BlockSchema
 from backend.data.model import APIKeyCredentials, SchemaField
 from backend.util.file import store_media_file
+from backend.util.request import Requests
 from backend.util.type import MediaFileType
 
 from ._auth import (
@@ -114,10 +114,9 @@ class ReadDiscordMessagesBlock(Block):
             if message.attachments:
                 attachment = message.attachments[0]  # Process the first attachment
                 if attachment.filename.endswith((".txt", ".py")):
-                    async with aiohttp.ClientSession() as session:
-                        async with session.get(attachment.url) as response:
-                            file_content = response.text()
-                            self.output_data += f"\n\nFile from user: {attachment.filename}\nContent: {file_content}"
+                    response = await Requests().get(attachment.url)
+                    file_content = response.text()
+                    self.output_data += f"\n\nFile from user: {attachment.filename}\nContent: {file_content}"
 
             await client.close()
 
@@ -171,11 +170,11 @@ class SendDiscordMessageBlock(Block):
             description="The content of the message to send"
         )
         channel_name: str = SchemaField(
-            description="The name of the channel the message will be sent to"
+            description="Channel ID or channel name to send the message to"
         )
         server_name: str = SchemaField(
-            description="The name of the server where the channel is located",
-            advanced=True,  # Optional field for server name
+            description="Server name (only needed if using channel name)",
+            advanced=True,
             default="",
         )
 
@@ -231,25 +230,49 @@ class SendDiscordMessageBlock(Block):
         @client.event
         async def on_ready():
             print(f"Logged in as {client.user}")
-            for guild in client.guilds:
-                if server_name and guild.name != server_name:
-                    continue
-                for channel in guild.text_channels:
-                    if channel.name == channel_name:
-                        # Split message into chunks if it exceeds 2000 characters
-                        chunks = self.chunk_message(message_content)
-                        last_message = None
-                        for chunk in chunks:
-                            last_message = await channel.send(chunk)
-                        result["status"] = "Message sent"
-                        result["message_id"] = (
-                            str(last_message.id) if last_message else ""
-                        )
-                        result["channel_id"] = str(channel.id)
-                        await client.close()
-                        return
+            channel = None
 
-            result["status"] = "Channel not found"
+            # Try to parse as channel ID first
+            try:
+                channel_id = int(channel_name)
+                channel = client.get_channel(channel_id)
+            except ValueError:
+                # Not a valid ID, will try name lookup
+                pass
+
+            # If not found by ID (or not an ID), try name lookup
+            if not channel:
+                for guild in client.guilds:
+                    if server_name and guild.name != server_name:
+                        continue
+                    for ch in guild.text_channels:
+                        if ch.name == channel_name:
+                            channel = ch
+                            break
+                    if channel:
+                        break
+
+            if not channel:
+                result["status"] = f"Channel not found: {channel_name}"
+                await client.close()
+                return
+
+            # Type check - ensure it's a text channel that can send messages
+            if not hasattr(channel, "send"):
+                result["status"] = (
+                    f"Channel {channel_name} cannot receive messages (not a text channel)"
+                )
+                await client.close()
+                return
+
+            # Split message into chunks if it exceeds 2000 characters
+            chunks = self.chunk_message(message_content)
+            last_message = None
+            for chunk in chunks:
+                last_message = await channel.send(chunk)  # type: ignore
+            result["status"] = "Message sent"
+            result["message_id"] = str(last_message.id) if last_message else ""
+            result["channel_id"] = str(channel.id)
             await client.close()
 
         await client.start(token)
@@ -675,16 +698,15 @@ class SendDiscordFileBlock(Block):
 
                 elif file.startswith(("http://", "https://")):
                     # URL - download the file
-                    async with aiohttp.ClientSession() as session:
-                        async with session.get(file) as response:
-                            file_bytes = await response.read()
+                    response = await Requests().get(file)
+                    file_bytes = response.content
 
-                            # Try to get filename from URL if not provided
-                            if not filename:
-                                from urllib.parse import urlparse
+                    # Try to get filename from URL if not provided
+                    if not filename:
+                        from urllib.parse import urlparse
 
-                                path = urlparse(file).path
-                                detected_filename = Path(path).name or "download"
+                        path = urlparse(file).path
+                        detected_filename = Path(path).name or "download"
                 else:
                     # Local file path - read from stored media file
                     # This would be a path from a previous block's output

autogpt_platform/backend/backend/blocks/firecrawl/_api.py

@@ -0,0 +1,12 @@
+from enum import Enum
+
+
+class ScrapeFormat(Enum):
+    MARKDOWN = "markdown"
+    HTML = "html"
+    RAW_HTML = "rawHtml"
+    LINKS = "links"
+    SCREENSHOT = "screenshot"
+    SCREENSHOT_FULL_PAGE = "screenshot@fullPage"
+    JSON = "json"
+    CHANGE_TRACKING = "changeTracking"

autogpt_platform/backend/backend/blocks/firecrawl/_format_utils.py

@@ -0,0 +1,28 @@
+"""Utility functions for converting between our ScrapeFormat enum and firecrawl FormatOption types."""
+
+from typing import List
+
+from firecrawl.v2.types import FormatOption, ScreenshotFormat
+
+from backend.blocks.firecrawl._api import ScrapeFormat
+
+
+def convert_to_format_options(
+    formats: List[ScrapeFormat],
+) -> List[FormatOption]:
+    """Convert our ScrapeFormat enum values to firecrawl FormatOption types.
+
+    Handles special cases like screenshot@fullPage which needs to be converted
+    to a ScreenshotFormat object.
+    """
+    result: List[FormatOption] = []
+
+    for format_enum in formats:
+        if format_enum.value == "screenshot@fullPage":
+            # Special case: convert to ScreenshotFormat with full_page=True
+            result.append(ScreenshotFormat(type="screenshot", full_page=True))
+        else:
+            # Regular string literals
+            result.append(format_enum.value)
+
+    return result

autogpt_platform/backend/backend/blocks/firecrawl/crawl.py

@@ -1,8 +1,9 @@
-from enum import Enum
 from typing import Any
 
-from firecrawl import FirecrawlApp, ScrapeOptions
+from firecrawl import FirecrawlApp
+from firecrawl.v2.types import ScrapeOptions
 
+from backend.blocks.firecrawl._api import ScrapeFormat
 from backend.sdk import (
     APIKeyCredentials,
     Block,
@@ -14,21 +15,10 @@ from backend.sdk import (
 )
 
 from ._config import firecrawl
-
-
-class ScrapeFormat(Enum):
-    MARKDOWN = "markdown"
-    HTML = "html"
-    RAW_HTML = "rawHtml"
-    LINKS = "links"
-    SCREENSHOT = "screenshot"
-    SCREENSHOT_FULL_PAGE = "screenshot@fullPage"
-    JSON = "json"
-    CHANGE_TRACKING = "changeTracking"
+from ._format_utils import convert_to_format_options
 
 
 class FirecrawlCrawlBlock(Block):
-
     class Input(BlockSchema):
         credentials: CredentialsMetaInput = firecrawl.credentials_field()
         url: str = SchemaField(description="The URL to crawl")
@@ -78,18 +68,17 @@ class FirecrawlCrawlBlock(Block):
     async def run(
         self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs
     ) -> BlockOutput:
-
         app = FirecrawlApp(api_key=credentials.api_key.get_secret_value())
 
         # Sync call
-        crawl_result = app.crawl_url(
+        crawl_result = app.crawl(
             input_data.url,
             limit=input_data.limit,
             scrape_options=ScrapeOptions(
-                formats=[format.value for format in input_data.formats],
-                onlyMainContent=input_data.only_main_content,
-                maxAge=input_data.max_age,
-                waitFor=input_data.wait_for,
+                formats=convert_to_format_options(input_data.formats),
+                only_main_content=input_data.only_main_content,
+                max_age=input_data.max_age,
+                wait_for=input_data.wait_for,
             ),
         )
         yield "data", crawl_result.data
@@ -101,7 +90,7 @@ class FirecrawlCrawlBlock(Block):
                 elif f == ScrapeFormat.HTML:
                     yield "html", data.html
                 elif f == ScrapeFormat.RAW_HTML:
-                    yield "raw_html", data.rawHtml
+                    yield "raw_html", data.raw_html
                 elif f == ScrapeFormat.LINKS:
                     yield "links", data.links
                 elif f == ScrapeFormat.SCREENSHOT:
@@ -109,6 +98,6 @@ class FirecrawlCrawlBlock(Block):
                 elif f == ScrapeFormat.SCREENSHOT_FULL_PAGE:
                     yield "screenshot_full_page", data.screenshot
                 elif f == ScrapeFormat.CHANGE_TRACKING:
-                    yield "change_tracking", data.changeTracking
+                    yield "change_tracking", data.change_tracking
                 elif f == ScrapeFormat.JSON:
                     yield "json", data.json

autogpt_platform/backend/backend/blocks/firecrawl/extract.py

@@ -20,7 +20,6 @@ from ._config import firecrawl
 
 @cost(BlockCost(2, BlockCostType.RUN))
 class FirecrawlExtractBlock(Block):
-
     class Input(BlockSchema):
         credentials: CredentialsMetaInput = firecrawl.credentials_field()
         urls: list[str] = SchemaField(
@@ -53,7 +52,6 @@ class FirecrawlExtractBlock(Block):
     async def run(
         self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs
     ) -> BlockOutput:
-
         app = FirecrawlApp(api_key=credentials.api_key.get_secret_value())
 
         extract_result = app.extract(

autogpt_platform/backend/backend/blocks/firecrawl/map.py

@@ -1,3 +1,5 @@
+from typing import Any
+
 from firecrawl import FirecrawlApp
 
 from backend.sdk import (
@@ -14,14 +16,16 @@ from ._config import firecrawl
 
 
 class FirecrawlMapWebsiteBlock(Block):
-
     class Input(BlockSchema):
         credentials: CredentialsMetaInput = firecrawl.credentials_field()
 
         url: str = SchemaField(description="The website url to map")
 
     class Output(BlockSchema):
-        links: list[str] = SchemaField(description="The links of the website")
+        links: list[str] = SchemaField(description="List of URLs found on the website")
+        results: list[dict[str, Any]] = SchemaField(
+            description="List of search results with url, title, and description"
+        )
 
     def __init__(self):
         super().__init__(
@@ -35,12 +39,22 @@ class FirecrawlMapWebsiteBlock(Block):
     async def run(
         self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs
     ) -> BlockOutput:
-
         app = FirecrawlApp(api_key=credentials.api_key.get_secret_value())
 
         # Sync call
-        map_result = app.map_url(
+        map_result = app.map(
             url=input_data.url,
         )
 
-        yield "links", map_result.links
+        # Convert SearchResult objects to dicts
+        results_data = [
+            {
+                "url": link.url,
+                "title": link.title,
+                "description": link.description,
+            }
+            for link in map_result.links
+        ]
+
+        yield "links", [link.url for link in map_result.links]
+        yield "results", results_data

autogpt_platform/backend/backend/blocks/firecrawl/scrape.py

@@ -1,8 +1,8 @@
-from enum import Enum
 from typing import Any
 
 from firecrawl import FirecrawlApp
 
+from backend.blocks.firecrawl._api import ScrapeFormat
 from backend.sdk import (
     APIKeyCredentials,
     Block,
@@ -14,21 +14,10 @@ from backend.sdk import (
 )
 
 from ._config import firecrawl
-
-
-class ScrapeFormat(Enum):
-    MARKDOWN = "markdown"
-    HTML = "html"
-    RAW_HTML = "rawHtml"
-    LINKS = "links"
-    SCREENSHOT = "screenshot"
-    SCREENSHOT_FULL_PAGE = "screenshot@fullPage"
-    JSON = "json"
-    CHANGE_TRACKING = "changeTracking"
+from ._format_utils import convert_to_format_options
 
 
 class FirecrawlScrapeBlock(Block):
-
     class Input(BlockSchema):
         credentials: CredentialsMetaInput = firecrawl.credentials_field()
         url: str = SchemaField(description="The URL to crawl")
@@ -78,12 +67,11 @@ class FirecrawlScrapeBlock(Block):
     async def run(
         self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs
     ) -> BlockOutput:
-
         app = FirecrawlApp(api_key=credentials.api_key.get_secret_value())
 
-        scrape_result = app.scrape_url(
+        scrape_result = app.scrape(
             input_data.url,
-            formats=[format.value for format in input_data.formats],
+            formats=convert_to_format_options(input_data.formats),
             only_main_content=input_data.only_main_content,
             max_age=input_data.max_age,
             wait_for=input_data.wait_for,
@@ -96,7 +84,7 @@ class FirecrawlScrapeBlock(Block):
             elif f == ScrapeFormat.HTML:
                 yield "html", scrape_result.html
             elif f == ScrapeFormat.RAW_HTML:
-                yield "raw_html", scrape_result.rawHtml
+                yield "raw_html", scrape_result.raw_html
             elif f == ScrapeFormat.LINKS:
                 yield "links", scrape_result.links
             elif f == ScrapeFormat.SCREENSHOT:
@@ -104,6 +92,6 @@ class FirecrawlScrapeBlock(Block):
             elif f == ScrapeFormat.SCREENSHOT_FULL_PAGE:
                 yield "screenshot_full_page", scrape_result.screenshot
             elif f == ScrapeFormat.CHANGE_TRACKING:
-                yield "change_tracking", scrape_result.changeTracking
+                yield "change_tracking", scrape_result.change_tracking
             elif f == ScrapeFormat.JSON:
                 yield "json", scrape_result.json

autogpt_platform/backend/backend/blocks/firecrawl/search.py

@@ -1,8 +1,9 @@
-from enum import Enum
 from typing import Any
 
-from firecrawl import FirecrawlApp, ScrapeOptions
+from firecrawl import FirecrawlApp
+from firecrawl.v2.types import ScrapeOptions
 
+from backend.blocks.firecrawl._api import ScrapeFormat
 from backend.sdk import (
     APIKeyCredentials,
     Block,
@@ -14,21 +15,10 @@ from backend.sdk import (
 )
 
 from ._config import firecrawl
-
-
-class ScrapeFormat(Enum):
-    MARKDOWN = "markdown"
-    HTML = "html"
-    RAW_HTML = "rawHtml"
-    LINKS = "links"
-    SCREENSHOT = "screenshot"
-    SCREENSHOT_FULL_PAGE = "screenshot@fullPage"
-    JSON = "json"
-    CHANGE_TRACKING = "changeTracking"
+from ._format_utils import convert_to_format_options
 
 
 class FirecrawlSearchBlock(Block):
-
     class Input(BlockSchema):
         credentials: CredentialsMetaInput = firecrawl.credentials_field()
         query: str = SchemaField(description="The query to search for")
@@ -61,7 +51,6 @@ class FirecrawlSearchBlock(Block):
     async def run(
         self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs
     ) -> BlockOutput:
-
         app = FirecrawlApp(api_key=credentials.api_key.get_secret_value())
 
         # Sync call
@@ -69,11 +58,12 @@ class FirecrawlSearchBlock(Block):
             input_data.query,
             limit=input_data.limit,
             scrape_options=ScrapeOptions(
-                formats=[format.value for format in input_data.formats],
-                maxAge=input_data.max_age,
-                waitFor=input_data.wait_for,
+                formats=convert_to_format_options(input_data.formats) or None,
+                max_age=input_data.max_age,
+                wait_for=input_data.wait_for,
             ),
         )
         yield "data", scrape_result
-        for site in scrape_result.data:
-            yield "site", site
+        if hasattr(scrape_result, "web") and scrape_result.web:
+            for site in scrape_result.web:
+                yield "site", site

autogpt_platform/backend/backend/blocks/io.py

@@ -554,6 +554,89 @@ class AgentToggleInputBlock(AgentInputBlock):
         )
 
 
+class AgentTableInputBlock(AgentInputBlock):
+    """
+    This block allows users to input data in a table format.
+
+    Configure the table columns at build time, then users can input
+    rows of data at runtime. Each row is output as a dictionary
+    with column names as keys.
+    """
+
+    class Input(AgentInputBlock.Input):
+        value: Optional[list[dict[str, Any]]] = SchemaField(
+            description="The table data as a list of dictionaries.",
+            default=None,
+            advanced=False,
+            title="Default Value",
+        )
+        column_headers: list[str] = SchemaField(
+            description="Column headers for the table.",
+            default_factory=lambda: ["Column 1", "Column 2", "Column 3"],
+            advanced=False,
+            title="Column Headers",
+        )
+
+        def generate_schema(self):
+            """Generate schema for the value field with table format."""
+            schema = super().generate_schema()
+            schema["type"] = "array"
+            schema["format"] = "table"
+            schema["items"] = {
+                "type": "object",
+                "properties": {
+                    header: {"type": "string"}
+                    for header in (
+                        self.column_headers or ["Column 1", "Column 2", "Column 3"]
+                    )
+                },
+            }
+            if self.value is not None:
+                schema["default"] = self.value
+            return schema
+
+    class Output(AgentInputBlock.Output):
+        result: list[dict[str, Any]] = SchemaField(
+            description="The table data as a list of dictionaries with headers as keys."
+        )
+
+    def __init__(self):
+        super().__init__(
+            id="5603b273-f41e-4020-af7d-fbc9c6a8d928",
+            description="Block for table data input with customizable headers.",
+            disabled=not config.enable_agent_input_subtype_blocks,
+            input_schema=AgentTableInputBlock.Input,
+            output_schema=AgentTableInputBlock.Output,
+            test_input=[
+                {
+                    "name": "test_table",
+                    "column_headers": ["Name", "Age", "City"],
+                    "value": [
+                        {"Name": "John", "Age": "30", "City": "New York"},
+                        {"Name": "Jane", "Age": "25", "City": "London"},
+                    ],
+                    "description": "Example table input",
+                }
+            ],
+            test_output=[
+                (
+                    "result",
+                    [
+                        {"Name": "John", "Age": "30", "City": "New York"},
+                        {"Name": "Jane", "Age": "25", "City": "London"},
+                    ],
+                )
+            ],
+        )
+
+    async def run(self, input_data: Input, *args, **kwargs) -> BlockOutput:
+        """
+        Yields the table data as a list of dictionaries.
+        """
+        # Pass through the value, defaulting to empty list if None
+        yield "result", input_data.value if input_data.value is not None else []
+
+
 IO_BLOCK_IDs = [
     AgentInputBlock().id,
     AgentOutputBlock().id,
@@ -565,4 +648,5 @@ IO_BLOCK_IDs = [
     AgentFileInputBlock().id,
     AgentDropdownInputBlock().id,
     AgentToggleInputBlock().id,
+    AgentTableInputBlock().id,
 ]

autogpt_platform/backend/backend/blocks/iteration.py

@@ -2,7 +2,7 @@ from typing import Any
 
 from backend.data.block import Block, BlockCategory, BlockOutput, BlockSchema
 from backend.data.model import SchemaField
-from backend.util.json import json
+from backend.util.json import loads
 
 
 class StepThroughItemsBlock(Block):
@@ -54,20 +54,43 @@ class StepThroughItemsBlock(Block):
         )
 
     async def run(self, input_data: Input, **kwargs) -> BlockOutput:
+        # Security fix: Add limits to prevent DoS from large iterations
+        MAX_ITEMS = 10000  # Maximum items to iterate
+        MAX_ITEM_SIZE = 1024 * 1024  # 1MB per item
+
         for data in [input_data.items, input_data.items_object, input_data.items_str]:
             if not data:
                 continue
+
+            # Limit string size before parsing
             if isinstance(data, str):
-                items = json.loads(data)
+                if len(data) > MAX_ITEM_SIZE:
+                    raise ValueError(
+                        f"Input too large: {len(data)} bytes > {MAX_ITEM_SIZE} bytes"
+                    )
+                items = loads(data)
             else:
                 items = data
+
+            # Check total item count
+            if isinstance(items, (list, dict)):
+                if len(items) > MAX_ITEMS:
+                    raise ValueError(f"Too many items: {len(items)} > {MAX_ITEMS}")
+
+            iteration_count = 0
             if isinstance(items, dict):
                 # If items is a dictionary, iterate over its values
-                for item in items.values():
-                    yield "item", item
-                    yield "key", item
+                for key, value in items.items():
+                    if iteration_count >= MAX_ITEMS:
+                        break
+                    yield "item", value
+                    yield "key", key  # Fixed: should yield key, not item
+                    iteration_count += 1
             else:
                 # If items is a list, iterate over the list
                 for index, item in enumerate(items):
+                    if iteration_count >= MAX_ITEMS:
+                        break
                     yield "item", item
                     yield "key", index
+                    iteration_count += 1

autogpt_platform/backend/backend/blocks/jina/fact_checker.py

@@ -1,5 +1,8 @@
+from typing import List
 from urllib.parse import quote
 
+from typing_extensions import TypedDict
+
 from backend.blocks.jina._auth import (
     JinaCredentials,
     JinaCredentialsField,
@@ -10,6 +13,12 @@ from backend.data.model import SchemaField
 from backend.util.request import Requests
 
 
+class Reference(TypedDict):
+    url: str
+    keyQuote: str
+    isSupportive: bool
+
+
 class FactCheckerBlock(Block):
     class Input(BlockSchema):
         statement: str = SchemaField(
@@ -23,6 +32,10 @@ class FactCheckerBlock(Block):
         )
         result: bool = SchemaField(description="The result of the factuality check")
         reason: str = SchemaField(description="The reason for the factuality result")
+        references: List[Reference] = SchemaField(
+            description="List of references supporting or contradicting the statement",
+            default=[],
+        )
         error: str = SchemaField(description="Error message if the check fails")
 
     def __init__(self):
@@ -53,5 +66,11 @@ class FactCheckerBlock(Block):
             yield "factuality", data["factuality"]
             yield "result", data["result"]
             yield "reason", data["reason"]
+
+            # Yield references if present in the response
+            if "references" in data:
+                yield "references", data["references"]
+            else:
+                yield "references", []
         else:
             raise RuntimeError(f"Expected 'data' key not found in response: {data}")

autogpt_platform/backend/backend/blocks/linear/_config.py

@@ -62,10 +62,10 @@ TEST_CREDENTIALS_OAUTH = OAuth2Credentials(
     title="Mock Linear API key",
     username="mock-linear-username",
     access_token=SecretStr("mock-linear-access-token"),
-    access_token_expires_at=None,
+    access_token_expires_at=1672531200,  # Mock expiration time for short-lived token
     refresh_token=SecretStr("mock-linear-refresh-token"),
     refresh_token_expires_at=None,
-    scopes=["mock-linear-scopes"],
+    scopes=["read", "write"],
 )
 
 TEST_CREDENTIALS_API_KEY = APIKeyCredentials(

autogpt_platform/backend/backend/blocks/linear/_oauth.py

@@ -2,7 +2,9 @@
 Linear OAuth handler implementation.
 """
 
+import base64
 import json
+import time
 from typing import Optional
 from urllib.parse import urlencode
 
@@ -38,8 +40,9 @@ class LinearOAuthHandler(BaseOAuthHandler):
         self.client_secret = client_secret
         self.redirect_uri = redirect_uri
         self.auth_base_url = "https://linear.app/oauth/authorize"
-        self.token_url = "https://api.linear.app/oauth/token"  # Correct token URL
+        self.token_url = "https://api.linear.app/oauth/token"
         self.revoke_url = "https://api.linear.app/oauth/revoke"
+        self.migrate_url = "https://api.linear.app/oauth/migrate_old_token"
 
     def get_login_url(
         self, scopes: list[str], state: str, code_challenge: Optional[str]
@@ -82,19 +85,84 @@ class LinearOAuthHandler(BaseOAuthHandler):
 
         return True  # Linear doesn't return JSON on successful revoke
 
+    async def migrate_old_token(
+        self, credentials: OAuth2Credentials
+    ) -> OAuth2Credentials:
+        """
+        Migrate an old long-lived token to a new short-lived token with refresh token.
+
+        This uses Linear's /oauth/migrate_old_token endpoint to exchange current
+        long-lived tokens for short-lived tokens with refresh tokens without
+        requiring users to re-authorize.
+        """
+        if not credentials.access_token:
+            raise ValueError("No access token to migrate")
+
+        request_body = {
+            "client_id": self.client_id,
+            "client_secret": self.client_secret,
+        }
+
+        headers = {
+            "Authorization": f"Bearer {credentials.access_token.get_secret_value()}",
+            "Content-Type": "application/x-www-form-urlencoded",
+        }
+
+        response = await Requests().post(
+            self.migrate_url, data=request_body, headers=headers
+        )
+
+        if not response.ok:
+            try:
+                error_data = response.json()
+                error_message = error_data.get("error", "Unknown error")
+                error_description = error_data.get("error_description", "")
+                if error_description:
+                    error_message = f"{error_message}: {error_description}"
+            except json.JSONDecodeError:
+                error_message = response.text
+            raise LinearAPIException(
+                f"Failed to migrate Linear token ({response.status}): {error_message}",
+                response.status,
+            )
+
+        token_data = response.json()
+
+        # Extract token expiration
+        now = int(time.time())
+        expires_in = token_data.get("expires_in")
+        access_token_expires_at = None
+        if expires_in:
+            access_token_expires_at = now + expires_in
+
+        new_credentials = OAuth2Credentials(
+            provider=self.PROVIDER_NAME,
+            title=credentials.title,
+            username=credentials.username,
+            access_token=token_data["access_token"],
+            scopes=credentials.scopes,  # Preserve original scopes
+            refresh_token=token_data.get("refresh_token"),
+            access_token_expires_at=access_token_expires_at,
+            refresh_token_expires_at=None,
+        )
+
+        new_credentials.id = credentials.id
+        return new_credentials
+
     async def _refresh_tokens(
         self, credentials: OAuth2Credentials
     ) -> OAuth2Credentials:
         if not credentials.refresh_token:
             raise ValueError(
-                "No refresh token available."
-            )  # Linear uses non-expiring tokens
+                "No refresh token available. Token may need to be migrated to the new refresh token system."
+            )
 
         return await self._request_tokens(
             {
                 "refresh_token": credentials.refresh_token.get_secret_value(),
                 "grant_type": "refresh_token",
-            }
+            },
+            current_credentials=credentials,
         )
 
     async def _request_tokens(
@@ -102,16 +170,33 @@ class LinearOAuthHandler(BaseOAuthHandler):
         params: dict[str, str],
         current_credentials: Optional[OAuth2Credentials] = None,
     ) -> OAuth2Credentials:
+        # Determine if this is a refresh token request
+        is_refresh = params.get("grant_type") == "refresh_token"
+
+        # Build request body with appropriate grant_type
         request_body = {
             "client_id": self.client_id,
             "client_secret": self.client_secret,
-            "grant_type": "authorization_code",  # Ensure grant_type is correct
             **params,
         }
 
-        headers = {
-            "Content-Type": "application/x-www-form-urlencoded"
-        }  # Correct header for token request
+        # Set default grant_type if not provided
+        if "grant_type" not in request_body:
+            request_body["grant_type"] = "authorization_code"
+
+        headers = {"Content-Type": "application/x-www-form-urlencoded"}
+
+        # For refresh token requests, support HTTP Basic Authentication as recommended
+        if is_refresh:
+            # Option 1: Use HTTP Basic Auth (preferred by Linear)
+            client_credentials = f"{self.client_id}:{self.client_secret}"
+            encoded_credentials = base64.b64encode(client_credentials.encode()).decode()
+            headers["Authorization"] = f"Basic {encoded_credentials}"
+
+            # Remove client credentials from body when using Basic Auth
+            request_body.pop("client_id", None)
+            request_body.pop("client_secret", None)
+
         response = await Requests().post(
             self.token_url, data=request_body, headers=headers
         )
@@ -120,6 +205,9 @@ class LinearOAuthHandler(BaseOAuthHandler):
             try:
                 error_data = response.json()
                 error_message = error_data.get("error", "Unknown error")
+                error_description = error_data.get("error_description", "")
+                if error_description:
+                    error_message = f"{error_message}: {error_description}"
             except json.JSONDecodeError:
                 error_message = response.text
             raise LinearAPIException(
@@ -129,27 +217,84 @@ class LinearOAuthHandler(BaseOAuthHandler):
 
         token_data = response.json()
 
-        # Note: Linear access tokens do not expire, so we set expires_at to None
+        # Extract token expiration if provided (for new refresh token implementation)
+        now = int(time.time())
+        expires_in = token_data.get("expires_in")
+        access_token_expires_at = None
+        if expires_in:
+            access_token_expires_at = now + expires_in
+
+        # Get username - preserve from current credentials if refreshing
+        username = None
+        if current_credentials and is_refresh:
+            username = current_credentials.username
+        elif "user" in token_data:
+            username = token_data["user"].get("name", "Unknown User")
+        else:
+            # Fetch username using the access token
+            username = await self._request_username(token_data["access_token"])
+
         new_credentials = OAuth2Credentials(
             provider=self.PROVIDER_NAME,
             title=current_credentials.title if current_credentials else None,
-            username=token_data.get("user", {}).get(
-                "name", "Unknown User"
-            ),  # extract name or set appropriate
+            username=username or "Unknown User",
             access_token=token_data["access_token"],
-            scopes=token_data["scope"].split(
-                ","
-            ),  # Linear returns comma-separated scopes
-            refresh_token=token_data.get(
-                "refresh_token"
-            ),  # Linear uses non-expiring tokens so this might be null
-            access_token_expires_at=None,
-            refresh_token_expires_at=None,
+            scopes=(
+                token_data["scope"].split(",")
+                if "scope" in token_data
+                else (current_credentials.scopes if current_credentials else [])
+            ),
+            refresh_token=token_data.get("refresh_token"),
+            access_token_expires_at=access_token_expires_at,
+            refresh_token_expires_at=None,  # Linear doesn't provide refresh token expiration
         )
+
         if current_credentials:
             new_credentials.id = current_credentials.id
+
         return new_credentials
 
+    async def get_access_token(self, credentials: OAuth2Credentials) -> str:
+        """
+        Returns a valid access token, handling migration and refresh as needed.
+
+        This overrides the base implementation to handle Linear's token migration
+        from old long-lived tokens to new short-lived tokens with refresh tokens.
+        """
+        # If token has no expiration and no refresh token, it might be an old token
+        # that needs migration
+        if (
+            credentials.access_token_expires_at is None
+            and credentials.refresh_token is None
+        ):
+            try:
+                # Attempt to migrate the old token
+                migrated_credentials = await self.migrate_old_token(credentials)
+                # Update the credentials store would need to be handled by the caller
+                # For now, use the migrated credentials for this request
+                credentials = migrated_credentials
+            except LinearAPIException:
+                # Migration failed, try to use the old token as-is
+                # This maintains backward compatibility
+                pass
+
+        # Use the standard refresh logic from the base class
+        if self.needs_refresh(credentials):
+            credentials = await self.refresh_tokens(credentials)
+
+        return credentials.access_token.get_secret_value()
+
+    def needs_migration(self, credentials: OAuth2Credentials) -> bool:
+        """
+        Check if credentials represent an old long-lived token that needs migration.
+
+        Old tokens have no expiration time and no refresh token.
+        """
+        return (
+            credentials.access_token_expires_at is None
+            and credentials.refresh_token is None
+        )
+
     async def _request_username(self, access_token: str) -> Optional[str]:
         # Use the LinearClient to fetch user details using GraphQL
         from ._api import LinearClient

autogpt_platform/backend/backend/blocks/linear/models.py

@@ -37,5 +37,5 @@ class Project(BaseModel):
     name: str
     description: str
     priority: int
-    progress: int
-    content: str
+    progress: float
+    content: str | None

autogpt_platform/backend/backend/blocks/llm.py

@@ -101,9 +101,9 @@ class LlmModel(str, Enum, metaclass=LlmModelMeta):
     CLAUDE_4_1_OPUS = "claude-opus-4-1-20250805"
     CLAUDE_4_OPUS = "claude-opus-4-20250514"
     CLAUDE_4_SONNET = "claude-sonnet-4-20250514"
+    CLAUDE_4_5_SONNET = "claude-sonnet-4-5-20250929"
+    CLAUDE_4_5_HAIKU = "claude-haiku-4-5-20251001"
     CLAUDE_3_7_SONNET = "claude-3-7-sonnet-20250219"
-    CLAUDE_3_5_SONNET = "claude-3-5-sonnet-latest"
-    CLAUDE_3_5_HAIKU = "claude-3-5-haiku-latest"
     CLAUDE_3_HAIKU = "claude-3-haiku-20240307"
     # AI/ML API models
     AIML_API_QWEN2_5_72B = "Qwen/Qwen2.5-72B-Instruct-Turbo"
@@ -213,15 +213,15 @@ MODEL_METADATA = {
     LlmModel.CLAUDE_4_SONNET: ModelMetadata(
         "anthropic", 200000, 64000
     ),  # claude-4-sonnet-20250514
+    LlmModel.CLAUDE_4_5_SONNET: ModelMetadata(
+        "anthropic", 200000, 64000
+    ),  # claude-sonnet-4-5-20250929
+    LlmModel.CLAUDE_4_5_HAIKU: ModelMetadata(
+        "anthropic", 200000, 64000
+    ),  # claude-haiku-4-5-20251001
     LlmModel.CLAUDE_3_7_SONNET: ModelMetadata(
         "anthropic", 200000, 64000
     ),  # claude-3-7-sonnet-20250219
-    LlmModel.CLAUDE_3_5_SONNET: ModelMetadata(
-        "anthropic", 200000, 8192
-    ),  # claude-3-5-sonnet-20241022
-    LlmModel.CLAUDE_3_5_HAIKU: ModelMetadata(
-        "anthropic", 200000, 8192
-    ),  # claude-3-5-haiku-20241022
     LlmModel.CLAUDE_3_HAIKU: ModelMetadata(
         "anthropic", 200000, 4096
     ),  # claude-3-haiku-20240307
@@ -1400,11 +1400,27 @@ class AITextSummarizerBlock(AIBlockBase):
 
     @staticmethod
     def _split_text(text: str, max_tokens: int, overlap: int) -> list[str]:
+        # Security fix: Add validation to prevent DoS attacks
+        # Limit text size to prevent memory exhaustion
+        MAX_TEXT_LENGTH = 1_000_000  # 1MB character limit
+        MAX_CHUNKS = 100  # Maximum number of chunks to prevent excessive memory use
+
+        if len(text) > MAX_TEXT_LENGTH:
+            text = text[:MAX_TEXT_LENGTH]
+
+        # Ensure chunk_size is at least 1 to prevent infinite loops
+        chunk_size = max(1, max_tokens - overlap)
+
+        # Ensure overlap is less than max_tokens to prevent invalid configurations
+        if overlap >= max_tokens:
+            overlap = max(0, max_tokens - 1)
+
         words = text.split()
         chunks = []
-        chunk_size = max_tokens - overlap
 
         for i in range(0, len(words), chunk_size):
+            if len(chunks) >= MAX_CHUNKS:
+                break  # Limit the number of chunks to prevent memory exhaustion
             chunk = " ".join(words[i : i + max_tokens])
             chunks.append(chunk)
 
@@ -1538,7 +1554,9 @@ class AIConversationBlock(AIBlockBase):
                 ("prompt", list),
             ],
             test_mock={
-                "llm_call": lambda *args, **kwargs: "The 2020 World Series was played at Globe Life Field in Arlington, Texas."
+                "llm_call": lambda *args, **kwargs: dict(
+                    response="The 2020 World Series was played at Globe Life Field in Arlington, Texas."
+                )
             },
         )
 
@@ -1567,7 +1585,7 @@ class AIConversationBlock(AIBlockBase):
             ),
             credentials=credentials,
         )
-        yield "response", response
+        yield "response", response["response"]
         yield "prompt", self.prompt
 
 

autogpt_platform/backend/backend/blocks/perplexity.py

@@ -0,0 +1,226 @@
+# flake8: noqa: E501
+import logging
+from enum import Enum
+from typing import Any, Literal
+
+import openai
+from pydantic import SecretStr
+
+from backend.data.block import Block, BlockCategory, BlockOutput, BlockSchema
+from backend.data.model import (
+    APIKeyCredentials,
+    CredentialsField,
+    CredentialsMetaInput,
+    NodeExecutionStats,
+    SchemaField,
+)
+from backend.integrations.providers import ProviderName
+from backend.util.logging import TruncatedLogger
+
+logger = TruncatedLogger(logging.getLogger(__name__), "[Perplexity-Block]")
+
+
+class PerplexityModel(str, Enum):
+    """Perplexity sonar models available via OpenRouter"""
+
+    SONAR = "perplexity/sonar"
+    SONAR_PRO = "perplexity/sonar-pro"
+    SONAR_DEEP_RESEARCH = "perplexity/sonar-deep-research"
+
+
+PerplexityCredentials = CredentialsMetaInput[
+    Literal[ProviderName.OPEN_ROUTER], Literal["api_key"]
+]
+
+TEST_CREDENTIALS = APIKeyCredentials(
+    id="test-perplexity-creds",
+    provider="open_router",
+    api_key=SecretStr("mock-openrouter-api-key"),
+    title="Mock OpenRouter API key",
+    expires_at=None,
+)
+TEST_CREDENTIALS_INPUT = {
+    "provider": TEST_CREDENTIALS.provider,
+    "id": TEST_CREDENTIALS.id,
+    "type": TEST_CREDENTIALS.type,
+    "title": TEST_CREDENTIALS.title,
+}
+
+
+def PerplexityCredentialsField() -> PerplexityCredentials:
+    return CredentialsField(
+        description="OpenRouter API key for accessing Perplexity models.",
+    )
+
+
+class PerplexityBlock(Block):
+    class Input(BlockSchema):
+        prompt: str = SchemaField(
+            description="The query to send to the Perplexity model.",
+            placeholder="Enter your query here...",
+        )
+        model: PerplexityModel = SchemaField(
+            title="Perplexity Model",
+            default=PerplexityModel.SONAR,
+            description="The Perplexity sonar model to use.",
+            advanced=False,
+        )
+        credentials: PerplexityCredentials = PerplexityCredentialsField()
+        system_prompt: str = SchemaField(
+            title="System Prompt",
+            default="",
+            description="Optional system prompt to provide context to the model.",
+            advanced=True,
+        )
+        max_tokens: int | None = SchemaField(
+            advanced=True,
+            default=None,
+            description="The maximum number of tokens to generate.",
+        )
+
+    class Output(BlockSchema):
+        response: str = SchemaField(
+            description="The response from the Perplexity model."
+        )
+        annotations: list[dict[str, Any]] = SchemaField(
+            description="List of URL citations and annotations from the response."
+        )
+        error: str = SchemaField(description="Error message if the API call failed.")
+
+    def __init__(self):
+        super().__init__(
+            id="c8a5f2e9-8b3d-4a7e-9f6c-1d5e3c9b7a4f",
+            description="Query Perplexity's sonar models with real-time web search capabilities and receive annotated responses with source citations.",
+            categories={BlockCategory.AI, BlockCategory.SEARCH},
+            input_schema=PerplexityBlock.Input,
+            output_schema=PerplexityBlock.Output,
+            test_input={
+                "prompt": "What is the weather today?",
+                "model": PerplexityModel.SONAR,
+                "credentials": TEST_CREDENTIALS_INPUT,
+            },
+            test_credentials=TEST_CREDENTIALS,
+            test_output=[
+                ("response", "The weather varies by location..."),
+                ("annotations", list),
+            ],
+            test_mock={
+                "call_perplexity": lambda *args, **kwargs: {
+                    "response": "The weather varies by location...",
+                    "annotations": [
+                        {
+                            "type": "url_citation",
+                            "url_citation": {
+                                "title": "weather.com",
+                                "url": "https://weather.com",
+                            },
+                        }
+                    ],
+                }
+            },
+        )
+        self.execution_stats = NodeExecutionStats()
+
+    async def call_perplexity(
+        self,
+        credentials: APIKeyCredentials,
+        model: PerplexityModel,
+        prompt: str,
+        system_prompt: str = "",
+        max_tokens: int | None = None,
+    ) -> dict[str, Any]:
+        """Call Perplexity via OpenRouter and extract annotations."""
+        client = openai.AsyncOpenAI(
+            base_url="https://openrouter.ai/api/v1",
+            api_key=credentials.api_key.get_secret_value(),
+        )
+
+        messages = []
+        if system_prompt:
+            messages.append({"role": "system", "content": system_prompt})
+        messages.append({"role": "user", "content": prompt})
+
+        try:
+            response = await client.chat.completions.create(
+                extra_headers={
+                    "HTTP-Referer": "https://agpt.co",
+                    "X-Title": "AutoGPT",
+                },
+                model=model.value,
+                messages=messages,
+                max_tokens=max_tokens,
+            )
+
+            if not response.choices:
+                raise ValueError("No response from Perplexity via OpenRouter.")
+
+            # Extract the response content
+            response_content = response.choices[0].message.content or ""
+
+            # Extract annotations if present in the message
+            annotations = []
+            if hasattr(response.choices[0].message, "annotations"):
+                # If annotations are directly available
+                annotations = response.choices[0].message.annotations
+            else:
+                # Check if there's a raw response with annotations
+                raw = getattr(response.choices[0].message, "_raw_response", None)
+                if isinstance(raw, dict) and "annotations" in raw:
+                    annotations = raw["annotations"]
+
+            if not annotations and hasattr(response, "model_extra"):
+                # Check model_extra for annotations
+                model_extra = response.model_extra
+                if isinstance(model_extra, dict):
+                    # Check in choices
+                    if "choices" in model_extra and len(model_extra["choices"]) > 0:
+                        choice = model_extra["choices"][0]
+                        if "message" in choice and "annotations" in choice["message"]:
+                            annotations = choice["message"]["annotations"]
+
+            # Also check the raw response object for annotations
+            if not annotations:
+                raw = getattr(response, "_raw_response", None)
+                if isinstance(raw, dict):
+                    # Check various possible locations for annotations
+                    if "annotations" in raw:
+                        annotations = raw["annotations"]
+                    elif "choices" in raw and len(raw["choices"]) > 0:
+                        choice = raw["choices"][0]
+                        if "message" in choice and "annotations" in choice["message"]:
+                            annotations = choice["message"]["annotations"]
+
+            # Update execution stats
+            if response.usage:
+                self.execution_stats.input_token_count = response.usage.prompt_tokens
+                self.execution_stats.output_token_count = (
+                    response.usage.completion_tokens
+                )
+
+            return {"response": response_content, "annotations": annotations or []}
+
+        except Exception as e:
+            logger.error(f"Error calling Perplexity: {e}")
+            raise
+
+    async def run(
+        self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs
+    ) -> BlockOutput:
+        logger.debug(f"Running Perplexity block with model: {input_data.model}")
+
+        try:
+            result = await self.call_perplexity(
+                credentials=credentials,
+                model=input_data.model,
+                prompt=input_data.prompt,
+                system_prompt=input_data.system_prompt,
+                max_tokens=input_data.max_tokens,
+            )
+
+            yield "response", result["response"]
+            yield "annotations", result["annotations"]
+
+        except Exception as e:
+            error_msg = f"Error calling Perplexity: {str(e)}"
+            logger.error(error_msg)
+            yield "error", error_msg

autogpt_platform/backend/backend/blocks/rss.py

@@ -1,4 +1,5 @@
 import asyncio
+import logging
 from datetime import datetime, timedelta, timezone
 from typing import Any
 
@@ -7,6 +8,7 @@ import pydantic
 
 from backend.data.block import Block, BlockCategory, BlockOutput, BlockSchema
 from backend.data.model import SchemaField
+from backend.util.request import Requests
 
 
 class RSSEntry(pydantic.BaseModel):
@@ -100,8 +102,33 @@ class ReadRSSFeedBlock(Block):
         )
 
     @staticmethod
-    def parse_feed(url: str) -> dict[str, Any]:
-        return feedparser.parse(url)  # type: ignore
+    async def parse_feed(url: str) -> dict[str, Any]:
+        # Security fix: Add protection against memory exhaustion attacks
+        MAX_FEED_SIZE = 10 * 1024 * 1024  # 10MB limit for RSS feeds
+
+        # Download feed content with size limit
+        try:
+            response = await Requests(raise_for_status=True).get(url)
+
+            # Check content length if available
+            content_length = response.headers.get("Content-Length")
+            if content_length and int(content_length) > MAX_FEED_SIZE:
+                raise ValueError(
+                    f"Feed too large: {content_length} bytes exceeds {MAX_FEED_SIZE} limit"
+                )
+
+            # Get content with size limit
+            content = response.content
+            if len(content) > MAX_FEED_SIZE:
+                raise ValueError(f"Feed too large: exceeds {MAX_FEED_SIZE} byte limit")
+
+            # Parse with feedparser using the validated content
+            # feedparser has built-in protection against XML attacks
+            return feedparser.parse(content)  # type: ignore
+        except Exception as e:
+            # Log error and return empty feed
+            logging.warning(f"Failed to parse RSS feed from {url}: {e}")
+            return {"entries": []}
 
     async def run(self, input_data: Input, **kwargs) -> BlockOutput:
         keep_going = True
@@ -111,7 +138,7 @@ class ReadRSSFeedBlock(Block):
         while keep_going:
             keep_going = input_data.run_continuously
 
-            feed = self.parse_feed(input_data.rss_url)
+            feed = await self.parse_feed(input_data.rss_url)
             all_entries = []
 
             for entry in feed["entries"]:

autogpt_platform/backend/backend/blocks/smart_decision_maker.py

@@ -13,6 +13,11 @@ from backend.data.block import (
     BlockSchema,
     BlockType,
 )
+from backend.data.dynamic_fields import (
+    extract_base_field_name,
+    get_dynamic_field_description,
+    is_dynamic_field,
+)
 from backend.data.model import NodeExecutionStats, SchemaField
 from backend.util import json
 from backend.util.clients import get_database_manager_async_client
@@ -98,6 +103,22 @@ def _create_tool_response(call_id: str, output: Any) -> dict[str, Any]:
     return {"role": "tool", "tool_call_id": call_id, "content": content}
 
 
+def _convert_raw_response_to_dict(raw_response: Any) -> dict[str, Any]:
+    """
+    Safely convert raw_response to dictionary format for conversation history.
+    Handles different response types from different LLM providers.
+    """
+    if isinstance(raw_response, str):
+        # Ollama returns a string, convert to dict format
+        return {"role": "assistant", "content": raw_response}
+    elif isinstance(raw_response, dict):
+        # Already a dict (from tests or some providers)
+        return raw_response
+    else:
+        # OpenAI/Anthropic return objects, convert with json.to_dict
+        return json.to_dict(raw_response)
+
+
 def get_pending_tool_calls(conversation_history: list[Any]) -> dict[str, int]:
     """
     All the tool calls entry in the conversation history requires a response.
@@ -261,6 +282,7 @@ class SmartDecisionMakerBlock(Block):
 
     @staticmethod
     def cleanup(s: str):
+        """Clean up block names for use as tool function names."""
         return re.sub(r"[^a-zA-Z0-9_-]", "_", s).lower()
 
     @staticmethod
@@ -288,41 +310,66 @@ class SmartDecisionMakerBlock(Block):
         }
         sink_block_input_schema = block.input_schema
         properties = {}
+        field_mapping = {}  # clean_name -> original_name
 
         for link in links:
-            sink_name = SmartDecisionMakerBlock.cleanup(link.sink_name)
+            field_name = link.sink_name
+            is_dynamic = is_dynamic_field(field_name)
+            # Clean property key to ensure Anthropic API compatibility for ALL fields
+            clean_field_name = SmartDecisionMakerBlock.cleanup(field_name)
+            field_mapping[clean_field_name] = field_name
 
-            # Handle dynamic fields (e.g., values_#_*, items_$_*, etc.)
-            # These are fields that get merged by the executor into their base field
-            if (
-                "_#_" in link.sink_name
-                or "_$_" in link.sink_name
-                or "_@_" in link.sink_name
-            ):
-                # For dynamic fields, provide a generic string schema
-                # The executor will handle merging these into the appropriate structure
-                properties[sink_name] = {
+            if is_dynamic:
+                # For dynamic fields, use cleaned name but preserve original in description
+                properties[clean_field_name] = {
                     "type": "string",
-                    "description": f"Dynamic value for {link.sink_name}",
+                    "description": get_dynamic_field_description(field_name),
                 }
             else:
-                # For regular fields, use the block's schema
+                # For regular fields, use the block's schema directly
                 try:
-                    properties[sink_name] = sink_block_input_schema.get_field_schema(
-                        link.sink_name
+                    properties[clean_field_name] = (
+                        sink_block_input_schema.get_field_schema(field_name)
                     )
                 except (KeyError, AttributeError):
-                    # If the field doesn't exist in the schema, provide a generic schema
-                    properties[sink_name] = {
+                    # If field doesn't exist in schema, provide a generic one
+                    properties[clean_field_name] = {
                         "type": "string",
-                        "description": f"Value for {link.sink_name}",
+                        "description": f"Value for {field_name}",
                     }
 
+        # Build the parameters schema using a single unified path
+        base_schema = block.input_schema.jsonschema()
+        base_required = set(base_schema.get("required", []))
+
+        # Compute required fields at the leaf level:
+        # - If a linked field is dynamic and its base is required in the block schema, require the leaf
+        # - If a linked field is regular and is required in the block schema, require the leaf
+        required_fields: set[str] = set()
+        for link in links:
+            field_name = link.sink_name
+            is_dynamic = is_dynamic_field(field_name)
+            # Always use cleaned field name for property key (Anthropic API compliance)
+            clean_field_name = SmartDecisionMakerBlock.cleanup(field_name)
+
+            if is_dynamic:
+                base_name = extract_base_field_name(field_name)
+                if base_name in base_required:
+                    required_fields.add(clean_field_name)
+            else:
+                if field_name in base_required:
+                    required_fields.add(clean_field_name)
+
         tool_function["parameters"] = {
-            **block.input_schema.jsonschema(),
+            "type": "object",
             "properties": properties,
+            "additionalProperties": False,
+            "required": sorted(required_fields),
         }
 
+        # Store field mapping for later use in output processing
+        tool_function["_field_mapping"] = field_mapping
+
         return {"type": "function", "function": tool_function}
 
     @staticmethod
@@ -366,13 +413,12 @@ class SmartDecisionMakerBlock(Block):
             sink_block_properties = sink_block_input_schema.get("properties", {}).get(
                 link.sink_name, {}
             )
-            sink_name = SmartDecisionMakerBlock.cleanup(link.sink_name)
             description = (
                 sink_block_properties["description"]
                 if "description" in sink_block_properties
                 else f"The {link.sink_name} of the tool"
             )
-            properties[sink_name] = {
+            properties[link.sink_name] = {
                 "type": "string",
                 "description": description,
                 "default": json.dumps(sink_block_properties.get("default", None)),
@@ -388,24 +434,17 @@ class SmartDecisionMakerBlock(Block):
         return {"type": "function", "function": tool_function}
 
     @staticmethod
-    async def _create_function_signature(node_id: str) -> list[dict[str, Any]]:
+    async def _create_function_signature(
+        node_id: str,
+    ) -> list[dict[str, Any]]:
         """
-        Creates function signatures for tools linked to a specified node within a graph.
-
-        This method filters the graph links to identify those that are tools and are
-        connected to the given node_id. It then constructs function signatures for each
-        tool based on the metadata and input schema of the linked nodes.
+        Creates function signatures for connected tools.
 
         Args:
             node_id: The node_id for which to create function signatures.
 
         Returns:
-            list[dict[str, Any]]: A list of dictionaries, each representing a function signature
-                                  for a tool, including its name, description, and parameters.
-
-        Raises:
-            ValueError: If no tool links are found for the specified node_id, or if a sink node
-                        or its metadata cannot be found.
+            List of function signatures for tools
         """
         db_client = get_database_manager_async_client()
         tools = [
@@ -430,20 +469,116 @@ class SmartDecisionMakerBlock(Block):
                 raise ValueError(f"Sink node not found: {links[0].sink_id}")
 
             if sink_node.block_id == AgentExecutorBlock().id:
-                return_tool_functions.append(
+                tool_func = (
                     await SmartDecisionMakerBlock._create_agent_function_signature(
                         sink_node, links
                     )
                 )
+                return_tool_functions.append(tool_func)
             else:
-                return_tool_functions.append(
+                tool_func = (
                     await SmartDecisionMakerBlock._create_block_function_signature(
                         sink_node, links
                     )
                 )
+                return_tool_functions.append(tool_func)
 
         return return_tool_functions
 
+    async def _attempt_llm_call_with_validation(
+        self,
+        credentials: llm.APIKeyCredentials,
+        input_data: Input,
+        current_prompt: list[dict],
+        tool_functions: list[dict[str, Any]],
+    ):
+        """
+        Attempt a single LLM call with tool validation.
+
+        Returns the response if successful, raises ValueError if validation fails.
+        """
+        resp = await llm.llm_call(
+            credentials=credentials,
+            llm_model=input_data.model,
+            prompt=current_prompt,
+            max_tokens=input_data.max_tokens,
+            tools=tool_functions,
+            ollama_host=input_data.ollama_host,
+            parallel_tool_calls=input_data.multiple_tool_calls,
+        )
+
+        # Track LLM usage stats per call
+        self.merge_stats(
+            NodeExecutionStats(
+                input_token_count=resp.prompt_tokens,
+                output_token_count=resp.completion_tokens,
+                llm_call_count=1,
+            )
+        )
+
+        if not resp.tool_calls:
+            return resp
+        validation_errors_list: list[str] = []
+        for tool_call in resp.tool_calls:
+            tool_name = tool_call.function.name
+            try:
+                tool_args = json.loads(tool_call.function.arguments)
+            except Exception as e:
+                validation_errors_list.append(
+                    f"Tool call '{tool_name}' has invalid JSON arguments: {e}"
+                )
+                continue
+
+            # Find the tool definition to get the expected arguments
+            tool_def = next(
+                (
+                    tool
+                    for tool in tool_functions
+                    if tool["function"]["name"] == tool_name
+                ),
+                None,
+            )
+            if tool_def is None and len(tool_functions) == 1:
+                tool_def = tool_functions[0]
+
+            # Get parameters schema from tool definition
+            if (
+                tool_def
+                and "function" in tool_def
+                and "parameters" in tool_def["function"]
+            ):
+                parameters = tool_def["function"]["parameters"]
+                expected_args = parameters.get("properties", {})
+                required_params = set(parameters.get("required", []))
+            else:
+                expected_args = {arg: {} for arg in tool_args.keys()}
+                required_params = set()
+
+            # Validate tool call arguments
+            provided_args = set(tool_args.keys())
+            expected_args_set = set(expected_args.keys())
+
+            # Check for unexpected arguments (typos)
+            unexpected_args = provided_args - expected_args_set
+            # Only check for missing REQUIRED parameters
+            missing_required_args = required_params - provided_args
+
+            if unexpected_args or missing_required_args:
+                error_msg = f"Tool call '{tool_name}' has parameter errors:"
+                if unexpected_args:
+                    error_msg += f" Unknown parameters: {sorted(unexpected_args)}."
+                if missing_required_args:
+                    error_msg += f" Missing required parameters: {sorted(missing_required_args)}."
+                error_msg += f" Expected parameters: {sorted(expected_args_set)}."
+                if required_params:
+                    error_msg += f" Required parameters: {sorted(required_params)}."
+                validation_errors_list.append(error_msg)
+
+        if validation_errors_list:
+            raise ValueError("; ".join(validation_errors_list))
+
+        return resp
+
     async def run(
         self,
         input_data: Input,
@@ -466,27 +601,19 @@ class SmartDecisionMakerBlock(Block):
         if pending_tool_calls and input_data.last_tool_output is None:
             raise ValueError(f"Tool call requires an output for {pending_tool_calls}")
 
-        # Only assign the last tool output to the first pending tool call
         tool_output = []
         if pending_tool_calls and input_data.last_tool_output is not None:
-            # Get the first pending tool call ID
             first_call_id = next(iter(pending_tool_calls.keys()))
             tool_output.append(
                 _create_tool_response(first_call_id, input_data.last_tool_output)
             )
 
-            # Add tool output to prompt right away
             prompt.extend(tool_output)
-
-            # Check if there are still pending tool calls after handling the first one
             remaining_pending_calls = get_pending_tool_calls(prompt)
 
-            # If there are still pending tool calls, yield the conversation and return early
             if remaining_pending_calls:
                 yield "conversations", prompt
                 return
-
-        # Fallback on adding tool output in the conversation history as user prompt.
         elif input_data.last_tool_output:
             logger.error(
                 f"[SmartDecisionMakerBlock-node_exec_id={node_exec_id}] "
@@ -519,24 +646,33 @@ class SmartDecisionMakerBlock(Block):
         ):
             prompt.append({"role": "user", "content": prefix + input_data.prompt})
 
-        response = await llm.llm_call(
-            credentials=credentials,
-            llm_model=input_data.model,
-            prompt=prompt,
-            max_tokens=input_data.max_tokens,
-            tools=tool_functions,
-            ollama_host=input_data.ollama_host,
-            parallel_tool_calls=input_data.multiple_tool_calls,
-        )
+        current_prompt = list(prompt)
+        max_attempts = max(1, int(input_data.retry))
+        response = None
 
-        # Track LLM usage stats
-        self.merge_stats(
-            NodeExecutionStats(
-                input_token_count=response.prompt_tokens,
-                output_token_count=response.completion_tokens,
-                llm_call_count=1,
+        last_error = None
+        for attempt in range(max_attempts):
+            try:
+                response = await self._attempt_llm_call_with_validation(
+                    credentials, input_data, current_prompt, tool_functions
+                )
+                break
+
+            except ValueError as e:
+                last_error = e
+                error_feedback = (
+                    "Your tool call had parameter errors. Please fix the following issues and try again:\n"
+                    + f"- {str(e)}\n"
+                    + "\nPlease make sure to use the exact parameter names as specified in the function schema."
+                )
+                current_prompt = list(current_prompt) + [
+                    {"role": "user", "content": error_feedback}
+                ]
+
+        if response is None:
+            raise last_error or ValueError(
+                "Failed to get valid response after all retry attempts"
             )
-        )
 
         if not response.tool_calls:
             yield "finished", response.response
@@ -546,7 +682,6 @@ class SmartDecisionMakerBlock(Block):
             tool_name = tool_call.function.name
             tool_args = json.loads(tool_call.function.arguments)
 
-            # Find the tool definition to get the expected arguments
             tool_def = next(
                 (
                     tool
@@ -555,7 +690,6 @@ class SmartDecisionMakerBlock(Block):
                 ),
                 None,
             )
-
             if (
                 tool_def
                 and "function" in tool_def
@@ -563,20 +697,38 @@ class SmartDecisionMakerBlock(Block):
             ):
                 expected_args = tool_def["function"]["parameters"].get("properties", {})
             else:
-                expected_args = tool_args.keys()
+                expected_args = {arg: {} for arg in tool_args.keys()}
 
-            # Yield provided arguments and None for missing ones
-            for arg_name in expected_args:
-                if arg_name in tool_args:
-                    yield f"tools_^_{tool_name}_~_{arg_name}", tool_args[arg_name]
-                else:
-                    yield f"tools_^_{tool_name}_~_{arg_name}", None
+            # Get field mapping from tool definition
+            field_mapping = (
+                tool_def.get("function", {}).get("_field_mapping", {})
+                if tool_def
+                else {}
+            )
+
+            for clean_arg_name in expected_args:
+                # arg_name is now always the cleaned field name (for Anthropic API compliance)
+                # Get the original field name from field mapping for proper emit key generation
+                original_field_name = field_mapping.get(clean_arg_name, clean_arg_name)
+                arg_value = tool_args.get(clean_arg_name)
+
+                sanitized_tool_name = self.cleanup(tool_name)
+                sanitized_arg_name = self.cleanup(original_field_name)
+                emit_key = f"tools_^_{sanitized_tool_name}_~_{sanitized_arg_name}"
+
+                logger.debug(
+                    "[SmartDecisionMakerBlock|geid:%s|neid:%s] emit %s",
+                    graph_exec_id,
+                    node_exec_id,
+                    emit_key,
+                )
+                yield emit_key, arg_value
 
-        # Add reasoning to conversation history if available
         if response.reasoning:
             prompt.append(
                 {"role": "assistant", "content": f"[Reasoning]: {response.reasoning}"}
             )
 
-        prompt.append(response.raw_response)
+        prompt.append(_convert_raw_response_to_dict(response.raw_response))
+
         yield "conversations", prompt

autogpt_platform/backend/backend/blocks/stagehand/blocks.py

@@ -1,6 +1,7 @@
 import logging
 import signal
 import threading
+import warnings
 from contextlib import contextmanager
 from enum import Enum
 
@@ -26,6 +27,13 @@ from backend.sdk import (
     SchemaField,
 )
 
+# Suppress false positive cleanup warning of litellm (a dependency of stagehand)
+warnings.filterwarnings(
+    "ignore",
+    message="coroutine 'close_litellm_async_clients' was never awaited",
+    category=RuntimeWarning,
+)
+
 # Store the original method
 original_register_signal_handlers = stagehand.main.Stagehand._register_signal_handlers
 

autogpt_platform/backend/backend/blocks/test/test_block.py

@@ -19,7 +19,7 @@ async def test_block_ids_valid(block: Type[Block]):
     # Skip list for blocks with known invalid UUIDs
     skip_blocks = {
         "GetWeatherInformationBlock",
-        "CodeExecutionBlock",
+        "ExecuteCodeBlock",
         "CountdownTimerBlock",
         "TwitterGetListTweetsBlock",
         "TwitterRemoveListMemberBlock",

autogpt_platform/backend/backend/blocks/test/test_blocks_dos_vulnerability.py

@@ -0,0 +1,269 @@
+"""
+Test security fixes for various DoS vulnerabilities.
+"""
+
+import asyncio
+from unittest.mock import patch
+
+import pytest
+
+from backend.blocks.code_extraction_block import CodeExtractionBlock
+from backend.blocks.iteration import StepThroughItemsBlock
+from backend.blocks.llm import AITextSummarizerBlock
+from backend.blocks.text import ExtractTextInformationBlock
+from backend.blocks.xml_parser import XMLParserBlock
+from backend.util.file import store_media_file
+from backend.util.type import MediaFileType
+
+
+class TestCodeExtractionBlockSecurity:
+    """Test ReDoS fixes in CodeExtractionBlock."""
+
+    async def test_redos_protection(self):
+        """Test that the regex patterns don't cause ReDoS."""
+        block = CodeExtractionBlock()
+
+        # Test with input that would previously cause ReDoS
+        malicious_input = "```python" + " " * 10000  # Large spaces
+
+        result = []
+        async for output_name, output_data in block.run(
+            CodeExtractionBlock.Input(text=malicious_input)
+        ):
+            result.append((output_name, output_data))
+
+        # Should complete without hanging
+        assert len(result) >= 1
+        assert any(name == "remaining_text" for name, _ in result)
+
+
+class TestAITextSummarizerBlockSecurity:
+    """Test memory exhaustion fixes in AITextSummarizerBlock."""
+
+    def test_split_text_limits(self):
+        """Test that _split_text has proper limits."""
+        # Test text size limit
+        large_text = "a" * 2_000_000  # 2MB text
+        result = AITextSummarizerBlock._split_text(large_text, 1000, 100)
+
+        # Should be truncated to 1MB
+        total_chars = sum(len(chunk) for chunk in result)
+        assert total_chars <= 1_000_000 + 1000  # Allow for chunk boundary
+
+        # Test chunk count limit
+        result = AITextSummarizerBlock._split_text("word " * 10000, 10, 9)
+        assert len(result) <= 100  # MAX_CHUNKS limit
+
+        # Test parameter validation
+        result = AITextSummarizerBlock._split_text(
+            "test", 10, 15
+        )  # overlap > max_tokens
+        assert len(result) >= 1  # Should still work
+
+
+class TestExtractTextInformationBlockSecurity:
+    """Test ReDoS and memory exhaustion fixes in ExtractTextInformationBlock."""
+
+    async def test_text_size_limits(self):
+        """Test text size limits."""
+        block = ExtractTextInformationBlock()
+
+        # Test with large input
+        large_text = "a" * 2_000_000  # 2MB
+
+        results = []
+        async for output_name, output_data in block.run(
+            ExtractTextInformationBlock.Input(
+                text=large_text, pattern=r"a+", find_all=True, group=0
+            )
+        ):
+            results.append((output_name, output_data))
+
+        # Should complete and have limits applied
+        matched_results = [r for name, r in results if name == "matched_results"]
+        if matched_results:
+            assert len(matched_results[0]) <= 1000  # MAX_MATCHES limit
+
+    async def test_dangerous_pattern_timeout(self):
+        """Test timeout protection for dangerous patterns."""
+        block = ExtractTextInformationBlock()
+
+        # Test with potentially dangerous lookahead pattern
+        test_input = "a" * 1000
+
+        # This should complete quickly due to timeout protection
+        start_time = asyncio.get_event_loop().time()
+        results = []
+        async for output_name, output_data in block.run(
+            ExtractTextInformationBlock.Input(
+                text=test_input, pattern=r"(?=.+)", find_all=True, group=0
+            )
+        ):
+            results.append((output_name, output_data))
+
+        end_time = asyncio.get_event_loop().time()
+        # Should complete within reasonable time (much less than 5s timeout)
+        assert (end_time - start_time) < 10
+
+    async def test_redos_catastrophic_backtracking(self):
+        """Test that ReDoS patterns with catastrophic backtracking are handled."""
+        block = ExtractTextInformationBlock()
+
+        # Pattern that causes catastrophic backtracking: (a+)+b
+        # With input "aaaaaaaaaaaaaaaaaaaaaaaaaaaa" (no 'b'), this causes exponential time
+        dangerous_pattern = r"(a+)+b"
+        test_input = "a" * 30  # 30 'a's without a 'b' at the end
+
+        # This should be handled by timeout protection or pattern detection
+        start_time = asyncio.get_event_loop().time()
+        results = []
+
+        async for output_name, output_data in block.run(
+            ExtractTextInformationBlock.Input(
+                text=test_input, pattern=dangerous_pattern, find_all=True, group=0
+            )
+        ):
+            results.append((output_name, output_data))
+
+        end_time = asyncio.get_event_loop().time()
+        elapsed = end_time - start_time
+
+        # Should complete within timeout (6 seconds to be safe)
+        # The current threading.Timer approach doesn't work, so this will likely fail
+        # demonstrating the need for a fix
+        assert elapsed < 6, f"Regex took {elapsed}s, timeout mechanism failed"
+
+        # Should return empty results on timeout or no match
+        matched_results = [r for name, r in results if name == "matched_results"]
+        assert matched_results[0] == []  # No matches expected
+
+
+class TestStepThroughItemsBlockSecurity:
+    """Test iteration limits in StepThroughItemsBlock."""
+
+    async def test_item_count_limits(self):
+        """Test maximum item count limits."""
+        block = StepThroughItemsBlock()
+
+        # Test with too many items
+        large_list = list(range(20000))  # Exceeds MAX_ITEMS (10000)
+
+        with pytest.raises(ValueError, match="Too many items"):
+            async for _ in block.run(StepThroughItemsBlock.Input(items=large_list)):
+                pass
+
+    async def test_string_size_limits(self):
+        """Test string input size limits."""
+        block = StepThroughItemsBlock()
+
+        # Test with large JSON string
+        large_string = '["item"]' * 200000  # Large JSON string
+
+        with pytest.raises(ValueError, match="Input too large"):
+            async for _ in block.run(
+                StepThroughItemsBlock.Input(items_str=large_string)
+            ):
+                pass
+
+    async def test_normal_iteration_works(self):
+        """Test that normal iteration still works."""
+        block = StepThroughItemsBlock()
+
+        results = []
+        async for output_name, output_data in block.run(
+            StepThroughItemsBlock.Input(items=[1, 2, 3])
+        ):
+            results.append((output_name, output_data))
+
+        # Should have 6 outputs (item, key for each of 3 items)
+        assert len(results) == 6
+        items = [data for name, data in results if name == "item"]
+        assert items == [1, 2, 3]
+
+
+class TestXMLParserBlockSecurity:
+    """Test XML size limits in XMLParserBlock."""
+
+    async def test_xml_size_limits(self):
+        """Test XML input size limits."""
+        block = XMLParserBlock()
+
+        # Test with large XML - need to exceed 10MB limit
+        # Each "<item>data</item>" is 17 chars, need ~620K items for >10MB
+        large_xml = "<root>" + "<item>data</item>" * 620000 + "</root>"
+
+        with pytest.raises(ValueError, match="XML too large"):
+            async for _ in block.run(XMLParserBlock.Input(input_xml=large_xml)):
+                pass
+
+
+class TestStoreMediaFileSecurity:
+    """Test file storage security limits."""
+
+    @patch("backend.util.file.scan_content_safe")
+    @patch("backend.util.file.get_cloud_storage_handler")
+    async def test_file_size_limits(self, mock_cloud_storage, mock_scan):
+        """Test file size limits."""
+        # Mock cloud storage handler - get_cloud_storage_handler is async
+        # but is_cloud_path and parse_cloud_path are sync methods
+        from unittest.mock import MagicMock
+
+        mock_handler = MagicMock()
+        mock_handler.is_cloud_path.return_value = False
+
+        # Make get_cloud_storage_handler an async function that returns the mock handler
+        async def async_get_handler():
+            return mock_handler
+
+        mock_cloud_storage.side_effect = async_get_handler
+        mock_scan.return_value = None
+
+        # Test with large base64 content
+        large_content = "a" * (200 * 1024 * 1024)  # 200MB
+        large_data_uri = f"data:text/plain;base64,{large_content}"
+
+        with pytest.raises(ValueError, match="File too large"):
+            await store_media_file(
+                graph_exec_id="test",
+                file=MediaFileType(large_data_uri),
+                user_id="test_user",
+            )
+
+    @patch("backend.util.file.Path")
+    @patch("backend.util.file.scan_content_safe")
+    @patch("backend.util.file.get_cloud_storage_handler")
+    async def test_directory_size_limits(self, mock_cloud_storage, mock_scan, MockPath):
+        """Test directory size limits."""
+        from unittest.mock import MagicMock
+
+        mock_handler = MagicMock()
+        mock_handler.is_cloud_path.return_value = False
+
+        async def async_get_handler():
+            return mock_handler
+
+        mock_cloud_storage.side_effect = async_get_handler
+        mock_scan.return_value = None
+
+        # Create mock path instance for the execution directory
+        mock_path_instance = MagicMock()
+        mock_path_instance.exists.return_value = True
+
+        # Mock glob to return files that total > 1GB
+        mock_file = MagicMock()
+        mock_file.is_file.return_value = True
+        mock_file.stat.return_value.st_size = 2 * 1024 * 1024 * 1024  # 2GB
+        mock_path_instance.glob.return_value = [mock_file]
+
+        # Make Path() return our mock
+        MockPath.return_value = mock_path_instance
+
+        # Should raise an error when directory size exceeds limit
+        with pytest.raises(ValueError, match="Disk usage limit exceeded"):
+            await store_media_file(
+                graph_exec_id="test",
+                file=MediaFileType(
+                    "data:text/plain;base64,dGVzdA=="
+                ),  # Small test file
+                user_id="test_user",
+            )

autogpt_platform/backend/backend/blocks/test/test_llm.py

@@ -362,7 +362,7 @@ class TestLLMStatsTracking:
         assert block.execution_stats.llm_call_count == 1
 
         # Check output
-        assert outputs["response"] == {"response": "AI response to conversation"}
+        assert outputs["response"] == "AI response to conversation"
 
     @pytest.mark.asyncio
     async def test_ai_list_generator_with_retries(self):

autogpt_platform/backend/backend/blocks/test/test_smart_decision_maker.py

@@ -216,8 +216,17 @@ async def test_smart_decision_maker_tracks_llm_stats():
     }
 
     # Mock the _create_function_signature method to avoid database calls
-    with patch("backend.blocks.llm.llm_call", return_value=mock_response), patch.object(
-        SmartDecisionMakerBlock, "_create_function_signature", return_value=[]
+    from unittest.mock import AsyncMock
+
+    with patch(
+        "backend.blocks.llm.llm_call",
+        new_callable=AsyncMock,
+        return_value=mock_response,
+    ), patch.object(
+        SmartDecisionMakerBlock,
+        "_create_function_signature",
+        new_callable=AsyncMock,
+        return_value=[],
     ):
 
         # Create test input
@@ -249,3 +258,471 @@ async def test_smart_decision_maker_tracks_llm_stats():
         # Verify outputs
         assert "finished" in outputs  # Should have finished since no tool calls
         assert outputs["finished"] == "I need to think about this."
+
+
+@pytest.mark.asyncio
+async def test_smart_decision_maker_parameter_validation():
+    """Test that SmartDecisionMakerBlock correctly validates tool call parameters."""
+    from unittest.mock import MagicMock, patch
+
+    import backend.blocks.llm as llm_module
+    from backend.blocks.smart_decision_maker import SmartDecisionMakerBlock
+
+    block = SmartDecisionMakerBlock()
+
+    # Mock tool functions with specific parameter schema
+    mock_tool_functions = [
+        {
+            "type": "function",
+            "function": {
+                "name": "search_keywords",
+                "description": "Search for keywords with difficulty filtering",
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "query": {"type": "string", "description": "Search query"},
+                        "max_keyword_difficulty": {
+                            "type": "integer",
+                            "description": "Maximum keyword difficulty (required)",
+                        },
+                        "optional_param": {
+                            "type": "string",
+                            "description": "Optional parameter with default",
+                            "default": "default_value",
+                        },
+                    },
+                    "required": ["query", "max_keyword_difficulty"],
+                },
+            },
+        }
+    ]
+
+    # Test case 1: Tool call with TYPO in parameter name (should retry and eventually fail)
+    mock_tool_call_with_typo = MagicMock()
+    mock_tool_call_with_typo.function.name = "search_keywords"
+    mock_tool_call_with_typo.function.arguments = '{"query": "test", "maximum_keyword_difficulty": 50}'  # TYPO: maximum instead of max
+
+    mock_response_with_typo = MagicMock()
+    mock_response_with_typo.response = None
+    mock_response_with_typo.tool_calls = [mock_tool_call_with_typo]
+    mock_response_with_typo.prompt_tokens = 50
+    mock_response_with_typo.completion_tokens = 25
+    mock_response_with_typo.reasoning = None
+    mock_response_with_typo.raw_response = {"role": "assistant", "content": None}
+
+    from unittest.mock import AsyncMock
+
+    with patch(
+        "backend.blocks.llm.llm_call",
+        new_callable=AsyncMock,
+        return_value=mock_response_with_typo,
+    ) as mock_llm_call, patch.object(
+        SmartDecisionMakerBlock,
+        "_create_function_signature",
+        new_callable=AsyncMock,
+        return_value=mock_tool_functions,
+    ):
+
+        input_data = SmartDecisionMakerBlock.Input(
+            prompt="Search for keywords",
+            model=llm_module.LlmModel.GPT4O,
+            credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
+            retry=2,  # Set retry to 2 for testing
+        )
+
+        # Should raise ValueError after retries due to typo'd parameter name
+        with pytest.raises(ValueError) as exc_info:
+            outputs = {}
+            async for output_name, output_data in block.run(
+                input_data,
+                credentials=llm_module.TEST_CREDENTIALS,
+                graph_id="test-graph-id",
+                node_id="test-node-id",
+                graph_exec_id="test-exec-id",
+                node_exec_id="test-node-exec-id",
+                user_id="test-user-id",
+            ):
+                outputs[output_name] = output_data
+
+        # Verify error message contains details about the typo
+        error_msg = str(exc_info.value)
+        assert "Tool call 'search_keywords' has parameter errors" in error_msg
+        assert "Unknown parameters: ['maximum_keyword_difficulty']" in error_msg
+
+        # Verify that LLM was called the expected number of times (retries)
+        assert mock_llm_call.call_count == 2  # Should retry based on input_data.retry
+
+    # Test case 2: Tool call missing REQUIRED parameter (should raise ValueError)
+    mock_tool_call_missing_required = MagicMock()
+    mock_tool_call_missing_required.function.name = "search_keywords"
+    mock_tool_call_missing_required.function.arguments = (
+        '{"query": "test"}'  # Missing required max_keyword_difficulty
+    )
+
+    mock_response_missing_required = MagicMock()
+    mock_response_missing_required.response = None
+    mock_response_missing_required.tool_calls = [mock_tool_call_missing_required]
+    mock_response_missing_required.prompt_tokens = 50
+    mock_response_missing_required.completion_tokens = 25
+    mock_response_missing_required.reasoning = None
+    mock_response_missing_required.raw_response = {"role": "assistant", "content": None}
+
+    from unittest.mock import AsyncMock
+
+    with patch(
+        "backend.blocks.llm.llm_call",
+        new_callable=AsyncMock,
+        return_value=mock_response_missing_required,
+    ), patch.object(
+        SmartDecisionMakerBlock,
+        "_create_function_signature",
+        new_callable=AsyncMock,
+        return_value=mock_tool_functions,
+    ):
+
+        input_data = SmartDecisionMakerBlock.Input(
+            prompt="Search for keywords",
+            model=llm_module.LlmModel.GPT4O,
+            credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
+        )
+
+        # Should raise ValueError due to missing required parameter
+        with pytest.raises(ValueError) as exc_info:
+            outputs = {}
+            async for output_name, output_data in block.run(
+                input_data,
+                credentials=llm_module.TEST_CREDENTIALS,
+                graph_id="test-graph-id",
+                node_id="test-node-id",
+                graph_exec_id="test-exec-id",
+                node_exec_id="test-node-exec-id",
+                user_id="test-user-id",
+            ):
+                outputs[output_name] = output_data
+
+        error_msg = str(exc_info.value)
+        assert "Tool call 'search_keywords' has parameter errors" in error_msg
+        assert "Missing required parameters: ['max_keyword_difficulty']" in error_msg
+
+    # Test case 3: Valid tool call with OPTIONAL parameter missing (should succeed)
+    mock_tool_call_valid = MagicMock()
+    mock_tool_call_valid.function.name = "search_keywords"
+    mock_tool_call_valid.function.arguments = '{"query": "test", "max_keyword_difficulty": 50}'  # optional_param missing, but that's OK
+
+    mock_response_valid = MagicMock()
+    mock_response_valid.response = None
+    mock_response_valid.tool_calls = [mock_tool_call_valid]
+    mock_response_valid.prompt_tokens = 50
+    mock_response_valid.completion_tokens = 25
+    mock_response_valid.reasoning = None
+    mock_response_valid.raw_response = {"role": "assistant", "content": None}
+
+    from unittest.mock import AsyncMock
+
+    with patch(
+        "backend.blocks.llm.llm_call",
+        new_callable=AsyncMock,
+        return_value=mock_response_valid,
+    ), patch.object(
+        SmartDecisionMakerBlock,
+        "_create_function_signature",
+        new_callable=AsyncMock,
+        return_value=mock_tool_functions,
+    ):
+
+        input_data = SmartDecisionMakerBlock.Input(
+            prompt="Search for keywords",
+            model=llm_module.LlmModel.GPT4O,
+            credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
+        )
+
+        # Should succeed - optional parameter missing is OK
+        outputs = {}
+        async for output_name, output_data in block.run(
+            input_data,
+            credentials=llm_module.TEST_CREDENTIALS,
+            graph_id="test-graph-id",
+            node_id="test-node-id",
+            graph_exec_id="test-exec-id",
+            node_exec_id="test-node-exec-id",
+            user_id="test-user-id",
+        ):
+            outputs[output_name] = output_data
+
+        # Verify tool outputs were generated correctly
+        assert "tools_^_search_keywords_~_query" in outputs
+        assert outputs["tools_^_search_keywords_~_query"] == "test"
+        assert "tools_^_search_keywords_~_max_keyword_difficulty" in outputs
+        assert outputs["tools_^_search_keywords_~_max_keyword_difficulty"] == 50
+        # Optional parameter should be None when not provided
+        assert "tools_^_search_keywords_~_optional_param" in outputs
+        assert outputs["tools_^_search_keywords_~_optional_param"] is None
+
+    # Test case 4: Valid tool call with ALL parameters (should succeed)
+    mock_tool_call_all_params = MagicMock()
+    mock_tool_call_all_params.function.name = "search_keywords"
+    mock_tool_call_all_params.function.arguments = '{"query": "test", "max_keyword_difficulty": 50, "optional_param": "custom_value"}'
+
+    mock_response_all_params = MagicMock()
+    mock_response_all_params.response = None
+    mock_response_all_params.tool_calls = [mock_tool_call_all_params]
+    mock_response_all_params.prompt_tokens = 50
+    mock_response_all_params.completion_tokens = 25
+    mock_response_all_params.reasoning = None
+    mock_response_all_params.raw_response = {"role": "assistant", "content": None}
+
+    from unittest.mock import AsyncMock
+
+    with patch(
+        "backend.blocks.llm.llm_call",
+        new_callable=AsyncMock,
+        return_value=mock_response_all_params,
+    ), patch.object(
+        SmartDecisionMakerBlock,
+        "_create_function_signature",
+        new_callable=AsyncMock,
+        return_value=mock_tool_functions,
+    ):
+
+        input_data = SmartDecisionMakerBlock.Input(
+            prompt="Search for keywords",
+            model=llm_module.LlmModel.GPT4O,
+            credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
+        )
+
+        # Should succeed with all parameters
+        outputs = {}
+        async for output_name, output_data in block.run(
+            input_data,
+            credentials=llm_module.TEST_CREDENTIALS,
+            graph_id="test-graph-id",
+            node_id="test-node-id",
+            graph_exec_id="test-exec-id",
+            node_exec_id="test-node-exec-id",
+            user_id="test-user-id",
+        ):
+            outputs[output_name] = output_data
+
+        # Verify all tool outputs were generated correctly
+        assert outputs["tools_^_search_keywords_~_query"] == "test"
+        assert outputs["tools_^_search_keywords_~_max_keyword_difficulty"] == 50
+        assert outputs["tools_^_search_keywords_~_optional_param"] == "custom_value"
+
+
+@pytest.mark.asyncio
+async def test_smart_decision_maker_raw_response_conversion():
+    """Test that SmartDecisionMaker correctly handles different raw_response types with retry mechanism."""
+    from unittest.mock import MagicMock, patch
+
+    import backend.blocks.llm as llm_module
+    from backend.blocks.smart_decision_maker import SmartDecisionMakerBlock
+
+    block = SmartDecisionMakerBlock()
+
+    # Mock tool functions
+    mock_tool_functions = [
+        {
+            "type": "function",
+            "function": {
+                "name": "test_tool",
+                "parameters": {
+                    "type": "object",
+                    "properties": {"param": {"type": "string"}},
+                    "required": ["param"],
+                },
+            },
+        }
+    ]
+
+    # Test case 1: Simulate ChatCompletionMessage raw_response that caused the original error
+    class MockChatCompletionMessage:
+        """Simulate OpenAI's ChatCompletionMessage object that lacks .get() method"""
+
+        def __init__(self, role, content, tool_calls=None):
+            self.role = role
+            self.content = content
+            self.tool_calls = tool_calls or []
+
+        # This is what caused the error - no .get() method
+        # def get(self, key, default=None):  # Intentionally missing
+
+    # First response: has invalid parameter name (triggers retry)
+    mock_tool_call_invalid = MagicMock()
+    mock_tool_call_invalid.function.name = "test_tool"
+    mock_tool_call_invalid.function.arguments = (
+        '{"wrong_param": "test_value"}'  # Invalid parameter name
+    )
+
+    mock_response_retry = MagicMock()
+    mock_response_retry.response = None
+    mock_response_retry.tool_calls = [mock_tool_call_invalid]
+    mock_response_retry.prompt_tokens = 50
+    mock_response_retry.completion_tokens = 25
+    mock_response_retry.reasoning = None
+    # This would cause the original error without our fix
+    mock_response_retry.raw_response = MockChatCompletionMessage(
+        role="assistant", content=None, tool_calls=[mock_tool_call_invalid]
+    )
+
+    # Second response: successful (correct parameter name)
+    mock_tool_call_valid = MagicMock()
+    mock_tool_call_valid.function.name = "test_tool"
+    mock_tool_call_valid.function.arguments = (
+        '{"param": "test_value"}'  # Correct parameter name
+    )
+
+    mock_response_success = MagicMock()
+    mock_response_success.response = None
+    mock_response_success.tool_calls = [mock_tool_call_valid]
+    mock_response_success.prompt_tokens = 50
+    mock_response_success.completion_tokens = 25
+    mock_response_success.reasoning = None
+    mock_response_success.raw_response = MockChatCompletionMessage(
+        role="assistant", content=None, tool_calls=[mock_tool_call_valid]
+    )
+
+    # Mock llm_call to return different responses on different calls
+    from unittest.mock import AsyncMock
+
+    with patch(
+        "backend.blocks.llm.llm_call", new_callable=AsyncMock
+    ) as mock_llm_call, patch.object(
+        SmartDecisionMakerBlock,
+        "_create_function_signature",
+        new_callable=AsyncMock,
+        return_value=mock_tool_functions,
+    ):
+        # First call returns response that will trigger retry due to validation error
+        # Second call returns successful response
+        mock_llm_call.side_effect = [mock_response_retry, mock_response_success]
+
+        input_data = SmartDecisionMakerBlock.Input(
+            prompt="Test prompt",
+            model=llm_module.LlmModel.GPT4O,
+            credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
+            retry=2,
+        )
+
+        # Should succeed after retry, demonstrating our helper function works
+        outputs = {}
+        async for output_name, output_data in block.run(
+            input_data,
+            credentials=llm_module.TEST_CREDENTIALS,
+            graph_id="test-graph-id",
+            node_id="test-node-id",
+            graph_exec_id="test-exec-id",
+            node_exec_id="test-node-exec-id",
+            user_id="test-user-id",
+        ):
+            outputs[output_name] = output_data
+
+        # Verify the tool output was generated successfully
+        assert "tools_^_test_tool_~_param" in outputs
+        assert outputs["tools_^_test_tool_~_param"] == "test_value"
+
+        # Verify conversation history was properly maintained
+        assert "conversations" in outputs
+        conversations = outputs["conversations"]
+        assert len(conversations) > 0
+
+        # The conversations should contain properly converted raw_response objects as dicts
+        # This would have failed with the original bug due to ChatCompletionMessage.get() error
+        for msg in conversations:
+            assert isinstance(msg, dict), f"Expected dict, got {type(msg)}"
+            if msg.get("role") == "assistant":
+                # Should have been converted from ChatCompletionMessage to dict
+                assert "role" in msg
+
+        # Verify LLM was called twice (initial + 1 retry)
+        assert mock_llm_call.call_count == 2
+
+    # Test case 2: Test with different raw_response types (Ollama string, dict)
+    # Test Ollama string response
+    mock_response_ollama = MagicMock()
+    mock_response_ollama.response = "I'll help you with that."
+    mock_response_ollama.tool_calls = None
+    mock_response_ollama.prompt_tokens = 30
+    mock_response_ollama.completion_tokens = 15
+    mock_response_ollama.reasoning = None
+    mock_response_ollama.raw_response = (
+        "I'll help you with that."  # Ollama returns string
+    )
+
+    from unittest.mock import AsyncMock
+
+    with patch(
+        "backend.blocks.llm.llm_call",
+        new_callable=AsyncMock,
+        return_value=mock_response_ollama,
+    ), patch.object(
+        SmartDecisionMakerBlock,
+        "_create_function_signature",
+        new_callable=AsyncMock,
+        return_value=[],  # No tools for this test
+    ):
+        input_data = SmartDecisionMakerBlock.Input(
+            prompt="Simple prompt",
+            model=llm_module.LlmModel.GPT4O,
+            credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
+        )
+
+        outputs = {}
+        async for output_name, output_data in block.run(
+            input_data,
+            credentials=llm_module.TEST_CREDENTIALS,
+            graph_id="test-graph-id",
+            node_id="test-node-id",
+            graph_exec_id="test-exec-id",
+            node_exec_id="test-node-exec-id",
+            user_id="test-user-id",
+        ):
+            outputs[output_name] = output_data
+
+        # Should finish since no tool calls
+        assert "finished" in outputs
+        assert outputs["finished"] == "I'll help you with that."
+
+    # Test case 3: Test with dict raw_response (some providers/tests)
+    mock_response_dict = MagicMock()
+    mock_response_dict.response = "Test response"
+    mock_response_dict.tool_calls = None
+    mock_response_dict.prompt_tokens = 25
+    mock_response_dict.completion_tokens = 10
+    mock_response_dict.reasoning = None
+    mock_response_dict.raw_response = {
+        "role": "assistant",
+        "content": "Test response",
+    }  # Dict format
+
+    from unittest.mock import AsyncMock
+
+    with patch(
+        "backend.blocks.llm.llm_call",
+        new_callable=AsyncMock,
+        return_value=mock_response_dict,
+    ), patch.object(
+        SmartDecisionMakerBlock,
+        "_create_function_signature",
+        new_callable=AsyncMock,
+        return_value=[],
+    ):
+        input_data = SmartDecisionMakerBlock.Input(
+            prompt="Another test",
+            model=llm_module.LlmModel.GPT4O,
+            credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
+        )
+
+        outputs = {}
+        async for output_name, output_data in block.run(
+            input_data,
+            credentials=llm_module.TEST_CREDENTIALS,
+            graph_id="test-graph-id",
+            node_id="test-node-id",
+            graph_exec_id="test-exec-id",
+            node_exec_id="test-node-exec-id",
+            user_id="test-user-id",
+        ):
+            outputs[output_name] = output_data
+
+        assert "finished" in outputs
+        assert outputs["finished"] == "Test response"

autogpt_platform/backend/backend/blocks/test/test_smart_decision_maker_dict.py

@@ -48,16 +48,24 @@ async def test_smart_decision_maker_handles_dynamic_dict_fields():
     assert "parameters" in signature["function"]
     assert "properties" in signature["function"]["parameters"]
 
-    # Check that dynamic fields are handled
+    # Check that dynamic fields are handled with original names
     properties = signature["function"]["parameters"]["properties"]
     assert len(properties) == 3  # Should have all three fields
 
-    # Each dynamic field should have proper schema
-    for prop_value in properties.values():
+    # Check that field names are cleaned (for Anthropic API compatibility)
+    assert "values___name" in properties
+    assert "values___age" in properties
+    assert "values___city" in properties
+
+    # Each dynamic field should have proper schema with descriptive text
+    for field_name, prop_value in properties.items():
         assert "type" in prop_value
         assert prop_value["type"] == "string"  # Dynamic fields get string type
         assert "description" in prop_value
-        assert "Dynamic value for" in prop_value["description"]
+        # Check that descriptions properly explain the dynamic field
+        if field_name == "values___name":
+            assert "Dictionary field 'name'" in prop_value["description"]
+            assert "values['name']" in prop_value["description"]
 
 
 @pytest.mark.asyncio
@@ -96,10 +104,18 @@ async def test_smart_decision_maker_handles_dynamic_list_fields():
     properties = signature["function"]["parameters"]["properties"]
     assert len(properties) == 2  # Should have both list items
 
-    # Each dynamic field should have proper schema
-    for prop_value in properties.values():
+    # Check that field names are cleaned (for Anthropic API compatibility)
+    assert "entries___0" in properties
+    assert "entries___1" in properties
+
+    # Each dynamic field should have proper schema with descriptive text
+    for field_name, prop_value in properties.items():
         assert prop_value["type"] == "string"
-        assert "Dynamic value for" in prop_value["description"]
+        assert "description" in prop_value
+        # Check that descriptions properly explain the list field
+        if field_name == "entries___0":
+            assert "List item 0" in prop_value["description"]
+            assert "entries[0]" in prop_value["description"]
 
 
 @pytest.mark.asyncio

autogpt_platform/backend/backend/blocks/test/test_smart_decision_maker_dynamic_fields.py

@@ -0,0 +1,553 @@
+"""Comprehensive tests for SmartDecisionMakerBlock dynamic field handling."""
+
+import json
+from unittest.mock import AsyncMock, Mock, patch
+
+import pytest
+
+from backend.blocks.data_manipulation import AddToListBlock, CreateDictionaryBlock
+from backend.blocks.smart_decision_maker import SmartDecisionMakerBlock
+from backend.blocks.text import MatchTextPatternBlock
+from backend.data.dynamic_fields import get_dynamic_field_description
+
+
+@pytest.mark.asyncio
+async def test_dynamic_field_description_generation():
+    """Test that dynamic field descriptions are generated correctly."""
+    # Test dictionary field description
+    desc = get_dynamic_field_description("values_#_name")
+    assert "Dictionary field 'name' for base field 'values'" in desc
+    assert "values['name']" in desc
+
+    # Test list field description
+    desc = get_dynamic_field_description("items_$_0")
+    assert "List item 0 for base field 'items'" in desc
+    assert "items[0]" in desc
+
+    # Test object field description
+    desc = get_dynamic_field_description("user_@_email")
+    assert "Object attribute 'email' for base field 'user'" in desc
+    assert "user.email" in desc
+
+    # Test regular field fallback
+    desc = get_dynamic_field_description("regular_field")
+    assert desc == "Value for regular_field"
+
+
+@pytest.mark.asyncio
+async def test_create_block_function_signature_with_dict_fields():
+    """Test that function signatures are created correctly for dictionary dynamic fields."""
+    block = SmartDecisionMakerBlock()
+
+    # Create a mock node for CreateDictionaryBlock
+    mock_node = Mock()
+    mock_node.block = CreateDictionaryBlock()
+    mock_node.block_id = CreateDictionaryBlock().id
+    mock_node.input_default = {}
+
+    # Create mock links with dynamic dictionary fields (source sanitized, sink original)
+    mock_links = [
+        Mock(
+            source_name="tools_^_create_dict_~_values___name",  # Sanitized source
+            sink_name="values_#_name",  # Original sink
+            sink_id="dict_node_id",
+            source_id="smart_decision_node_id",
+        ),
+        Mock(
+            source_name="tools_^_create_dict_~_values___age",  # Sanitized source
+            sink_name="values_#_age",  # Original sink
+            sink_id="dict_node_id",
+            source_id="smart_decision_node_id",
+        ),
+        Mock(
+            source_name="tools_^_create_dict_~_values___email",  # Sanitized source
+            sink_name="values_#_email",  # Original sink
+            sink_id="dict_node_id",
+            source_id="smart_decision_node_id",
+        ),
+    ]
+
+    # Generate function signature
+    signature = await block._create_block_function_signature(mock_node, mock_links)  # type: ignore
+
+    # Verify the signature structure
+    assert signature["type"] == "function"
+    assert "function" in signature
+    assert "parameters" in signature["function"]
+    assert "properties" in signature["function"]["parameters"]
+
+    # Check that dynamic fields are handled with original names
+    properties = signature["function"]["parameters"]["properties"]
+    assert len(properties) == 3
+
+    # Check cleaned field names (for Anthropic API compatibility)
+    assert "values___name" in properties
+    assert "values___age" in properties
+    assert "values___email" in properties
+
+    # Check descriptions mention they are dictionary fields
+    assert "Dictionary field" in properties["values___name"]["description"]
+    assert "values['name']" in properties["values___name"]["description"]
+
+    assert "Dictionary field" in properties["values___age"]["description"]
+    assert "values['age']" in properties["values___age"]["description"]
+
+    assert "Dictionary field" in properties["values___email"]["description"]
+    assert "values['email']" in properties["values___email"]["description"]
+
+
+@pytest.mark.asyncio
+async def test_create_block_function_signature_with_list_fields():
+    """Test that function signatures are created correctly for list dynamic fields."""
+    block = SmartDecisionMakerBlock()
+
+    # Create a mock node for AddToListBlock
+    mock_node = Mock()
+    mock_node.block = AddToListBlock()
+    mock_node.block_id = AddToListBlock().id
+    mock_node.input_default = {}
+
+    # Create mock links with dynamic list fields
+    mock_links = [
+        Mock(
+            source_name="tools_^_add_list_~_0",
+            sink_name="entries_$_0",  # Dynamic list field
+            sink_id="list_node_id",
+            source_id="smart_decision_node_id",
+        ),
+        Mock(
+            source_name="tools_^_add_list_~_1",
+            sink_name="entries_$_1",  # Dynamic list field
+            sink_id="list_node_id",
+            source_id="smart_decision_node_id",
+        ),
+        Mock(
+            source_name="tools_^_add_list_~_2",
+            sink_name="entries_$_2",  # Dynamic list field
+            sink_id="list_node_id",
+            source_id="smart_decision_node_id",
+        ),
+    ]
+
+    # Generate function signature
+    signature = await block._create_block_function_signature(mock_node, mock_links)  # type: ignore
+
+    # Verify the signature structure
+    assert signature["type"] == "function"
+    properties = signature["function"]["parameters"]["properties"]
+
+    # Check cleaned field names (for Anthropic API compatibility)
+    assert "entries___0" in properties
+    assert "entries___1" in properties
+    assert "entries___2" in properties
+
+    # Check descriptions mention they are list items
+    assert "List item 0" in properties["entries___0"]["description"]
+    assert "entries[0]" in properties["entries___0"]["description"]
+
+    assert "List item 1" in properties["entries___1"]["description"]
+    assert "entries[1]" in properties["entries___1"]["description"]
+
+
+@pytest.mark.asyncio
+async def test_create_block_function_signature_with_object_fields():
+    """Test that function signatures are created correctly for object dynamic fields."""
+    block = SmartDecisionMakerBlock()
+
+    # Create a mock node for MatchTextPatternBlock (simulating object fields)
+    mock_node = Mock()
+    mock_node.block = MatchTextPatternBlock()
+    mock_node.block_id = MatchTextPatternBlock().id
+    mock_node.input_default = {}
+
+    # Create mock links with dynamic object fields
+    mock_links = [
+        Mock(
+            source_name="tools_^_extract_~_user_name",
+            sink_name="data_@_user_name",  # Dynamic object field
+            sink_id="extract_node_id",
+            source_id="smart_decision_node_id",
+        ),
+        Mock(
+            source_name="tools_^_extract_~_user_email",
+            sink_name="data_@_user_email",  # Dynamic object field
+            sink_id="extract_node_id",
+            source_id="smart_decision_node_id",
+        ),
+    ]
+
+    # Generate function signature
+    signature = await block._create_block_function_signature(mock_node, mock_links)  # type: ignore
+
+    # Verify the signature structure
+    properties = signature["function"]["parameters"]["properties"]
+
+    # Check cleaned field names (for Anthropic API compatibility)
+    assert "data___user_name" in properties
+    assert "data___user_email" in properties
+
+    # Check descriptions mention they are object attributes
+    assert "Object attribute" in properties["data___user_name"]["description"]
+    assert "data.user_name" in properties["data___user_name"]["description"]
+
+
+@pytest.mark.asyncio
+async def test_create_function_signature():
+    """Test that the mapping between sanitized and original field names is built correctly."""
+    block = SmartDecisionMakerBlock()
+
+    # Mock the database client and connected nodes
+    with patch(
+        "backend.blocks.smart_decision_maker.get_database_manager_async_client"
+    ) as mock_db:
+        mock_client = AsyncMock()
+        mock_db.return_value = mock_client
+
+        # Create mock nodes and links
+        mock_dict_node = Mock()
+        mock_dict_node.block = CreateDictionaryBlock()
+        mock_dict_node.block_id = CreateDictionaryBlock().id
+        mock_dict_node.input_default = {}
+
+        mock_list_node = Mock()
+        mock_list_node.block = AddToListBlock()
+        mock_list_node.block_id = AddToListBlock().id
+        mock_list_node.input_default = {}
+
+        # Mock links with dynamic fields
+        dict_link1 = Mock(
+            source_name="tools_^_create_dictionary_~_name",
+            sink_name="values_#_name",
+            sink_id="dict_node_id",
+            source_id="test_node_id",
+        )
+        dict_link2 = Mock(
+            source_name="tools_^_create_dictionary_~_age",
+            sink_name="values_#_age",
+            sink_id="dict_node_id",
+            source_id="test_node_id",
+        )
+        list_link = Mock(
+            source_name="tools_^_add_to_list_~_0",
+            sink_name="entries_$_0",
+            sink_id="list_node_id",
+            source_id="test_node_id",
+        )
+
+        mock_client.get_connected_output_nodes.return_value = [
+            (dict_link1, mock_dict_node),
+            (dict_link2, mock_dict_node),
+            (list_link, mock_list_node),
+        ]
+
+        # Call the method that builds signatures
+        tool_functions = await block._create_function_signature("test_node_id")
+
+        # Verify we got 2 tool functions (one for dict, one for list)
+        assert len(tool_functions) == 2
+
+        # Verify the tool functions contain the dynamic field names
+        dict_tool = next(
+            (
+                tool
+                for tool in tool_functions
+                if tool["function"]["name"] == "createdictionaryblock"
+            ),
+            None,
+        )
+        assert dict_tool is not None
+        dict_properties = dict_tool["function"]["parameters"]["properties"]
+        assert "values___name" in dict_properties
+        assert "values___age" in dict_properties
+
+        list_tool = next(
+            (
+                tool
+                for tool in tool_functions
+                if tool["function"]["name"] == "addtolistblock"
+            ),
+            None,
+        )
+        assert list_tool is not None
+        list_properties = list_tool["function"]["parameters"]["properties"]
+        assert "entries___0" in list_properties
+
+
+@pytest.mark.asyncio
+async def test_output_yielding_with_dynamic_fields():
+    """Test that outputs are yielded correctly with dynamic field names mapped back."""
+    block = SmartDecisionMakerBlock()
+
+    # No more sanitized mapping needed since we removed sanitization
+
+    # Mock LLM response with tool calls
+    mock_response = Mock()
+    mock_response.tool_calls = [
+        Mock(
+            function=Mock(
+                arguments=json.dumps(
+                    {
+                        "values___name": "Alice",
+                        "values___age": 30,
+                        "values___email": "alice@example.com",
+                    }
+                ),
+            )
+        )
+    ]
+    # Ensure function name is a real string, not a Mock name
+    mock_response.tool_calls[0].function.name = "createdictionaryblock"
+    mock_response.reasoning = "Creating a dictionary with user information"
+    mock_response.raw_response = {"role": "assistant", "content": "test"}
+    mock_response.prompt_tokens = 100
+    mock_response.completion_tokens = 50
+
+    # Mock the LLM call
+    with patch(
+        "backend.blocks.smart_decision_maker.llm.llm_call", new_callable=AsyncMock
+    ) as mock_llm:
+        mock_llm.return_value = mock_response
+
+        # Mock the function signature creation
+        with patch.object(
+            block, "_create_function_signature", new_callable=AsyncMock
+        ) as mock_sig:
+            mock_sig.return_value = [
+                {
+                    "type": "function",
+                    "function": {
+                        "name": "createdictionaryblock",
+                        "parameters": {
+                            "type": "object",
+                            "properties": {
+                                "values___name": {"type": "string"},
+                                "values___age": {"type": "number"},
+                                "values___email": {"type": "string"},
+                            },
+                        },
+                    },
+                }
+            ]
+
+            # Create input data
+            from backend.blocks import llm
+
+            input_data = block.input_schema(
+                prompt="Create a user dictionary",
+                credentials=llm.TEST_CREDENTIALS_INPUT,
+                model=llm.LlmModel.GPT4O,
+            )
+
+            # Run the block
+            outputs = {}
+            async for output_name, output_value in block.run(
+                input_data,
+                credentials=llm.TEST_CREDENTIALS,
+                graph_id="test_graph",
+                node_id="test_node",
+                graph_exec_id="test_exec",
+                node_exec_id="test_node_exec",
+                user_id="test_user",
+            ):
+                outputs[output_name] = output_value
+
+            # Verify the outputs use sanitized field names (matching frontend normalizeToolName)
+            assert "tools_^_createdictionaryblock_~_values___name" in outputs
+            assert outputs["tools_^_createdictionaryblock_~_values___name"] == "Alice"
+
+            assert "tools_^_createdictionaryblock_~_values___age" in outputs
+            assert outputs["tools_^_createdictionaryblock_~_values___age"] == 30
+
+            assert "tools_^_createdictionaryblock_~_values___email" in outputs
+            assert (
+                outputs["tools_^_createdictionaryblock_~_values___email"]
+                == "alice@example.com"
+            )
+
+
+@pytest.mark.asyncio
+async def test_mixed_regular_and_dynamic_fields():
+    """Test handling of blocks with both regular and dynamic fields."""
+    block = SmartDecisionMakerBlock()
+
+    # Create a mock node
+    mock_node = Mock()
+    mock_node.block = Mock()
+    mock_node.block.name = "TestBlock"
+    mock_node.block.description = "A test block"
+    mock_node.block.input_schema = Mock()
+
+    # Mock the get_field_schema to return a proper schema for regular fields
+    def get_field_schema(field_name):
+        if field_name == "regular_field":
+            return {"type": "string", "description": "A regular field"}
+        elif field_name == "values":
+            return {"type": "object", "description": "A dictionary field"}
+        else:
+            raise KeyError(f"Field {field_name} not found")
+
+    mock_node.block.input_schema.get_field_schema = get_field_schema
+    mock_node.block.input_schema.jsonschema = Mock(
+        return_value={"properties": {}, "required": []}
+    )
+
+    # Create links with both regular and dynamic fields
+    mock_links = [
+        Mock(
+            source_name="tools_^_test_~_regular",
+            sink_name="regular_field",  # Regular field
+            sink_id="test_node_id",
+            source_id="smart_decision_node_id",
+        ),
+        Mock(
+            source_name="tools_^_test_~_dict_key",
+            sink_name="values_#_key1",  # Dynamic dict field
+            sink_id="test_node_id",
+            source_id="smart_decision_node_id",
+        ),
+        Mock(
+            source_name="tools_^_test_~_dict_key2",
+            sink_name="values_#_key2",  # Dynamic dict field
+            sink_id="test_node_id",
+            source_id="smart_decision_node_id",
+        ),
+    ]
+
+    # Generate function signature
+    signature = await block._create_block_function_signature(mock_node, mock_links)  # type: ignore
+
+    # Check properties
+    properties = signature["function"]["parameters"]["properties"]
+    assert len(properties) == 3
+
+    # Regular field should have its original schema
+    assert "regular_field" in properties
+    assert properties["regular_field"]["description"] == "A regular field"
+
+    # Dynamic fields should have generated descriptions
+    assert "values___key1" in properties
+    assert "Dictionary field" in properties["values___key1"]["description"]
+
+    assert "values___key2" in properties
+    assert "Dictionary field" in properties["values___key2"]["description"]
+
+
+@pytest.mark.asyncio
+async def test_validation_errors_dont_pollute_conversation():
+    """Test that validation errors are only used during retries and don't pollute the conversation."""
+    block = SmartDecisionMakerBlock()
+
+    # Track conversation history changes
+    conversation_snapshots = []
+
+    # Mock response with invalid tool call (missing required parameter)
+    invalid_response = Mock()
+    invalid_response.tool_calls = [
+        Mock(
+            function=Mock(
+                arguments=json.dumps({"wrong_param": "value"}),  # Wrong parameter name
+            )
+        )
+    ]
+    # Ensure function name is a real string, not a Mock name
+    invalid_response.tool_calls[0].function.name = "test_tool"
+    invalid_response.reasoning = None
+    invalid_response.raw_response = {"role": "assistant", "content": "invalid"}
+    invalid_response.prompt_tokens = 100
+    invalid_response.completion_tokens = 50
+
+    # Mock valid response after retry
+    valid_response = Mock()
+    valid_response.tool_calls = [
+        Mock(function=Mock(arguments=json.dumps({"correct_param": "value"})))
+    ]
+    # Ensure function name is a real string, not a Mock name
+    valid_response.tool_calls[0].function.name = "test_tool"
+    valid_response.reasoning = None
+    valid_response.raw_response = {"role": "assistant", "content": "valid"}
+    valid_response.prompt_tokens = 100
+    valid_response.completion_tokens = 50
+
+    call_count = 0
+
+    async def mock_llm_call(**kwargs):
+        nonlocal call_count
+        # Capture conversation state
+        conversation_snapshots.append(kwargs.get("prompt", []).copy())
+        call_count += 1
+        if call_count == 1:
+            return invalid_response
+        else:
+            return valid_response
+
+    # Mock the LLM call
+    with patch(
+        "backend.blocks.smart_decision_maker.llm.llm_call", new_callable=AsyncMock
+    ) as mock_llm:
+        mock_llm.side_effect = mock_llm_call
+
+        # Mock the function signature creation
+        with patch.object(
+            block, "_create_function_signature", new_callable=AsyncMock
+        ) as mock_sig:
+            mock_sig.return_value = [
+                {
+                    "type": "function",
+                    "function": {
+                        "name": "test_tool",
+                        "parameters": {
+                            "type": "object",
+                            "properties": {
+                                "correct_param": {
+                                    "type": "string",
+                                    "description": "The correct parameter",
+                                }
+                            },
+                            "required": ["correct_param"],
+                        },
+                    },
+                }
+            ]
+
+            # Create input data
+            from backend.blocks import llm
+
+            input_data = block.input_schema(
+                prompt="Test prompt",
+                credentials=llm.TEST_CREDENTIALS_INPUT,
+                model=llm.LlmModel.GPT4O,
+                retry=3,  # Allow retries
+            )
+
+            # Run the block
+            outputs = {}
+            async for output_name, output_value in block.run(
+                input_data,
+                credentials=llm.TEST_CREDENTIALS,
+                graph_id="test_graph",
+                node_id="test_node",
+                graph_exec_id="test_exec",
+                node_exec_id="test_node_exec",
+                user_id="test_user",
+            ):
+                outputs[output_name] = output_value
+
+            # Verify we had 2 LLM calls (initial + retry)
+            assert call_count == 2
+
+            # Check the final conversation output
+            final_conversation = outputs.get("conversations", [])
+
+            # The final conversation should NOT contain the validation error message
+            error_messages = [
+                msg
+                for msg in final_conversation
+                if msg.get("role") == "user"
+                and "parameter errors" in msg.get("content", "")
+            ]
+            assert (
+                len(error_messages) == 0
+            ), "Validation error leaked into final conversation"
+
+            # The final conversation should only have the successful response
+            assert final_conversation[-1]["content"] == "valid"

autogpt_platform/backend/backend/blocks/test/test_table_input.py

@@ -0,0 +1,131 @@
+import pytest
+
+from backend.blocks.io import AgentTableInputBlock
+from backend.util.test import execute_block_test
+
+
+@pytest.mark.asyncio
+async def test_table_input_block():
+    """Test the AgentTableInputBlock with basic input/output."""
+    block = AgentTableInputBlock()
+    await execute_block_test(block)
+
+
+@pytest.mark.asyncio
+async def test_table_input_with_data():
+    """Test AgentTableInputBlock with actual table data."""
+    block = AgentTableInputBlock()
+
+    input_data = block.Input(
+        name="test_table",
+        column_headers=["Name", "Age", "City"],
+        value=[
+            {"Name": "John", "Age": "30", "City": "New York"},
+            {"Name": "Jane", "Age": "25", "City": "London"},
+            {"Name": "Bob", "Age": "35", "City": "Paris"},
+        ],
+    )
+
+    output_data = []
+    async for output_name, output_value in block.run(input_data):
+        output_data.append((output_name, output_value))
+
+    assert len(output_data) == 1
+    assert output_data[0][0] == "result"
+
+    result = output_data[0][1]
+    assert len(result) == 3
+    assert result[0]["Name"] == "John"
+    assert result[1]["Age"] == "25"
+    assert result[2]["City"] == "Paris"
+
+
+@pytest.mark.asyncio
+async def test_table_input_empty_data():
+    """Test AgentTableInputBlock with empty data."""
+    block = AgentTableInputBlock()
+
+    input_data = block.Input(
+        name="empty_table", column_headers=["Col1", "Col2"], value=[]
+    )
+
+    output_data = []
+    async for output_name, output_value in block.run(input_data):
+        output_data.append((output_name, output_value))
+
+    assert len(output_data) == 1
+    assert output_data[0][0] == "result"
+    assert output_data[0][1] == []
+
+
+@pytest.mark.asyncio
+async def test_table_input_with_missing_columns():
+    """Test AgentTableInputBlock passes through data with missing columns as-is."""
+    block = AgentTableInputBlock()
+
+    input_data = block.Input(
+        name="partial_table",
+        column_headers=["Name", "Age", "City"],
+        value=[
+            {"Name": "John", "Age": "30"},  # Missing City
+            {"Name": "Jane", "City": "London"},  # Missing Age
+            {"Age": "35", "City": "Paris"},  # Missing Name
+        ],
+    )
+
+    output_data = []
+    async for output_name, output_value in block.run(input_data):
+        output_data.append((output_name, output_value))
+
+    result = output_data[0][1]
+    assert len(result) == 3
+
+    # Check data is passed through as-is
+    assert result[0] == {"Name": "John", "Age": "30"}
+    assert result[1] == {"Name": "Jane", "City": "London"}
+    assert result[2] == {"Age": "35", "City": "Paris"}
+
+
+@pytest.mark.asyncio
+async def test_table_input_none_value():
+    """Test AgentTableInputBlock with None value returns empty list."""
+    block = AgentTableInputBlock()
+
+    input_data = block.Input(
+        name="none_table", column_headers=["Name", "Age"], value=None
+    )
+
+    output_data = []
+    async for output_name, output_value in block.run(input_data):
+        output_data.append((output_name, output_value))
+
+    assert len(output_data) == 1
+    assert output_data[0][0] == "result"
+    assert output_data[0][1] == []
+
+
+@pytest.mark.asyncio
+async def test_table_input_with_default_headers():
+    """Test AgentTableInputBlock with default column headers."""
+    block = AgentTableInputBlock()
+
+    # Don't specify column_headers, should use defaults
+    input_data = block.Input(
+        name="default_headers_table",
+        value=[
+            {"Column 1": "A", "Column 2": "B", "Column 3": "C"},
+            {"Column 1": "D", "Column 2": "E", "Column 3": "F"},
+        ],
+    )
+
+    output_data = []
+    async for output_name, output_value in block.run(input_data):
+        output_data.append((output_name, output_value))
+
+    assert len(output_data) == 1
+    assert output_data[0][0] == "result"
+
+    result = output_data[0][1]
+    assert len(result) == 2
+    assert result[0]["Column 1"] == "A"
+    assert result[1]["Column 3"] == "F"

autogpt_platform/backend/backend/blocks/text.py

@@ -2,6 +2,8 @@ import re
 from pathlib import Path
 from typing import Any
 
+import regex  # Has built-in timeout support
+
 from backend.data.block import Block, BlockCategory, BlockOutput, BlockSchema
 from backend.data.model import SchemaField
 from backend.util import json, text
@@ -137,6 +139,11 @@ class ExtractTextInformationBlock(Block):
         )
 
     async def run(self, input_data: Input, **kwargs) -> BlockOutput:
+        # Security fix: Add limits to prevent ReDoS and memory exhaustion
+        MAX_TEXT_LENGTH = 1_000_000  # 1MB character limit
+        MAX_MATCHES = 1000  # Maximum number of matches to prevent memory exhaustion
+        MAX_MATCH_LENGTH = 10_000  # Maximum length per match
+
         flags = 0
         if not input_data.case_sensitive:
             flags = flags | re.IGNORECASE
@@ -148,20 +155,85 @@ class ExtractTextInformationBlock(Block):
         else:
             txt = json.dumps(input_data.text)
 
-        matches = [
-            match.group(input_data.group)
-            for match in re.finditer(input_data.pattern, txt, flags)
-            if input_data.group <= len(match.groups())
-        ]
-        if not input_data.find_all:
-            matches = matches[:1]
-        for match in matches:
-            yield "positive", match
-        if not matches:
-            yield "negative", input_data.text
+        # Limit text size to prevent DoS
+        if len(txt) > MAX_TEXT_LENGTH:
+            txt = txt[:MAX_TEXT_LENGTH]
 
-        yield "matched_results", matches
-        yield "matched_count", len(matches)
+        # Validate regex pattern to prevent dangerous patterns
+        dangerous_patterns = [
+            r".*\+.*\+",  # Nested quantifiers
+            r".*\*.*\*",  # Nested quantifiers
+            r"(?=.*\+)",  # Lookahead with quantifier
+            r"(?=.*\*)",  # Lookahead with quantifier
+            r"\(.+\)\+",  # Group with nested quantifier
+            r"\(.+\)\*",  # Group with nested quantifier
+            r"\([^)]+\+\)\+",  # Nested quantifiers like (a+)+
+            r"\([^)]+\*\)\*",  # Nested quantifiers like (a*)*
+        ]
+
+        # Check if pattern is potentially dangerous
+        is_dangerous = any(
+            re.search(dangerous, input_data.pattern) for dangerous in dangerous_patterns
+        )
+
+        # Use regex module with timeout for dangerous patterns
+        # For safe patterns, use standard re module for compatibility
+        try:
+            matches = []
+            match_count = 0
+
+            if is_dangerous:
+                # Use regex module with timeout (5 seconds) for dangerous patterns
+                # The regex module supports timeout parameter in finditer
+                try:
+                    for match in regex.finditer(
+                        input_data.pattern, txt, flags=flags, timeout=5.0
+                    ):
+                        if match_count >= MAX_MATCHES:
+                            break
+                        if input_data.group <= len(match.groups()):
+                            match_text = match.group(input_data.group)
+                            # Limit match length to prevent memory exhaustion
+                            if len(match_text) > MAX_MATCH_LENGTH:
+                                match_text = match_text[:MAX_MATCH_LENGTH]
+                            matches.append(match_text)
+                            match_count += 1
+                except regex.error as e:
+                    # Timeout occurred or regex error
+                    if "timeout" in str(e).lower():
+                        # Timeout - return empty results
+                        pass
+                    else:
+                        # Other regex error
+                        raise
+            else:
+                # Use standard re module for non-dangerous patterns
+                for match in re.finditer(input_data.pattern, txt, flags):
+                    if match_count >= MAX_MATCHES:
+                        break
+                    if input_data.group <= len(match.groups()):
+                        match_text = match.group(input_data.group)
+                        # Limit match length to prevent memory exhaustion
+                        if len(match_text) > MAX_MATCH_LENGTH:
+                            match_text = match_text[:MAX_MATCH_LENGTH]
+                        matches.append(match_text)
+                        match_count += 1
+
+            if not input_data.find_all:
+                matches = matches[:1]
+
+            for match in matches:
+                yield "positive", match
+            if not matches:
+                yield "negative", input_data.text
+
+            yield "matched_results", matches
+            yield "matched_count", len(matches)
+        except Exception:
+            # Return empty results on any regex error
+            yield "negative", input_data.text
+            yield "matched_results", []
+            yield "matched_count", 0
 
 
 class FillTextTemplateBlock(Block):

autogpt_platform/backend/backend/blocks/time_blocks.py

@@ -270,13 +270,17 @@ class GetCurrentDateBlock(Block):
             test_output=[
                 (
                     "date",
-                    lambda t: abs(datetime.now() - datetime.strptime(t, "%Y-%m-%d"))
-                    < timedelta(days=8),  # 7 days difference + 1 day error margin.
+                    lambda t: abs(
+                        datetime.now().date() - datetime.strptime(t, "%Y-%m-%d").date()
+                    )
+                    <= timedelta(days=8),  # 7 days difference + 1 day error margin.
                 ),
                 (
                     "date",
-                    lambda t: abs(datetime.now() - datetime.strptime(t, "%m/%d/%Y"))
-                    < timedelta(days=8),
+                    lambda t: abs(
+                        datetime.now().date() - datetime.strptime(t, "%m/%d/%Y").date()
+                    )
+                    <= timedelta(days=8),
                     # 7 days difference + 1 day error margin.
                 ),
                 (
@@ -382,7 +386,7 @@ class GetCurrentDateAndTimeBlock(Block):
                     lambda t: abs(
                         datetime.now().date() - datetime.strptime(t, "%Y/%m/%d").date()
                     )
-                    < timedelta(days=1),  # Date format only, no time component
+                    <= timedelta(days=1),  # Date format only, no time component
                 ),
                 (
                     "date_time",

autogpt_platform/backend/backend/blocks/xml_parser.py

@@ -26,6 +26,14 @@ class XMLParserBlock(Block):
         )
 
     async def run(self, input_data: Input, **kwargs) -> BlockOutput:
+        # Security fix: Add size limits to prevent XML bomb attacks
+        MAX_XML_SIZE = 10 * 1024 * 1024  # 10MB limit for XML input
+
+        if len(input_data.input_xml) > MAX_XML_SIZE:
+            raise ValueError(
+                f"XML too large: {len(input_data.input_xml)} bytes > {MAX_XML_SIZE} bytes"
+            )
+
         try:
             tokens = tokenize(input_data.input_xml)
             parser = Parser(tokens)

autogpt_platform/backend/backend/blocks/youtube.py

@@ -1,6 +1,7 @@
 from urllib.parse import parse_qs, urlparse
 
 from youtube_transcript_api._api import YouTubeTranscriptApi
+from youtube_transcript_api._errors import NoTranscriptFound
 from youtube_transcript_api._transcripts import FetchedTranscript
 from youtube_transcript_api.formatters import TextFormatter
 
@@ -64,7 +65,29 @@ class TranscribeYoutubeVideoBlock(Block):
 
     @staticmethod
     def get_transcript(video_id: str) -> FetchedTranscript:
-        return YouTubeTranscriptApi().fetch(video_id=video_id)
+        """
+        Get transcript for a video, preferring English but falling back to any available language.
+
+        :param video_id: The YouTube video ID
+        :return: The fetched transcript
+        :raises: Any exception except NoTranscriptFound for requested languages
+        """
+        api = YouTubeTranscriptApi()
+        try:
+            # Try to get English transcript first (default behavior)
+            return api.fetch(video_id=video_id)
+        except NoTranscriptFound:
+            # If English is not available, get the first available transcript
+            transcript_list = api.list(video_id)
+            # Try manually created transcripts first, then generated ones
+            available_transcripts = list(
+                transcript_list._manually_created_transcripts.values()
+            ) + list(transcript_list._generated_transcripts.values())
+            if available_transcripts:
+                # Fetch the first available transcript
+                return available_transcripts[0].fetch()
+            # If no transcripts at all, re-raise the original error
+            raise
 
     @staticmethod
     def format_transcript(transcript: FetchedTranscript) -> str:

autogpt_platform/backend/backend/cli.py

@@ -45,9 +45,6 @@ class MainApp(AppProcess):
 
         app.main(silent=True)
 
-    def cleanup(self):
-        pass
-
 
 @click.group()
 def main():

autogpt_platform/backend/backend/data/api_key.py

@@ -9,6 +9,7 @@ from prisma.models import APIKey as PrismaAPIKey
 from prisma.types import APIKeyWhereUniqueInput
 from pydantic import BaseModel, Field
 
+from backend.data.includes import MAX_USER_API_KEYS_FETCH
 from backend.util.exceptions import NotAuthorizedError, NotFoundError
 
 logger = logging.getLogger(__name__)
@@ -178,9 +179,13 @@ async def revoke_api_key(key_id: str, user_id: str) -> APIKeyInfo:
     return APIKeyInfo.from_db(updated_api_key)
 
 
-async def list_user_api_keys(user_id: str) -> list[APIKeyInfo]:
+async def list_user_api_keys(
+    user_id: str, limit: int = MAX_USER_API_KEYS_FETCH
+) -> list[APIKeyInfo]:
     api_keys = await PrismaAPIKey.prisma().find_many(
-        where={"userId": user_id}, order={"createdAt": "desc"}
+        where={"userId": user_id},
+        order={"createdAt": "desc"},
+        take=limit,
     )
 
     return [APIKeyInfo.from_db(key) for key in api_keys]

autogpt_platform/backend/backend/data/block.py

@@ -20,7 +20,6 @@ from typing import (
 
 import jsonref
 import jsonschema
-from autogpt_libs.utils.cache import cached
 from prisma.models import AgentBlock
 from prisma.types import AgentBlockCreateInput
 from pydantic import BaseModel
@@ -28,6 +27,7 @@ from pydantic import BaseModel
 from backend.data.model import NodeExecutionStats
 from backend.integrations.providers import ProviderName
 from backend.util import json
+from backend.util.cache import cached
 from backend.util.settings import Config
 
 from .model import (
@@ -722,7 +722,7 @@ def get_block(block_id: str) -> Block[BlockSchema, BlockSchema] | None:
     return cls() if cls else None
 
 
-@cached()
+@cached(ttl_seconds=3600)
 def get_webhook_block_ids() -> Sequence[str]:
     return [
         id
@@ -731,7 +731,7 @@ def get_webhook_block_ids() -> Sequence[str]:
     ]
 
 
-@cached()
+@cached(ttl_seconds=3600)
 def get_io_block_ids() -> Sequence[str]:
     return [
         id

autogpt_platform/backend/backend/data/block_cost_config.py

@@ -1,7 +1,11 @@
 from typing import Type
 
 from backend.blocks.ai_music_generator import AIMusicGeneratorBlock
-from backend.blocks.ai_shortform_video_block import AIShortformVideoCreatorBlock
+from backend.blocks.ai_shortform_video_block import (
+    AIAdMakerVideoCreatorBlock,
+    AIScreenshotToVideoAdBlock,
+    AIShortformVideoCreatorBlock,
+)
 from backend.blocks.apollo.organization import SearchOrganizationsBlock
 from backend.blocks.apollo.people import SearchPeopleBlock
 from backend.blocks.apollo.person import GetPersonDetailBlock
@@ -69,9 +73,9 @@ MODEL_COST: dict[LlmModel, int] = {
     LlmModel.CLAUDE_4_1_OPUS: 21,
     LlmModel.CLAUDE_4_OPUS: 21,
     LlmModel.CLAUDE_4_SONNET: 5,
+    LlmModel.CLAUDE_4_5_HAIKU: 4,
+    LlmModel.CLAUDE_4_5_SONNET: 9,
     LlmModel.CLAUDE_3_7_SONNET: 5,
-    LlmModel.CLAUDE_3_5_SONNET: 4,
-    LlmModel.CLAUDE_3_5_HAIKU: 1,  # $0.80 / $4.00
     LlmModel.CLAUDE_3_HAIKU: 1,
     LlmModel.AIML_API_QWEN2_5_72B: 1,
     LlmModel.AIML_API_LLAMA3_1_70B: 1,
@@ -321,7 +325,31 @@ BLOCK_COSTS: dict[Type[Block], list[BlockCost]] = {
     ],
     AIShortformVideoCreatorBlock: [
         BlockCost(
-            cost_amount=50,
+            cost_amount=307,
+            cost_filter={
+                "credentials": {
+                    "id": revid_credentials.id,
+                    "provider": revid_credentials.provider,
+                    "type": revid_credentials.type,
+                }
+            },
+        )
+    ],
+    AIAdMakerVideoCreatorBlock: [
+        BlockCost(
+            cost_amount=714,
+            cost_filter={
+                "credentials": {
+                    "id": revid_credentials.id,
+                    "provider": revid_credentials.provider,
+                    "type": revid_credentials.type,
+                }
+            },
+        )
+    ],
+    AIScreenshotToVideoAdBlock: [
+        BlockCost(
+            cost_amount=612,
             cost_filter={
                 "credentials": {
                     "id": revid_credentials.id,

autogpt_platform/backend/backend/data/credit.py

@@ -5,7 +5,6 @@ from datetime import datetime, timezone
 from typing import TYPE_CHECKING, Any, cast
 
 import stripe
-from prisma import Json
 from prisma.enums import (
     CreditRefundRequestStatus,
     CreditTransactionType,
@@ -13,16 +12,13 @@ from prisma.enums import (
     OnboardingStep,
 )
 from prisma.errors import UniqueViolationError
-from prisma.models import CreditRefundRequest, CreditTransaction, User
-from prisma.types import (
-    CreditRefundRequestCreateInput,
-    CreditTransactionCreateInput,
-    CreditTransactionWhereInput,
-)
+from prisma.models import CreditRefundRequest, CreditTransaction, User, UserBalance
+from prisma.types import CreditRefundRequestCreateInput, CreditTransactionWhereInput
 from pydantic import BaseModel
 
-from backend.data import db
 from backend.data.block_cost_config import BLOCK_COSTS
+from backend.data.db import query_raw_with_schema
+from backend.data.includes import MAX_CREDIT_REFUND_REQUESTS_FETCH
 from backend.data.model import (
     AutoTopUpConfig,
     RefundRequest,
@@ -35,7 +31,8 @@ from backend.data.user import get_user_by_id, get_user_email_by_id
 from backend.notifications.notifications import queue_notification_async
 from backend.server.v2.admin.model import UserHistoryResponse
 from backend.util.exceptions import InsufficientBalanceError
-from backend.util.json import SafeJson
+from backend.util.feature_flag import Flag, is_feature_enabled
+from backend.util.json import SafeJson, dumps
 from backend.util.models import Pagination
 from backend.util.retry import func_retry
 from backend.util.settings import Settings
@@ -48,6 +45,10 @@ stripe.api_key = settings.secrets.stripe_api_key
 logger = logging.getLogger(__name__)
 base_url = settings.config.frontend_base_url or settings.config.platform_base_url
 
+# Constants for test compatibility
+POSTGRES_INT_MAX = 2147483647
+POSTGRES_INT_MIN = -2147483648
+
 
 class UsageTransactionMetadata(BaseModel):
     graph_exec_id: str | None = None
@@ -138,14 +139,20 @@ class UserCreditBase(ABC):
         pass
 
     @abstractmethod
-    async def onboarding_reward(self, user_id: str, credits: int, step: OnboardingStep):
+    async def onboarding_reward(
+        self, user_id: str, credits: int, step: OnboardingStep
+    ) -> bool:
         """
         Reward the user with credits for completing an onboarding step.
         Won't reward if the user has already received credits for the step.
 
         Args:
             user_id (str): The user ID.
+            credits (int): The amount to reward.
             step (OnboardingStep): The onboarding step.
+
+        Returns:
+            bool: True if rewarded, False if already rewarded.
         """
         pass
 
@@ -235,6 +242,12 @@ class UserCreditBase(ABC):
         """
         Returns the current balance of the user & the latest balance snapshot time.
         """
+        # Check UserBalance first for efficiency and consistency
+        user_balance = await UserBalance.prisma().find_unique(where={"userId": user_id})
+        if user_balance:
+            return user_balance.balance, user_balance.updatedAt
+
+        # Fallback to transaction history computation if UserBalance doesn't exist
         top_time = self.time_now()
         snapshot = await CreditTransaction.prisma().find_first(
             where={
@@ -249,72 +262,86 @@ class UserCreditBase(ABC):
         snapshot_balance = snapshot.runningBalance or 0 if snapshot else 0
         snapshot_time = snapshot.createdAt if snapshot else datetime_min
 
-        # Get transactions after the snapshot, this should not exist, but just in case.
-        transactions = await CreditTransaction.prisma().group_by(
-            by=["userId"],
-            sum={"amount": True},
-            max={"createdAt": True},
-            where={
-                "userId": user_id,
-                "createdAt": {
-                    "gt": snapshot_time,
-                    "lte": top_time,
-                },
-                "isActive": True,
-            },
-        )
-        transaction_balance = (
-            int(transactions[0].get("_sum", {}).get("amount", 0) + snapshot_balance)
-            if transactions
-            else snapshot_balance
-        )
-        transaction_time = (
-            datetime.fromisoformat(
-                str(transactions[0].get("_max", {}).get("createdAt", datetime_min))
-            )
-            if transactions
-            else snapshot_time
-        )
-        return transaction_balance, transaction_time
+        return snapshot_balance, snapshot_time
 
     @func_retry
     async def _enable_transaction(
         self,
         transaction_key: str,
         user_id: str,
-        metadata: Json,
+        metadata: SafeJson,
         new_transaction_key: str | None = None,
     ):
-        transaction = await CreditTransaction.prisma().find_first_or_raise(
-            where={"transactionKey": transaction_key, "userId": user_id}
+        # First check if transaction exists and is inactive (safety check)
+        transaction = await CreditTransaction.prisma().find_first(
+            where={
+                "transactionKey": transaction_key,
+                "userId": user_id,
+                "isActive": False,
+            }
         )
-        if transaction.isActive:
-            return
+        if not transaction:
+            # Transaction doesn't exist or is already active, return early
+            return None
 
-        async with db.locked_transaction(f"usr_trx_{user_id}"):
-
-            transaction = await CreditTransaction.prisma().find_first_or_raise(
-                where={"transactionKey": transaction_key, "userId": user_id}
+        # Atomic operation to enable transaction and update user balance using UserBalance
+        result = await query_raw_with_schema(
+            """
+            WITH user_balance_lock AS (
+                SELECT 
+                    $2::text as userId, 
+                    COALESCE(
+                        (SELECT balance FROM {schema_prefix}"UserBalance" WHERE "userId" = $2 FOR UPDATE),
+                        -- Fallback: compute balance from transaction history if UserBalance doesn't exist
+                        (SELECT COALESCE(ct."runningBalance", 0) 
+                         FROM {schema_prefix}"CreditTransaction" ct 
+                         WHERE ct."userId" = $2 
+                           AND ct."isActive" = true 
+                           AND ct."runningBalance" IS NOT NULL 
+                         ORDER BY ct."createdAt" DESC 
+                         LIMIT 1),
+                        0
+                    ) as balance
+            ),
+            transaction_check AS (
+                SELECT * FROM {schema_prefix}"CreditTransaction" 
+                WHERE "transactionKey" = $1 AND "userId" = $2 AND "isActive" = false
+            ),
+            balance_update AS (
+                INSERT INTO {schema_prefix}"UserBalance" ("userId", "balance", "updatedAt")
+                SELECT 
+                    $2::text,
+                    user_balance_lock.balance + transaction_check.amount,
+                    CURRENT_TIMESTAMP
+                FROM user_balance_lock, transaction_check
+                ON CONFLICT ("userId") DO UPDATE SET
+                    "balance" = EXCLUDED."balance",
+                    "updatedAt" = EXCLUDED."updatedAt"
+                RETURNING "balance", "updatedAt"
+            ),
+            transaction_update AS (
+                UPDATE {schema_prefix}"CreditTransaction"
+                SET "transactionKey" = COALESCE($4, $1),
+                    "isActive" = true,
+                    "runningBalance" = balance_update.balance,
+                    "createdAt" = balance_update."updatedAt",
+                    "metadata" = $3::jsonb
+                FROM balance_update, transaction_check
+                WHERE {schema_prefix}"CreditTransaction"."transactionKey" = transaction_check."transactionKey"
+                  AND {schema_prefix}"CreditTransaction"."userId" = transaction_check."userId"
+                RETURNING {schema_prefix}"CreditTransaction"."runningBalance"
             )
-            if transaction.isActive:
-                return
+            SELECT "runningBalance" as balance FROM transaction_update;
+            """,
+            transaction_key,  # $1
+            user_id,  # $2
+            dumps(metadata.data),  # $3 - use pre-serialized JSON string for JSONB
+            new_transaction_key,  # $4
+        )
 
-            user_balance, _ = await self._get_credits(user_id)
-            await CreditTransaction.prisma().update(
-                where={
-                    "creditTransactionIdentifier": {
-                        "transactionKey": transaction_key,
-                        "userId": user_id,
-                    }
-                },
-                data={
-                    "transactionKey": new_transaction_key or transaction_key,
-                    "isActive": True,
-                    "runningBalance": user_balance + transaction.amount,
-                    "createdAt": self.time_now(),
-                    "metadata": metadata,
-                },
-            )
+        if result:
+            # UserBalance is already updated by the CTE
+            return result[0]["balance"]
 
     async def _add_transaction(
         self,
@@ -325,12 +352,54 @@ class UserCreditBase(ABC):
         transaction_key: str | None = None,
         ceiling_balance: int | None = None,
         fail_insufficient_credits: bool = True,
-        metadata: Json = SafeJson({}),
+        metadata: SafeJson = SafeJson({}),
     ) -> tuple[int, str]:
         """
         Add a new transaction for the user.
         This is the only method that should be used to add a new transaction.
 
+        ATOMIC OPERATION DESIGN DECISION:
+        ================================
+        This method uses PostgreSQL row-level locking (FOR UPDATE) for atomic credit operations.
+        After extensive analysis of concurrency patterns and correctness requirements, we determined
+        that the FOR UPDATE approach is necessary despite the latency overhead.
+
+        WHY FOR UPDATE LOCKING IS REQUIRED:
+        ----------------------------------
+        1. **Data Consistency**: Credit operations must be ACID-compliant. The balance check,
+           calculation, and update must be atomic to prevent race conditions where:
+           - Multiple spend operations could exceed available balance
+           - Lost update problems could occur with concurrent top-ups
+           - Refunds could create negative balances incorrectly
+
+        2. **Serializability**: FOR UPDATE ensures operations are serialized at the database level,
+           guaranteeing that each transaction sees a consistent view of the balance before applying changes.
+
+        3. **Correctness Over Performance**: Financial operations require absolute correctness.
+           The ~10-50ms latency increase from row locking is acceptable for the guarantee that
+           no user will ever have an incorrect balance due to race conditions.
+
+        4. **PostgreSQL Optimization**: Modern PostgreSQL versions optimize row locks efficiently.
+           The performance cost is minimal compared to the complexity and risk of lock-free approaches.
+
+        ALTERNATIVES CONSIDERED AND REJECTED:
+        ------------------------------------
+        - **Optimistic Concurrency**: Using version numbers or timestamps would require complex
+          retry logic and could still fail under high contention scenarios.
+        - **Application-Level Locking**: Redis locks or similar would add network overhead and
+          single points of failure while being less reliable than database locks.
+        - **Event Sourcing**: Would require complete architectural changes and eventual consistency
+          models that don't fit our real-time balance requirements.
+
+        PERFORMANCE CHARACTERISTICS:
+        ---------------------------
+        - Single user operations: 10-50ms latency (acceptable for financial operations)
+        - Concurrent operations on same user: Serialized (prevents data corruption)
+        - Concurrent operations on different users: Fully parallel (no blocking)
+
+        This design prioritizes correctness and data integrity over raw performance,
+        which is the appropriate choice for a credit/payment system.
+
         Args:
             user_id (str): The user ID.
             amount (int): The amount of credits to add.
@@ -344,40 +413,142 @@ class UserCreditBase(ABC):
         Returns:
             tuple[int, str]: The new balance & the transaction key.
         """
-        async with db.locked_transaction(f"usr_trx_{user_id}"):
-            # Get latest balance snapshot
-            user_balance, _ = await self._get_credits(user_id)
-
-            if ceiling_balance and amount > 0 and user_balance >= ceiling_balance:
+        # Quick validation for ceiling balance to avoid unnecessary database operations
+        if ceiling_balance and amount > 0:
+            current_balance, _ = await self._get_credits(user_id)
+            if current_balance >= ceiling_balance:
                 raise ValueError(
-                    f"You already have enough balance of ${user_balance/100}, top-up is not required when you already have at least ${ceiling_balance/100}"
+                    f"You already have enough balance of ${current_balance/100}, top-up is not required when you already have at least ${ceiling_balance/100}"
                 )
 
-            if amount < 0 and user_balance + amount < 0:
-                if fail_insufficient_credits:
-                    raise InsufficientBalanceError(
-                        message=f"Insufficient balance of ${user_balance/100}, where this will cost ${abs(amount)/100}",
-                        user_id=user_id,
-                        balance=user_balance,
-                        amount=amount,
+        # Single unified atomic operation for all transaction types using UserBalance
+        try:
+            result = await query_raw_with_schema(
+                """
+                WITH user_balance_lock AS (
+                    SELECT 
+                        $1::text as userId, 
+                        -- CRITICAL: FOR UPDATE lock prevents concurrent modifications to the same user's balance
+                        -- This ensures atomic read-modify-write operations and prevents race conditions
+                        COALESCE(
+                            (SELECT balance FROM {schema_prefix}"UserBalance" WHERE "userId" = $1 FOR UPDATE),
+                            -- Fallback: compute balance from transaction history if UserBalance doesn't exist
+                            (SELECT COALESCE(ct."runningBalance", 0) 
+                             FROM {schema_prefix}"CreditTransaction" ct 
+                             WHERE ct."userId" = $1 
+                               AND ct."isActive" = true 
+                               AND ct."runningBalance" IS NOT NULL 
+                             ORDER BY ct."createdAt" DESC 
+                             LIMIT 1),
+                            0
+                        ) as balance
+                ),
+                balance_update AS (
+                    INSERT INTO {schema_prefix}"UserBalance" ("userId", "balance", "updatedAt")
+                    SELECT 
+                        $1::text,
+                        CASE 
+                            -- For inactive transactions: Don't update balance
+                            WHEN $5::boolean = false THEN user_balance_lock.balance
+                            -- For ceiling balance (amount > 0): Apply ceiling
+                            WHEN $2 > 0 AND $7::int IS NOT NULL AND user_balance_lock.balance > $7::int - $2 THEN $7::int
+                            -- For regular operations: Apply with overflow/underflow protection  
+                            WHEN user_balance_lock.balance + $2 > $6::int THEN $6::int
+                            WHEN user_balance_lock.balance + $2 < $10::int THEN $10::int
+                            ELSE user_balance_lock.balance + $2
+                        END,
+                        CURRENT_TIMESTAMP
+                    FROM user_balance_lock
+                    WHERE (
+                        $5::boolean = false OR  -- Allow inactive transactions
+                        $2 >= 0 OR              -- Allow positive amounts (top-ups, grants)
+                        $8::boolean = false OR  -- Allow when insufficient balance check is disabled
+                        user_balance_lock.balance + $2 >= 0  -- Allow spending only when sufficient balance
                     )
+                    ON CONFLICT ("userId") DO UPDATE SET
+                        "balance" = EXCLUDED."balance",
+                        "updatedAt" = EXCLUDED."updatedAt"
+                    RETURNING "balance", "updatedAt"
+                ),
+                transaction_insert AS (
+                    INSERT INTO {schema_prefix}"CreditTransaction" (
+                        "userId", "amount", "type", "runningBalance", 
+                        "metadata", "isActive", "createdAt", "transactionKey"
+                    )
+                    SELECT 
+                        $1::text,
+                        $2::int,
+                        $3::text::{schema_prefix}"CreditTransactionType",
+                        CASE 
+                            -- For inactive transactions: Set runningBalance to original balance (don't apply the change yet)
+                            WHEN $5::boolean = false THEN user_balance_lock.balance
+                            ELSE COALESCE(balance_update.balance, user_balance_lock.balance)
+                        END,
+                        $4::jsonb,
+                        $5::boolean,
+                        COALESCE(balance_update."updatedAt", CURRENT_TIMESTAMP),
+                        COALESCE($9, gen_random_uuid()::text)
+                    FROM user_balance_lock
+                    LEFT JOIN balance_update ON true
+                    WHERE (
+                        $5::boolean = false OR  -- Allow inactive transactions
+                        $2 >= 0 OR              -- Allow positive amounts (top-ups, grants)
+                        $8::boolean = false OR  -- Allow when insufficient balance check is disabled
+                        user_balance_lock.balance + $2 >= 0  -- Allow spending only when sufficient balance
+                    )
+                    RETURNING "runningBalance", "transactionKey"
+                )
+                SELECT "runningBalance" as balance, "transactionKey" FROM transaction_insert;
+                """,
+                user_id,  # $1
+                amount,  # $2
+                transaction_type.value,  # $3
+                dumps(metadata.data),  # $4 - use pre-serialized JSON string for JSONB
+                is_active,  # $5
+                POSTGRES_INT_MAX,  # $6 - overflow protection
+                ceiling_balance,  # $7 - ceiling balance (nullable)
+                fail_insufficient_credits,  # $8 - check balance for spending
+                transaction_key,  # $9 - transaction key (nullable)
+                POSTGRES_INT_MIN,  # $10 - underflow protection
+            )
+        except Exception as e:
+            # Convert raw SQL unique constraint violations to UniqueViolationError
+            # for consistent exception handling throughout the codebase
+            error_str = str(e).lower()
+            if (
+                "already exists" in error_str
+                or "duplicate key" in error_str
+                or "unique constraint" in error_str
+            ):
+                # Extract table and constraint info for better error messages
+                # Re-raise as a UniqueViolationError but with proper format
+                # Create a minimal data structure that the error constructor expects
+                raise UniqueViolationError({"error": str(e), "user_facing_error": {}})
+            # For any other error, re-raise as-is
+            raise
 
-                amount = min(-user_balance, 0)
+        if result:
+            new_balance, tx_key = result[0]["balance"], result[0]["transactionKey"]
+            # UserBalance is already updated by the CTE
+            return new_balance, tx_key
 
-            # Create the transaction
-            transaction_data: CreditTransactionCreateInput = {
-                "userId": user_id,
-                "amount": amount,
-                "runningBalance": user_balance + amount,
-                "type": transaction_type,
-                "metadata": metadata,
-                "isActive": is_active,
-                "createdAt": self.time_now(),
-            }
-            if transaction_key:
-                transaction_data["transactionKey"] = transaction_key
-            tx = await CreditTransaction.prisma().create(data=transaction_data)
-            return user_balance + amount, tx.transactionKey
+        # If no result, either user doesn't exist or insufficient balance
+        user = await User.prisma().find_unique(where={"id": user_id})
+        if not user:
+            raise ValueError(f"User {user_id} not found")
+
+        # Must be insufficient balance for spending operation
+        if amount < 0 and fail_insufficient_credits:
+            current_balance, _ = await self._get_credits(user_id)
+            raise InsufficientBalanceError(
+                message=f"Insufficient balance of ${current_balance/100}, where this will cost ${abs(amount)/100}",
+                user_id=user_id,
+                balance=current_balance,
+                amount=amount,
+            )
+
+        # Unexpected case
+        raise ValueError(f"Transaction failed for user {user_id}, amount {amount}")
 
 
 class UserCredit(UserCreditBase):
@@ -452,9 +623,10 @@ class UserCredit(UserCreditBase):
                     {"reason": f"Reward for completing {step.value} onboarding step."}
                 ),
             )
+            return True
         except UniqueViolationError:
-            # Already rewarded for this step
-            pass
+            # User already received this reward
+            return False
 
     async def top_up_refund(
         self, user_id: str, transaction_key: str, metadata: dict[str, str]
@@ -643,7 +815,7 @@ class UserCredit(UserCreditBase):
     ):
         # init metadata, without sharing it with the world
         metadata = metadata or {}
-        if not metadata["reason"]:
+        if not metadata.get("reason"):
             match top_up_type:
                 case TopUpType.MANUAL:
                     metadata["reason"] = {"reason": f"Top up credits for {user_id}"}
@@ -905,7 +1077,9 @@ class UserCredit(UserCreditBase):
             ),
         )
 
-    async def get_refund_requests(self, user_id: str) -> list[RefundRequest]:
+    async def get_refund_requests(
+        self, user_id: str, limit: int = MAX_CREDIT_REFUND_REQUESTS_FETCH
+    ) -> list[RefundRequest]:
         return [
             RefundRequest(
                 id=r.id,
@@ -921,6 +1095,7 @@ class UserCredit(UserCreditBase):
             for r in await CreditRefundRequest.prisma().find_many(
                 where={"userId": user_id},
                 order={"createdAt": "desc"},
+                take=limit,
             )
         ]
 
@@ -970,8 +1145,8 @@ class DisabledUserCredit(UserCreditBase):
     async def top_up_credits(self, *args, **kwargs):
         pass
 
-    async def onboarding_reward(self, *args, **kwargs):
-        pass
+    async def onboarding_reward(self, *args, **kwargs) -> bool:
+        return True
 
     async def top_up_intent(self, *args, **kwargs) -> str:
         return ""
@@ -989,14 +1164,31 @@ class DisabledUserCredit(UserCreditBase):
         pass
 
 
-def get_user_credit_model() -> UserCreditBase:
+async def get_user_credit_model(user_id: str) -> UserCreditBase:
+    """
+    Get the credit model for a user, considering LaunchDarkly flags.
+
+    Args:
+        user_id (str): The user ID to check flags for.
+
+    Returns:
+        UserCreditBase: The appropriate credit model for the user
+    """
     if not settings.config.enable_credit:
         return DisabledUserCredit()
 
-    if settings.config.enable_beta_monthly_credit:
-        return BetaUserCredit(settings.config.num_user_credits_refill)
+    # Check LaunchDarkly flag for payment pilot users
+    # Default to False (beta monthly credit behavior) to maintain current behavior
+    is_payment_enabled = await is_feature_enabled(
+        Flag.ENABLE_PLATFORM_PAYMENT, user_id, default=False
+    )
 
-    return UserCredit()
+    if is_payment_enabled:
+        # Payment enabled users get UserCredit (no monthly refills, enable payments)
+        return UserCredit()
+    else:
+        # Default behavior: users get beta monthly credits
+        return BetaUserCredit(settings.config.num_user_credits_refill)
 
 
 def get_block_costs() -> dict[str, list["BlockCost"]]:
@@ -1086,7 +1278,8 @@ async def admin_get_user_history(
             )
             reason = metadata.get("reason", "No reason provided")
 
-        balance, last_update = await get_user_credit_model()._get_credits(tx.userId)
+        user_credit_model = await get_user_credit_model(tx.userId)
+        balance, _ = await user_credit_model._get_credits(tx.userId)
 
         history.append(
             UserTransaction(

autogpt_platform/backend/backend/data/credit_ceiling_test.py

@@ -0,0 +1,172 @@
+"""
+Test ceiling balance functionality to ensure auto top-up limits work correctly.
+
+This test was added to cover a previously untested code path that could lead to
+incorrect balance capping behavior.
+"""
+
+from uuid import uuid4
+
+import pytest
+from prisma.enums import CreditTransactionType
+from prisma.errors import UniqueViolationError
+from prisma.models import CreditTransaction, User, UserBalance
+
+from backend.data.credit import UserCredit
+from backend.util.json import SafeJson
+from backend.util.test import SpinTestServer
+
+
+async def create_test_user(user_id: str) -> None:
+    """Create a test user for ceiling tests."""
+    try:
+        await User.prisma().create(
+            data={
+                "id": user_id,
+                "email": f"test-{user_id}@example.com",
+                "name": f"Test User {user_id[:8]}",
+            }
+        )
+    except UniqueViolationError:
+        # User already exists, continue
+        pass
+
+    await UserBalance.prisma().upsert(
+        where={"userId": user_id},
+        data={"create": {"userId": user_id, "balance": 0}, "update": {"balance": 0}},
+    )
+
+
+async def cleanup_test_user(user_id: str) -> None:
+    """Clean up test user and their transactions."""
+    try:
+        await CreditTransaction.prisma().delete_many(where={"userId": user_id})
+        await User.prisma().delete_many(where={"id": user_id})
+    except Exception as e:
+        # Log cleanup failures but don't fail the test
+        print(f"Warning: Failed to cleanup test user {user_id}: {e}")
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_ceiling_balance_rejects_when_above_threshold(server: SpinTestServer):
+    """Test that ceiling balance correctly rejects top-ups when balance is above threshold."""
+    credit_system = UserCredit()
+    user_id = f"ceiling-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Give user balance of 1000 ($10) using internal method (bypasses Stripe)
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=1000,
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "initial_balance"}),
+        )
+        current_balance = await credit_system.get_credits(user_id)
+        assert current_balance == 1000
+
+        # Try to add 200 more with ceiling of 800 (should reject since 1000 > 800)
+        with pytest.raises(ValueError, match="You already have enough balance"):
+            await credit_system._add_transaction(
+                user_id=user_id,
+                amount=200,
+                transaction_type=CreditTransactionType.TOP_UP,
+                ceiling_balance=800,  # Ceiling lower than current balance
+            )
+
+        # Balance should remain unchanged
+        final_balance = await credit_system.get_credits(user_id)
+        assert final_balance == 1000, f"Balance should remain 1000, got {final_balance}"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_ceiling_balance_clamps_when_would_exceed(server: SpinTestServer):
+    """Test that ceiling balance correctly clamps amounts that would exceed the ceiling."""
+    credit_system = UserCredit()
+    user_id = f"ceiling-clamp-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Give user balance of 500 ($5) using internal method (bypasses Stripe)
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=500,
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "initial_balance"}),
+        )
+
+        # Add 800 more with ceiling of 1000 (should clamp to 1000, not reach 1300)
+        final_balance, _ = await credit_system._add_transaction(
+            user_id=user_id,
+            amount=800,
+            transaction_type=CreditTransactionType.TOP_UP,
+            ceiling_balance=1000,  # Ceiling should clamp 500 + 800 = 1300 to 1000
+        )
+
+        # Balance should be clamped to ceiling
+        assert (
+            final_balance == 1000
+        ), f"Balance should be clamped to 1000, got {final_balance}"
+
+        # Verify with get_credits too
+        stored_balance = await credit_system.get_credits(user_id)
+        assert (
+            stored_balance == 1000
+        ), f"Stored balance should be 1000, got {stored_balance}"
+
+        # Verify transaction shows the clamped amount
+        transactions = await CreditTransaction.prisma().find_many(
+            where={"userId": user_id, "type": CreditTransactionType.TOP_UP},
+            order={"createdAt": "desc"},
+        )
+
+        # Should have 2 transactions: 500 + (500 to reach ceiling of 1000)
+        assert len(transactions) == 2
+
+        # The second transaction should show it only added 500, not 800
+        second_tx = transactions[0]  # Most recent
+        assert second_tx.runningBalance == 1000
+        # The actual amount recorded could be 800 (what was requested) but balance was clamped
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_ceiling_balance_allows_when_under_threshold(server: SpinTestServer):
+    """Test that ceiling balance allows top-ups when balance is under threshold."""
+    credit_system = UserCredit()
+    user_id = f"ceiling-under-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Give user balance of 300 ($3) using internal method (bypasses Stripe)
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=300,
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "initial_balance"}),
+        )
+
+        # Add 200 more with ceiling of 1000 (should succeed: 300 + 200 = 500 < 1000)
+        final_balance, _ = await credit_system._add_transaction(
+            user_id=user_id,
+            amount=200,
+            transaction_type=CreditTransactionType.TOP_UP,
+            ceiling_balance=1000,
+        )
+
+        # Balance should be exactly 500
+        assert final_balance == 500, f"Balance should be 500, got {final_balance}"
+
+        # Verify with get_credits too
+        stored_balance = await credit_system.get_credits(user_id)
+        assert (
+            stored_balance == 500
+        ), f"Stored balance should be 500, got {stored_balance}"
+
+    finally:
+        await cleanup_test_user(user_id)

autogpt_platform/backend/backend/data/credit_concurrency_test.py

@@ -0,0 +1,737 @@
+"""
+Concurrency and atomicity tests for the credit system.
+
+These tests ensure the credit system handles high-concurrency scenarios correctly
+without race conditions, deadlocks, or inconsistent state.
+"""
+
+import asyncio
+import random
+from uuid import uuid4
+
+import prisma.enums
+import pytest
+from prisma.enums import CreditTransactionType
+from prisma.errors import UniqueViolationError
+from prisma.models import CreditTransaction, User, UserBalance
+
+from backend.data.credit import POSTGRES_INT_MAX, UsageTransactionMetadata, UserCredit
+from backend.util.exceptions import InsufficientBalanceError
+from backend.util.json import SafeJson
+from backend.util.test import SpinTestServer
+
+# Test with both UserCredit and BetaUserCredit if needed
+credit_system = UserCredit()
+
+
+async def create_test_user(user_id: str) -> None:
+    """Create a test user with initial balance."""
+    try:
+        await User.prisma().create(
+            data={
+                "id": user_id,
+                "email": f"test-{user_id}@example.com",
+                "name": f"Test User {user_id[:8]}",
+            }
+        )
+    except UniqueViolationError:
+        # User already exists, continue
+        pass
+
+    # Ensure UserBalance record exists
+    await UserBalance.prisma().upsert(
+        where={"userId": user_id},
+        data={"create": {"userId": user_id, "balance": 0}, "update": {"balance": 0}},
+    )
+
+
+async def cleanup_test_user(user_id: str) -> None:
+    """Clean up test user and their transactions."""
+    try:
+        await CreditTransaction.prisma().delete_many(where={"userId": user_id})
+        await UserBalance.prisma().delete_many(where={"userId": user_id})
+        await User.prisma().delete_many(where={"id": user_id})
+    except Exception as e:
+        # Log cleanup failures but don't fail the test
+        print(f"Warning: Failed to cleanup test user {user_id}: {e}")
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_concurrent_spends_same_user(server: SpinTestServer):
+    """Test multiple concurrent spends from the same user don't cause race conditions."""
+    user_id = f"concurrent-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Give user initial balance using internal method (bypasses Stripe)
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=1000,
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "initial_balance"}),
+        )
+
+        # Try to spend 10 x $1 concurrently
+        async def spend_one_dollar(idx: int):
+            try:
+                return await credit_system.spend_credits(
+                    user_id,
+                    100,  # $1
+                    UsageTransactionMetadata(
+                        graph_exec_id=f"concurrent-{idx}",
+                        reason=f"Concurrent spend {idx}",
+                    ),
+                )
+            except InsufficientBalanceError:
+                return None
+
+        # Run 10 concurrent spends
+        results = await asyncio.gather(
+            *[spend_one_dollar(i) for i in range(10)], return_exceptions=True
+        )
+
+        # Count successful spends
+        successful = [
+            r for r in results if r is not None and not isinstance(r, Exception)
+        ]
+        failed = [r for r in results if isinstance(r, InsufficientBalanceError)]
+
+        # All 10 should succeed since we have exactly $10
+        assert len(successful) == 10, f"Expected 10 successful, got {len(successful)}"
+        assert len(failed) == 0, f"Expected 0 failures, got {len(failed)}"
+
+        # Final balance should be exactly 0
+        final_balance = await credit_system.get_credits(user_id)
+        assert final_balance == 0, f"Expected balance 0, got {final_balance}"
+
+        # Verify transaction history is consistent
+        transactions = await CreditTransaction.prisma().find_many(
+            where={"userId": user_id, "type": prisma.enums.CreditTransactionType.USAGE}
+        )
+        assert (
+            len(transactions) == 10
+        ), f"Expected 10 transactions, got {len(transactions)}"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_concurrent_spends_insufficient_balance(server: SpinTestServer):
+    """Test that concurrent spends correctly enforce balance limits."""
+    user_id = f"insufficient-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Give user limited balance using internal method (bypasses Stripe)
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=500,
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "limited_balance"}),
+        )
+
+        # Try to spend 10 x $1 concurrently (but only have $5)
+        async def spend_one_dollar(idx: int):
+            try:
+                return await credit_system.spend_credits(
+                    user_id,
+                    100,  # $1
+                    UsageTransactionMetadata(
+                        graph_exec_id=f"insufficient-{idx}",
+                        reason=f"Insufficient spend {idx}",
+                    ),
+                )
+            except InsufficientBalanceError:
+                return "FAILED"
+
+        # Run 10 concurrent spends
+        results = await asyncio.gather(
+            *[spend_one_dollar(i) for i in range(10)], return_exceptions=True
+        )
+
+        # Count successful vs failed
+        successful = [
+            r
+            for r in results
+            if r not in ["FAILED", None] and not isinstance(r, Exception)
+        ]
+        failed = [r for r in results if r == "FAILED"]
+
+        # Exactly 5 should succeed, 5 should fail
+        assert len(successful) == 5, f"Expected 5 successful, got {len(successful)}"
+        assert len(failed) == 5, f"Expected 5 failures, got {len(failed)}"
+
+        # Final balance should be exactly 0
+        final_balance = await credit_system.get_credits(user_id)
+        assert final_balance == 0, f"Expected balance 0, got {final_balance}"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_concurrent_mixed_operations(server: SpinTestServer):
+    """Test concurrent mix of spends, top-ups, and balance checks."""
+    user_id = f"mixed-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Initial balance using internal method (bypasses Stripe)
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=1000,
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "initial_balance"}),
+        )
+
+        # Mix of operations
+        async def mixed_operations():
+            operations = []
+
+            # 5 spends of $1 each
+            for i in range(5):
+                operations.append(
+                    credit_system.spend_credits(
+                        user_id,
+                        100,
+                        UsageTransactionMetadata(reason=f"Mixed spend {i}"),
+                    )
+                )
+
+            # 3 top-ups of $2 each using internal method
+            for i in range(3):
+                operations.append(
+                    credit_system._add_transaction(
+                        user_id=user_id,
+                        amount=200,
+                        transaction_type=CreditTransactionType.TOP_UP,
+                        metadata=SafeJson({"test": f"concurrent_topup_{i}"}),
+                    )
+                )
+
+            # 10 balance checks
+            for i in range(10):
+                operations.append(credit_system.get_credits(user_id))
+
+            return await asyncio.gather(*operations, return_exceptions=True)
+
+        results = await mixed_operations()
+
+        # Check no exceptions occurred
+        exceptions = [
+            r
+            for r in results
+            if isinstance(r, Exception) and not isinstance(r, InsufficientBalanceError)
+        ]
+        assert len(exceptions) == 0, f"Unexpected exceptions: {exceptions}"
+
+        # Final balance should be: 1000 - 500 + 600 = 1100
+        final_balance = await credit_system.get_credits(user_id)
+        assert final_balance == 1100, f"Expected balance 1100, got {final_balance}"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_race_condition_exact_balance(server: SpinTestServer):
+    """Test spending exact balance amount concurrently doesn't go negative."""
+    user_id = f"exact-balance-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Give exact amount using internal method (bypasses Stripe)
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=100,
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "exact_amount"}),
+        )
+
+        # Try to spend $1 twice concurrently
+        async def spend_exact():
+            try:
+                return await credit_system.spend_credits(
+                    user_id, 100, UsageTransactionMetadata(reason="Exact spend")
+                )
+            except InsufficientBalanceError:
+                return "FAILED"
+
+        # Both try to spend the full balance
+        result1, result2 = await asyncio.gather(spend_exact(), spend_exact())
+
+        # Exactly one should succeed
+        results = [result1, result2]
+        successful = [
+            r for r in results if r != "FAILED" and not isinstance(r, Exception)
+        ]
+        failed = [r for r in results if r == "FAILED"]
+
+        assert len(successful) == 1, f"Expected 1 success, got {len(successful)}"
+        assert len(failed) == 1, f"Expected 1 failure, got {len(failed)}"
+
+        # Balance should be exactly 0, never negative
+        final_balance = await credit_system.get_credits(user_id)
+        assert final_balance == 0, f"Expected balance 0, got {final_balance}"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_onboarding_reward_idempotency(server: SpinTestServer):
+    """Test that onboarding rewards are idempotent (can't be claimed twice)."""
+    user_id = f"onboarding-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Use WELCOME step which is defined in the OnboardingStep enum
+        # Try to claim same reward multiple times concurrently
+        async def claim_reward():
+            try:
+                result = await credit_system.onboarding_reward(
+                    user_id, 500, prisma.enums.OnboardingStep.WELCOME
+                )
+                return "SUCCESS" if result else "DUPLICATE"
+            except Exception as e:
+                print(f"Claim reward failed: {e}")
+                return "FAILED"
+
+        # Try 5 concurrent claims of the same reward
+        results = await asyncio.gather(*[claim_reward() for _ in range(5)])
+
+        # Count results
+        success_count = results.count("SUCCESS")
+        failed_count = results.count("FAILED")
+
+        # At least one should succeed, others should be duplicates
+        assert success_count >= 1, f"At least one claim should succeed, got {results}"
+        assert failed_count == 0, f"No claims should fail, got {results}"
+
+        # Check balance - should only have 500, not 2500
+        final_balance = await credit_system.get_credits(user_id)
+        assert final_balance == 500, f"Expected balance 500, got {final_balance}"
+
+        # Check only one transaction exists
+        transactions = await CreditTransaction.prisma().find_many(
+            where={
+                "userId": user_id,
+                "type": prisma.enums.CreditTransactionType.GRANT,
+                "transactionKey": f"REWARD-{user_id}-WELCOME",
+            }
+        )
+        assert (
+            len(transactions) == 1
+        ), f"Expected 1 reward transaction, got {len(transactions)}"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_integer_overflow_protection(server: SpinTestServer):
+    """Test that integer overflow is prevented by clamping to POSTGRES_INT_MAX."""
+    user_id = f"overflow-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Try to add amount that would overflow
+        max_int = POSTGRES_INT_MAX
+
+        # First, set balance near max
+        await UserBalance.prisma().upsert(
+            where={"userId": user_id},
+            data={
+                "create": {"userId": user_id, "balance": max_int - 100},
+                "update": {"balance": max_int - 100},
+            },
+        )
+
+        # Try to add more than possible - should clamp to POSTGRES_INT_MAX
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=200,
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "overflow_protection"}),
+        )
+
+        # Balance should be clamped to max_int, not overflowed
+        final_balance = await credit_system.get_credits(user_id)
+        assert (
+            final_balance == max_int
+        ), f"Balance should be clamped to {max_int}, got {final_balance}"
+
+        # Verify transaction was created with clamped amount
+        transactions = await CreditTransaction.prisma().find_many(
+            where={
+                "userId": user_id,
+                "type": prisma.enums.CreditTransactionType.TOP_UP,
+            },
+            order={"createdAt": "desc"},
+        )
+        assert len(transactions) > 0, "Transaction should be created"
+        assert (
+            transactions[0].runningBalance == max_int
+        ), "Transaction should show clamped balance"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_high_concurrency_stress(server: SpinTestServer):
+    """Stress test with many concurrent operations."""
+    user_id = f"stress-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Initial balance using internal method (bypasses Stripe)
+        initial_balance = 10000  # $100
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=initial_balance,
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "stress_test_balance"}),
+        )
+
+        # Run many concurrent operations
+        async def random_operation(idx: int):
+            operation = random.choice(["spend", "check"])
+
+            if operation == "spend":
+                amount = random.randint(1, 50)  # $0.01 to $0.50
+                try:
+                    return (
+                        "spend",
+                        amount,
+                        await credit_system.spend_credits(
+                            user_id,
+                            amount,
+                            UsageTransactionMetadata(reason=f"Stress {idx}"),
+                        ),
+                    )
+                except InsufficientBalanceError:
+                    return ("spend_failed", amount, None)
+            else:
+                balance = await credit_system.get_credits(user_id)
+                return ("check", 0, balance)
+
+        # Run 100 concurrent operations
+        results = await asyncio.gather(
+            *[random_operation(i) for i in range(100)], return_exceptions=True
+        )
+
+        # Calculate expected final balance
+        total_spent = sum(
+            r[1]
+            for r in results
+            if not isinstance(r, Exception) and isinstance(r, tuple) and r[0] == "spend"
+        )
+        expected_balance = initial_balance - total_spent
+
+        # Verify final balance
+        final_balance = await credit_system.get_credits(user_id)
+        assert (
+            final_balance == expected_balance
+        ), f"Expected {expected_balance}, got {final_balance}"
+        assert final_balance >= 0, "Balance went negative!"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_concurrent_multiple_spends_sufficient_balance(server: SpinTestServer):
+    """Test multiple concurrent spends when there's sufficient balance for all."""
+    user_id = f"multi-spend-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Give user 150 balance ($1.50) using internal method (bypasses Stripe)
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=150,
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "sufficient_balance"}),
+        )
+
+        # Track individual timing to see serialization
+        timings = {}
+
+        async def spend_with_detailed_timing(amount: int, label: str):
+            start = asyncio.get_event_loop().time()
+            try:
+                await credit_system.spend_credits(
+                    user_id,
+                    amount,
+                    UsageTransactionMetadata(
+                        graph_exec_id=f"concurrent-{label}",
+                        reason=f"Concurrent spend {label}",
+                    ),
+                )
+                end = asyncio.get_event_loop().time()
+                timings[label] = {"start": start, "end": end, "duration": end - start}
+                return f"{label}-SUCCESS"
+            except Exception as e:
+                end = asyncio.get_event_loop().time()
+                timings[label] = {
+                    "start": start,
+                    "end": end,
+                    "duration": end - start,
+                    "error": str(e),
+                }
+                return f"{label}-FAILED: {e}"
+
+        # Run concurrent spends: 10, 20, 30 (total 60, well under 150)
+        overall_start = asyncio.get_event_loop().time()
+        results = await asyncio.gather(
+            spend_with_detailed_timing(10, "spend-10"),
+            spend_with_detailed_timing(20, "spend-20"),
+            spend_with_detailed_timing(30, "spend-30"),
+            return_exceptions=True,
+        )
+        overall_end = asyncio.get_event_loop().time()
+
+        print(f"Results: {results}")
+        print(f"Overall duration: {overall_end - overall_start:.4f}s")
+
+        # Analyze timing to detect serialization vs true concurrency
+        print("\nTiming analysis:")
+        for label, timing in timings.items():
+            print(
+                f"  {label}: started at {timing['start']:.4f}, ended at {timing['end']:.4f}, duration {timing['duration']:.4f}s"
+            )
+
+        # Check if operations overlapped (true concurrency) or were serialized
+        sorted_timings = sorted(timings.items(), key=lambda x: x[1]["start"])
+        print("\nExecution order by start time:")
+        for i, (label, timing) in enumerate(sorted_timings):
+            print(f"  {i+1}. {label}: {timing['start']:.4f} -> {timing['end']:.4f}")
+
+        # Check for overlap (true concurrency) vs serialization
+        overlaps = []
+        for i in range(len(sorted_timings) - 1):
+            current = sorted_timings[i]
+            next_op = sorted_timings[i + 1]
+            if current[1]["end"] > next_op[1]["start"]:
+                overlaps.append(f"{current[0]} overlaps with {next_op[0]}")
+
+        if overlaps:
+            print(f"✅ TRUE CONCURRENCY detected: {overlaps}")
+        else:
+            print("🔒 SERIALIZATION detected: No overlapping execution times")
+
+        # Check final balance
+        final_balance = await credit_system.get_credits(user_id)
+        print(f"Final balance: {final_balance}")
+
+        # Count successes/failures
+        successful = [r for r in results if "SUCCESS" in str(r)]
+        failed = [r for r in results if "FAILED" in str(r)]
+
+        print(f"Successful: {len(successful)}, Failed: {len(failed)}")
+
+        # All should succeed since 150 - (10 + 20 + 30) = 90 > 0
+        assert (
+            len(successful) == 3
+        ), f"Expected all 3 to succeed, got {len(successful)} successes: {results}"
+        assert final_balance == 90, f"Expected balance 90, got {final_balance}"
+
+        # Check transaction timestamps to confirm database-level serialization
+        transactions = await CreditTransaction.prisma().find_many(
+            where={"userId": user_id, "type": prisma.enums.CreditTransactionType.USAGE},
+            order={"createdAt": "asc"},
+        )
+        print("\nDatabase transaction order (by createdAt):")
+        for i, tx in enumerate(transactions):
+            print(
+                f"  {i+1}. Amount {tx.amount}, Running balance: {tx.runningBalance}, Created: {tx.createdAt}"
+            )
+
+        # Verify running balances are chronologically consistent (ordered by createdAt)
+        actual_balances = [
+            tx.runningBalance for tx in transactions if tx.runningBalance is not None
+        ]
+        print(f"Running balances: {actual_balances}")
+
+        # The balances should be valid intermediate states regardless of execution order
+        # Starting balance: 150, spending 10+20+30=60, so final should be 90
+        # The intermediate balances depend on execution order but should all be valid
+        expected_possible_balances = {
+            # If order is 10, 20, 30: [140, 120, 90]
+            # If order is 10, 30, 20: [140, 110, 90]
+            # If order is 20, 10, 30: [130, 120, 90]
+            # If order is 20, 30, 10: [130, 100, 90]
+            # If order is 30, 10, 20: [120, 110, 90]
+            # If order is 30, 20, 10: [120, 100, 90]
+            90,
+            100,
+            110,
+            120,
+            130,
+            140,  # All possible intermediate balances
+        }
+
+        # Verify all balances are valid intermediate states
+        for balance in actual_balances:
+            assert (
+                balance in expected_possible_balances
+            ), f"Invalid balance {balance}, expected one of {expected_possible_balances}"
+
+        # Final balance should always be 90 (150 - 60)
+        assert (
+            min(actual_balances) == 90
+        ), f"Final balance should be 90, got {min(actual_balances)}"
+
+        # The final transaction should always have balance 90
+        # The other transactions should have valid intermediate balances
+        assert (
+            90 in actual_balances
+        ), f"Final balance 90 should be in actual_balances: {actual_balances}"
+
+        # All balances should be >= 90 (the final state)
+        assert all(
+            balance >= 90 for balance in actual_balances
+        ), f"All balances should be >= 90, got {actual_balances}"
+
+        # CRITICAL: Transactions are atomic but can complete in any order
+        # What matters is that all running balances are valid intermediate states
+        # Each balance should be between 90 (final) and 140 (after first transaction)
+        for balance in actual_balances:
+            assert (
+                90 <= balance <= 140
+            ), f"Balance {balance} is outside valid range [90, 140]"
+
+        # Final balance (minimum) should always be 90
+        assert (
+            min(actual_balances) == 90
+        ), f"Final balance should be 90, got {min(actual_balances)}"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_prove_database_locking_behavior(server: SpinTestServer):
+    """Definitively prove whether database locking causes waiting vs failures."""
+    user_id = f"locking-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Set balance to exact amount that can handle all spends using internal method (bypasses Stripe)
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=60,  # Exactly 10+20+30
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "exact_amount_test"}),
+        )
+
+        async def spend_with_precise_timing(amount: int, label: str):
+            request_start = asyncio.get_event_loop().time()
+            db_operation_start = asyncio.get_event_loop().time()
+            try:
+                # Add a small delay to increase chance of true concurrency
+                await asyncio.sleep(0.001)
+
+                db_operation_start = asyncio.get_event_loop().time()
+                await credit_system.spend_credits(
+                    user_id,
+                    amount,
+                    UsageTransactionMetadata(
+                        graph_exec_id=f"locking-{label}",
+                        reason=f"Locking test {label}",
+                    ),
+                )
+                db_operation_end = asyncio.get_event_loop().time()
+
+                return {
+                    "label": label,
+                    "status": "SUCCESS",
+                    "request_start": request_start,
+                    "db_start": db_operation_start,
+                    "db_end": db_operation_end,
+                    "db_duration": db_operation_end - db_operation_start,
+                }
+            except Exception as e:
+                db_operation_end = asyncio.get_event_loop().time()
+                return {
+                    "label": label,
+                    "status": "FAILED",
+                    "error": str(e),
+                    "request_start": request_start,
+                    "db_start": db_operation_start,
+                    "db_end": db_operation_end,
+                    "db_duration": db_operation_end - db_operation_start,
+                }
+
+        # Launch all requests simultaneously
+        results = await asyncio.gather(
+            spend_with_precise_timing(10, "A"),
+            spend_with_precise_timing(20, "B"),
+            spend_with_precise_timing(30, "C"),
+            return_exceptions=True,
+        )
+
+        print("\n🔍 LOCKING BEHAVIOR ANALYSIS:")
+        print("=" * 50)
+
+        successful = [
+            r for r in results if isinstance(r, dict) and r.get("status") == "SUCCESS"
+        ]
+        failed = [
+            r for r in results if isinstance(r, dict) and r.get("status") == "FAILED"
+        ]
+
+        print(f"✅ Successful operations: {len(successful)}")
+        print(f"❌ Failed operations: {len(failed)}")
+
+        if len(failed) > 0:
+            print(
+                "\n🚫 CONCURRENT FAILURES - Some requests failed due to insufficient balance:"
+            )
+            for result in failed:
+                if isinstance(result, dict):
+                    print(
+                        f"   {result['label']}: {result.get('error', 'Unknown error')}"
+                    )
+
+        if len(successful) == 3:
+            print(
+                "\n🔒 SERIALIZATION CONFIRMED - All requests succeeded, indicating they were queued:"
+            )
+
+            # Sort by actual execution time to see order
+            dict_results = [r for r in results if isinstance(r, dict)]
+            sorted_results = sorted(dict_results, key=lambda x: x["db_start"])
+
+            for i, result in enumerate(sorted_results):
+                print(
+                    f"   {i+1}. {result['label']}: DB operation took {result['db_duration']:.4f}s"
+                )
+
+            # Check if any operations overlapped at the database level
+            print("\n⏱️  Database operation timeline:")
+            for result in sorted_results:
+                print(
+                    f"   {result['label']}: {result['db_start']:.4f} -> {result['db_end']:.4f}"
+                )
+
+        # Verify final state
+        final_balance = await credit_system.get_credits(user_id)
+        print(f"\n💰 Final balance: {final_balance}")
+
+        if len(successful) == 3:
+            assert (
+                final_balance == 0
+            ), f"If all succeeded, balance should be 0, got {final_balance}"
+            print(
+                "✅ CONCLUSION: Database row locking causes requests to WAIT and execute serially"
+            )
+        else:
+            print(
+                "❌ CONCLUSION: Some requests failed, indicating different concurrency behavior"
+            )
+
+    finally:
+        await cleanup_test_user(user_id)

autogpt_platform/backend/backend/data/credit_integration_test.py

@@ -0,0 +1,277 @@
+"""
+Integration tests for credit system to catch SQL enum casting issues.
+
+These tests run actual database operations to ensure SQL queries work correctly,
+which would have caught the CreditTransactionType enum casting bug.
+"""
+
+import pytest
+from prisma.enums import CreditTransactionType
+from prisma.models import CreditTransaction, User, UserBalance
+
+from backend.data.credit import (
+    AutoTopUpConfig,
+    BetaUserCredit,
+    UsageTransactionMetadata,
+    get_auto_top_up,
+    set_auto_top_up,
+)
+from backend.util.json import SafeJson
+
+
+@pytest.fixture
+async def cleanup_test_user():
+    """Clean up test user data before and after tests."""
+    import uuid
+
+    user_id = str(uuid.uuid4())  # Use unique user ID for each test
+
+    # Create the user first
+    try:
+        await User.prisma().create(
+            data={
+                "id": user_id,
+                "email": f"test-{user_id}@example.com",
+                "topUpConfig": SafeJson({}),
+                "timezone": "UTC",
+            }
+        )
+    except Exception:
+        # User might already exist, that's fine
+        pass
+
+    yield user_id
+
+    # Cleanup after test
+    await CreditTransaction.prisma().delete_many(where={"userId": user_id})
+    await UserBalance.prisma().delete_many(where={"userId": user_id})
+    # Clear auto-top-up config before deleting user
+    await User.prisma().update(
+        where={"id": user_id}, data={"topUpConfig": SafeJson({})}
+    )
+    await User.prisma().delete(where={"id": user_id})
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_credit_transaction_enum_casting_integration(cleanup_test_user):
+    """
+    Integration test to verify CreditTransactionType enum casting works in SQL queries.
+
+    This test would have caught the enum casting bug where PostgreSQL expected
+    platform."CreditTransactionType" but got "CreditTransactionType".
+    """
+    user_id = cleanup_test_user
+    credit_system = BetaUserCredit(1000)
+
+    # Test each transaction type to ensure enum casting works
+    test_cases = [
+        (CreditTransactionType.TOP_UP, 100, "Test top-up"),
+        (CreditTransactionType.USAGE, -50, "Test usage"),
+        (CreditTransactionType.GRANT, 200, "Test grant"),
+        (CreditTransactionType.REFUND, -25, "Test refund"),
+        (CreditTransactionType.CARD_CHECK, 0, "Test card check"),
+    ]
+
+    for transaction_type, amount, reason in test_cases:
+        metadata = SafeJson({"reason": reason, "test": "enum_casting"})
+
+        # This call would fail with enum casting error before the fix
+        balance, tx_key = await credit_system._add_transaction(
+            user_id=user_id,
+            amount=amount,
+            transaction_type=transaction_type,
+            metadata=metadata,
+            is_active=True,
+        )
+
+        # Verify transaction was created with correct type
+        transaction = await CreditTransaction.prisma().find_first(
+            where={"userId": user_id, "transactionKey": tx_key}
+        )
+
+        assert transaction is not None
+        assert transaction.type == transaction_type
+        assert transaction.amount == amount
+        assert transaction.metadata is not None
+
+        # Verify metadata content
+        assert transaction.metadata["reason"] == reason
+        assert transaction.metadata["test"] == "enum_casting"
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_auto_top_up_integration(cleanup_test_user, monkeypatch):
+    """
+    Integration test for auto-top-up functionality that triggers enum casting.
+
+    This tests the complete auto-top-up flow which involves SQL queries with
+    CreditTransactionType enums, ensuring enum casting works end-to-end.
+    """
+    # Enable credits for this test
+    from backend.data.credit import settings
+
+    monkeypatch.setattr(settings.config, "enable_credit", True)
+    monkeypatch.setattr(settings.config, "enable_beta_monthly_credit", True)
+    monkeypatch.setattr(settings.config, "num_user_credits_refill", 1000)
+
+    user_id = cleanup_test_user
+    credit_system = BetaUserCredit(1000)
+
+    # First add some initial credits so we can test the configuration and subsequent behavior
+    balance, _ = await credit_system._add_transaction(
+        user_id=user_id,
+        amount=50,  # Below threshold that we'll set
+        transaction_type=CreditTransactionType.GRANT,
+        metadata=SafeJson({"reason": "Initial credits before auto top-up config"}),
+    )
+    assert balance == 50
+
+    # Configure auto top-up with threshold above current balance
+    config = AutoTopUpConfig(threshold=100, amount=500)
+    await set_auto_top_up(user_id, config)
+
+    # Verify configuration was saved but no immediate top-up occurred
+    current_balance = await credit_system.get_credits(user_id)
+    assert current_balance == 50  # Balance should be unchanged
+
+    # Simulate spending credits that would trigger auto top-up
+    # This involves multiple SQL operations with enum casting
+    try:
+        metadata = UsageTransactionMetadata(reason="Test spend to trigger auto top-up")
+        await credit_system.spend_credits(user_id=user_id, cost=10, metadata=metadata)
+
+        # The auto top-up mechanism should have been triggered
+        # Verify the transaction types were handled correctly
+        transactions = await CreditTransaction.prisma().find_many(
+            where={"userId": user_id}, order={"createdAt": "desc"}
+        )
+
+        # Should have at least: GRANT (initial), USAGE (spend), and TOP_UP (auto top-up)
+        assert len(transactions) >= 3
+
+        # Verify different transaction types exist and enum casting worked
+        transaction_types = {t.type for t in transactions}
+        assert CreditTransactionType.GRANT in transaction_types
+        assert CreditTransactionType.USAGE in transaction_types
+        assert (
+            CreditTransactionType.TOP_UP in transaction_types
+        )  # Auto top-up should have triggered
+
+    except Exception as e:
+        # If this fails with enum casting error, the test successfully caught the bug
+        if "CreditTransactionType" in str(e) and (
+            "cast" in str(e).lower() or "type" in str(e).lower()
+        ):
+            pytest.fail(f"Enum casting error detected: {e}")
+        else:
+            # Re-raise other unexpected errors
+            raise
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_enable_transaction_enum_casting_integration(cleanup_test_user):
+    """
+    Integration test for _enable_transaction with enum casting.
+
+    Tests the scenario where inactive transactions are enabled, which also
+    involves SQL queries with CreditTransactionType enum casting.
+    """
+    user_id = cleanup_test_user
+    credit_system = BetaUserCredit(1000)
+
+    # Create an inactive transaction
+    balance, tx_key = await credit_system._add_transaction(
+        user_id=user_id,
+        amount=100,
+        transaction_type=CreditTransactionType.TOP_UP,
+        metadata=SafeJson({"reason": "Inactive transaction test"}),
+        is_active=False,  # Create as inactive
+    )
+
+    # Balance should be 0 since transaction is inactive
+    assert balance == 0
+
+    # Enable the transaction with new metadata
+    enable_metadata = SafeJson(
+        {
+            "payment_method": "test_payment",
+            "activation_reason": "Integration test activation",
+        }
+    )
+
+    # This would fail with enum casting error before the fix
+    final_balance = await credit_system._enable_transaction(
+        transaction_key=tx_key,
+        user_id=user_id,
+        metadata=enable_metadata,
+    )
+
+    # Now balance should reflect the activated transaction
+    assert final_balance == 100
+
+    # Verify transaction was properly enabled with correct enum type
+    transaction = await CreditTransaction.prisma().find_first(
+        where={"userId": user_id, "transactionKey": tx_key}
+    )
+
+    assert transaction is not None
+    assert transaction.isActive is True
+    assert transaction.type == CreditTransactionType.TOP_UP
+    assert transaction.runningBalance == 100
+
+    # Verify metadata was updated
+    assert transaction.metadata is not None
+    assert transaction.metadata["payment_method"] == "test_payment"
+    assert transaction.metadata["activation_reason"] == "Integration test activation"
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_auto_top_up_configuration_storage(cleanup_test_user, monkeypatch):
+    """
+    Test that auto-top-up configuration is properly stored and retrieved.
+
+    The immediate top-up logic is handled by the API routes, not the core
+    set_auto_top_up function. This test verifies the configuration is correctly
+    saved and can be retrieved.
+    """
+    # Enable credits for this test
+    from backend.data.credit import settings
+
+    monkeypatch.setattr(settings.config, "enable_credit", True)
+    monkeypatch.setattr(settings.config, "enable_beta_monthly_credit", True)
+    monkeypatch.setattr(settings.config, "num_user_credits_refill", 1000)
+
+    user_id = cleanup_test_user
+    credit_system = BetaUserCredit(1000)
+
+    # Set initial balance
+    balance, _ = await credit_system._add_transaction(
+        user_id=user_id,
+        amount=50,
+        transaction_type=CreditTransactionType.GRANT,
+        metadata=SafeJson({"reason": "Initial balance for config test"}),
+    )
+
+    assert balance == 50
+
+    # Configure auto top-up
+    config = AutoTopUpConfig(threshold=100, amount=200)
+    await set_auto_top_up(user_id, config)
+
+    # Verify the configuration was saved
+    retrieved_config = await get_auto_top_up(user_id)
+    assert retrieved_config.threshold == config.threshold
+    assert retrieved_config.amount == config.amount
+
+    # Verify balance is unchanged (no immediate top-up from set_auto_top_up)
+    final_balance = await credit_system.get_credits(user_id)
+    assert final_balance == 50  # Should be unchanged
+
+    # Verify no immediate auto-top-up transaction was created by set_auto_top_up
+    transactions = await CreditTransaction.prisma().find_many(
+        where={"userId": user_id}, order={"createdAt": "desc"}
+    )
+
+    # Should only have the initial GRANT transaction
+    assert len(transactions) == 1
+    assert transactions[0].type == CreditTransactionType.GRANT

autogpt_platform/backend/backend/data/credit_metadata_test.py

@@ -0,0 +1,141 @@
+"""
+Tests for credit system metadata handling to ensure JSON casting works correctly.
+
+This test verifies that metadata parameters are properly serialized when passed
+to raw SQL queries with JSONB columns.
+"""
+
+# type: ignore
+
+from typing import Any
+
+import pytest
+from prisma.enums import CreditTransactionType
+from prisma.models import CreditTransaction, UserBalance
+
+from backend.data.credit import BetaUserCredit
+from backend.data.user import DEFAULT_USER_ID
+from backend.util.json import SafeJson
+
+
+@pytest.fixture
+async def setup_test_user():
+    """Setup test user and cleanup after test."""
+    user_id = DEFAULT_USER_ID
+
+    # Cleanup before test
+    await CreditTransaction.prisma().delete_many(where={"userId": user_id})
+    await UserBalance.prisma().delete_many(where={"userId": user_id})
+
+    yield user_id
+
+    # Cleanup after test
+    await CreditTransaction.prisma().delete_many(where={"userId": user_id})
+    await UserBalance.prisma().delete_many(where={"userId": user_id})
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_metadata_json_serialization(setup_test_user):
+    """Test that metadata is properly serialized for JSONB column in raw SQL."""
+    user_id = setup_test_user
+    credit_system = BetaUserCredit(1000)
+
+    # Test with complex metadata that would fail if not properly serialized
+    complex_metadata = SafeJson(
+        {
+            "graph_exec_id": "test-12345",
+            "reason": "Testing metadata serialization",
+            "nested_data": {
+                "key1": "value1",
+                "key2": ["array", "of", "values"],
+                "key3": {"deeply": {"nested": "object"}},
+            },
+            "special_chars": "Testing 'quotes' and \"double quotes\" and unicode: 🚀",
+        }
+    )
+
+    # This should work without throwing a JSONB casting error
+    balance, tx_key = await credit_system._add_transaction(
+        user_id=user_id,
+        amount=500,  # $5 top-up
+        transaction_type=CreditTransactionType.TOP_UP,
+        metadata=complex_metadata,
+        is_active=True,
+    )
+
+    # Verify the transaction was created successfully
+    assert balance == 500
+
+    # Verify the metadata was stored correctly in the database
+    transaction = await CreditTransaction.prisma().find_first(
+        where={"userId": user_id, "transactionKey": tx_key}
+    )
+
+    assert transaction is not None
+    assert transaction.metadata is not None
+
+    # Verify the metadata contains our complex data
+    metadata_dict: dict[str, Any] = dict(transaction.metadata)  # type: ignore
+    assert metadata_dict["graph_exec_id"] == "test-12345"
+    assert metadata_dict["reason"] == "Testing metadata serialization"
+    assert metadata_dict["nested_data"]["key1"] == "value1"
+    assert metadata_dict["nested_data"]["key3"]["deeply"]["nested"] == "object"
+    assert (
+        metadata_dict["special_chars"]
+        == "Testing 'quotes' and \"double quotes\" and unicode: 🚀"
+    )
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_enable_transaction_metadata_serialization(setup_test_user):
+    """Test that _enable_transaction also handles metadata JSON serialization correctly."""
+    user_id = setup_test_user
+    credit_system = BetaUserCredit(1000)
+
+    # First create an inactive transaction
+    balance, tx_key = await credit_system._add_transaction(
+        user_id=user_id,
+        amount=300,
+        transaction_type=CreditTransactionType.TOP_UP,
+        metadata=SafeJson({"initial": "inactive_transaction"}),
+        is_active=False,  # Create as inactive
+    )
+
+    # Initial balance should be 0 because transaction is inactive
+    assert balance == 0
+
+    # Now enable the transaction with new metadata
+    enable_metadata = SafeJson(
+        {
+            "payment_method": "stripe",
+            "payment_intent": "pi_test_12345",
+            "activation_reason": "Payment confirmed",
+            "complex_data": {"array": [1, 2, 3], "boolean": True, "null_value": None},
+        }
+    )
+
+    # This should work without JSONB casting errors
+    final_balance = await credit_system._enable_transaction(
+        transaction_key=tx_key,
+        user_id=user_id,
+        metadata=enable_metadata,
+    )
+
+    # Now balance should reflect the activated transaction
+    assert final_balance == 300
+
+    # Verify the metadata was updated correctly
+    transaction = await CreditTransaction.prisma().find_first(
+        where={"userId": user_id, "transactionKey": tx_key}
+    )
+
+    assert transaction is not None
+    assert transaction.isActive is True
+
+    # Verify the metadata was updated with enable_metadata
+    metadata_dict: dict[str, Any] = dict(transaction.metadata)  # type: ignore
+    assert metadata_dict["payment_method"] == "stripe"
+    assert metadata_dict["payment_intent"] == "pi_test_12345"
+    assert metadata_dict["complex_data"]["array"] == [1, 2, 3]
+    assert metadata_dict["complex_data"]["boolean"] is True
+    assert metadata_dict["complex_data"]["null_value"] is None

autogpt_platform/backend/backend/data/credit_refund_test.py

@@ -0,0 +1,372 @@
+"""
+Tests for credit system refund and dispute operations.
+
+These tests ensure that refund operations (deduct_credits, handle_dispute)
+are atomic and maintain data consistency.
+"""
+
+from datetime import datetime, timezone
+from unittest.mock import MagicMock, patch
+
+import pytest
+import stripe
+from prisma.enums import CreditTransactionType
+from prisma.models import CreditRefundRequest, CreditTransaction, User, UserBalance
+
+from backend.data.credit import UserCredit
+from backend.util.json import SafeJson
+from backend.util.test import SpinTestServer
+
+credit_system = UserCredit()
+
+# Test user ID for refund tests
+REFUND_TEST_USER_ID = "refund-test-user"
+
+
+async def setup_test_user_with_topup():
+    """Create a test user with initial balance and a top-up transaction."""
+    # Clean up any existing data
+    await CreditRefundRequest.prisma().delete_many(
+        where={"userId": REFUND_TEST_USER_ID}
+    )
+    await CreditTransaction.prisma().delete_many(where={"userId": REFUND_TEST_USER_ID})
+    await UserBalance.prisma().delete_many(where={"userId": REFUND_TEST_USER_ID})
+    await User.prisma().delete_many(where={"id": REFUND_TEST_USER_ID})
+
+    # Create user
+    await User.prisma().create(
+        data={
+            "id": REFUND_TEST_USER_ID,
+            "email": f"{REFUND_TEST_USER_ID}@example.com",
+            "name": "Refund Test User",
+        }
+    )
+
+    # Create user balance
+    await UserBalance.prisma().create(
+        data={
+            "userId": REFUND_TEST_USER_ID,
+            "balance": 1000,  # $10
+        }
+    )
+
+    # Create a top-up transaction that can be refunded
+    topup_tx = await CreditTransaction.prisma().create(
+        data={
+            "userId": REFUND_TEST_USER_ID,
+            "amount": 1000,
+            "type": CreditTransactionType.TOP_UP,
+            "transactionKey": "pi_test_12345",
+            "runningBalance": 1000,
+            "isActive": True,
+            "metadata": SafeJson({"stripe_payment_intent": "pi_test_12345"}),
+        }
+    )
+
+    return topup_tx
+
+
+async def cleanup_test_user():
+    """Clean up test data."""
+    await CreditRefundRequest.prisma().delete_many(
+        where={"userId": REFUND_TEST_USER_ID}
+    )
+    await CreditTransaction.prisma().delete_many(where={"userId": REFUND_TEST_USER_ID})
+    await UserBalance.prisma().delete_many(where={"userId": REFUND_TEST_USER_ID})
+    await User.prisma().delete_many(where={"id": REFUND_TEST_USER_ID})
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_deduct_credits_atomic(server: SpinTestServer):
+    """Test that deduct_credits is atomic and creates transaction correctly."""
+    topup_tx = await setup_test_user_with_topup()
+
+    try:
+        # Create a mock refund object
+        refund = MagicMock(spec=stripe.Refund)
+        refund.id = "re_test_refund_123"
+        refund.payment_intent = topup_tx.transactionKey
+        refund.amount = 500  # Refund $5 of the $10 top-up
+        refund.status = "succeeded"
+        refund.reason = "requested_by_customer"
+        refund.created = int(datetime.now(timezone.utc).timestamp())
+
+        # Create refund request record (simulating webhook flow)
+        await CreditRefundRequest.prisma().create(
+            data={
+                "userId": REFUND_TEST_USER_ID,
+                "amount": 500,
+                "transactionKey": topup_tx.transactionKey,  # Should match the original transaction
+                "reason": "Test refund",
+            }
+        )
+
+        # Call deduct_credits
+        await credit_system.deduct_credits(refund)
+
+        # Verify the user's balance was deducted
+        user_balance = await UserBalance.prisma().find_unique(
+            where={"userId": REFUND_TEST_USER_ID}
+        )
+        assert user_balance is not None
+        assert (
+            user_balance.balance == 500
+        ), f"Expected balance 500, got {user_balance.balance}"
+
+        # Verify refund transaction was created
+        refund_tx = await CreditTransaction.prisma().find_first(
+            where={
+                "userId": REFUND_TEST_USER_ID,
+                "type": CreditTransactionType.REFUND,
+                "transactionKey": refund.id,
+            }
+        )
+        assert refund_tx is not None
+        assert refund_tx.amount == -500
+        assert refund_tx.runningBalance == 500
+        assert refund_tx.isActive
+
+        # Verify refund request was updated
+        refund_request = await CreditRefundRequest.prisma().find_first(
+            where={
+                "userId": REFUND_TEST_USER_ID,
+                "transactionKey": topup_tx.transactionKey,
+            }
+        )
+        assert refund_request is not None
+        assert (
+            refund_request.result
+            == "The refund request has been approved, the amount will be credited back to your account."
+        )
+
+    finally:
+        await cleanup_test_user()
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_deduct_credits_user_not_found(server: SpinTestServer):
+    """Test that deduct_credits raises error if transaction not found (which means user doesn't exist)."""
+    # Create a mock refund object that references a non-existent payment intent
+    refund = MagicMock(spec=stripe.Refund)
+    refund.id = "re_test_refund_nonexistent"
+    refund.payment_intent = "pi_test_nonexistent"  # This payment intent doesn't exist
+    refund.amount = 500
+    refund.status = "succeeded"
+    refund.reason = "requested_by_customer"
+    refund.created = int(datetime.now(timezone.utc).timestamp())
+
+    # Should raise error for missing transaction
+    with pytest.raises(Exception):  # Should raise NotFoundError for missing transaction
+        await credit_system.deduct_credits(refund)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+@patch("backend.data.credit.settings")
+@patch("stripe.Dispute.modify")
+@patch("backend.data.credit.get_user_by_id")
+async def test_handle_dispute_with_sufficient_balance(
+    mock_get_user, mock_stripe_modify, mock_settings, server: SpinTestServer
+):
+    """Test handling dispute when user has sufficient balance (dispute gets closed)."""
+    topup_tx = await setup_test_user_with_topup()
+
+    try:
+        # Mock settings to have a low tolerance threshold
+        mock_settings.config.refund_credit_tolerance_threshold = 0
+
+        # Mock the user lookup
+        mock_user = MagicMock()
+        mock_user.email = f"{REFUND_TEST_USER_ID}@example.com"
+        mock_get_user.return_value = mock_user
+
+        # Create a mock dispute object for small amount (user has 1000, disputing 100)
+        dispute = MagicMock(spec=stripe.Dispute)
+        dispute.id = "dp_test_dispute_123"
+        dispute.payment_intent = topup_tx.transactionKey
+        dispute.amount = 100  # Small dispute amount
+        dispute.status = "pending"
+        dispute.reason = "fraudulent"
+        dispute.created = int(datetime.now(timezone.utc).timestamp())
+
+        # Mock the close method to prevent real API calls
+        dispute.close = MagicMock()
+
+        # Handle the dispute
+        await credit_system.handle_dispute(dispute)
+
+        # Verify dispute.close() was called (since user has sufficient balance)
+        dispute.close.assert_called_once()
+
+        # Verify no stripe evidence was added since dispute was closed
+        mock_stripe_modify.assert_not_called()
+
+        # Verify the user's balance was NOT deducted (dispute was closed)
+        user_balance = await UserBalance.prisma().find_unique(
+            where={"userId": REFUND_TEST_USER_ID}
+        )
+        assert user_balance is not None
+        assert (
+            user_balance.balance == 1000
+        ), f"Balance should remain 1000, got {user_balance.balance}"
+
+    finally:
+        await cleanup_test_user()
+
+
+@pytest.mark.asyncio(loop_scope="session")
+@patch("backend.data.credit.settings")
+@patch("stripe.Dispute.modify")
+@patch("backend.data.credit.get_user_by_id")
+async def test_handle_dispute_with_insufficient_balance(
+    mock_get_user, mock_stripe_modify, mock_settings, server: SpinTestServer
+):
+    """Test handling dispute when user has insufficient balance (evidence gets added)."""
+    topup_tx = await setup_test_user_with_topup()
+
+    # Save original method for restoration before any try blocks
+    original_get_history = credit_system.get_transaction_history
+
+    try:
+        # Mock settings to have a high tolerance threshold so dispute isn't closed
+        mock_settings.config.refund_credit_tolerance_threshold = 2000
+
+        # Mock the user lookup
+        mock_user = MagicMock()
+        mock_user.email = f"{REFUND_TEST_USER_ID}@example.com"
+        mock_get_user.return_value = mock_user
+
+        # Mock the transaction history method to return an async result
+        from unittest.mock import AsyncMock
+
+        mock_history = MagicMock()
+        mock_history.transactions = []
+        credit_system.get_transaction_history = AsyncMock(return_value=mock_history)
+
+        # Create a mock dispute object for full amount (user has 1000, disputing 1000)
+        dispute = MagicMock(spec=stripe.Dispute)
+        dispute.id = "dp_test_dispute_pending"
+        dispute.payment_intent = topup_tx.transactionKey
+        dispute.amount = 1000
+        dispute.status = "warning_needs_response"
+        dispute.created = int(datetime.now(timezone.utc).timestamp())
+
+        # Mock the close method to prevent real API calls
+        dispute.close = MagicMock()
+
+        # Handle the dispute (evidence should be added)
+        await credit_system.handle_dispute(dispute)
+
+        # Verify dispute.close() was NOT called (insufficient balance after tolerance)
+        dispute.close.assert_not_called()
+
+        # Verify stripe evidence was added since dispute wasn't closed
+        mock_stripe_modify.assert_called_once()
+
+        # Verify the user's balance was NOT deducted (handle_dispute doesn't deduct credits)
+        user_balance = await UserBalance.prisma().find_unique(
+            where={"userId": REFUND_TEST_USER_ID}
+        )
+        assert user_balance is not None
+        assert user_balance.balance == 1000, "Balance should remain unchanged"
+
+    finally:
+        credit_system.get_transaction_history = original_get_history
+        await cleanup_test_user()
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_concurrent_refunds(server: SpinTestServer):
+    """Test that concurrent refunds are handled atomically."""
+    import asyncio
+
+    topup_tx = await setup_test_user_with_topup()
+
+    try:
+        # Create multiple refund requests
+        refund_requests = []
+        for i in range(5):
+            req = await CreditRefundRequest.prisma().create(
+                data={
+                    "userId": REFUND_TEST_USER_ID,
+                    "amount": 100,  # $1 each
+                    "transactionKey": topup_tx.transactionKey,
+                    "reason": f"Test refund {i}",
+                }
+            )
+            refund_requests.append(req)
+
+        # Create refund tasks to run concurrently
+        async def process_refund(index: int):
+            refund = MagicMock(spec=stripe.Refund)
+            refund.id = f"re_test_concurrent_{index}"
+            refund.payment_intent = topup_tx.transactionKey
+            refund.amount = 100  # $1 refund
+            refund.status = "succeeded"
+            refund.reason = "requested_by_customer"
+            refund.created = int(datetime.now(timezone.utc).timestamp())
+
+            try:
+                await credit_system.deduct_credits(refund)
+                return "success"
+            except Exception as e:
+                return f"error: {e}"
+
+        # Run refunds concurrently
+        results = await asyncio.gather(
+            *[process_refund(i) for i in range(5)], return_exceptions=True
+        )
+
+        # All should succeed
+        assert all(r == "success" for r in results), f"Some refunds failed: {results}"
+
+        # Verify final balance - with non-atomic implementation, this will demonstrate race condition
+        # EXPECTED BEHAVIOR: Due to race conditions, not all refunds will be properly processed
+        # The balance will be incorrect (higher than expected) showing lost updates
+        user_balance = await UserBalance.prisma().find_unique(
+            where={"userId": REFUND_TEST_USER_ID}
+        )
+        assert user_balance is not None
+
+        # With atomic implementation, this should be 500 (1000 - 5*100)
+        # With current non-atomic implementation, this will likely be wrong due to race conditions
+        print(f"DEBUG: Final balance = {user_balance.balance}, expected = 500")
+
+        # With atomic implementation, all 5 refunds should process correctly
+        assert (
+            user_balance.balance == 500
+        ), f"Expected balance 500 after 5 refunds of 100 each, got {user_balance.balance}"
+
+        # Verify all refund transactions exist
+        refund_txs = await CreditTransaction.prisma().find_many(
+            where={
+                "userId": REFUND_TEST_USER_ID,
+                "type": CreditTransactionType.REFUND,
+            }
+        )
+        assert (
+            len(refund_txs) == 5
+        ), f"Expected 5 refund transactions, got {len(refund_txs)}"
+
+        running_balances: set[int] = {
+            tx.runningBalance for tx in refund_txs if tx.runningBalance is not None
+        }
+
+        # Verify all balances are valid intermediate states
+        for balance in running_balances:
+            assert (
+                500 <= balance <= 1000
+            ), f"Invalid balance {balance}, should be between 500 and 1000"
+
+        # Final balance should be present
+        assert (
+            500 in running_balances
+        ), f"Final balance 500 should be in {running_balances}"
+
+        # All balances should be unique and form a valid sequence
+        sorted_balances = sorted(running_balances, reverse=True)
+        assert (
+            len(sorted_balances) == 5
+        ), f"Expected 5 unique balances, got {len(sorted_balances)}"
+
+    finally:
+        await cleanup_test_user()

autogpt_platform/backend/backend/data/credit_test.py

@@ -1,8 +1,8 @@
-from datetime import datetime, timezone
+from datetime import datetime, timedelta, timezone
 
 import pytest
 from prisma.enums import CreditTransactionType
-from prisma.models import CreditTransaction
+from prisma.models import CreditTransaction, UserBalance
 
 from backend.blocks.llm import AITextGeneratorBlock
 from backend.data.block import get_block
@@ -19,14 +19,24 @@ user_credit = BetaUserCredit(REFILL_VALUE)
 
 async def disable_test_user_transactions():
     await CreditTransaction.prisma().delete_many(where={"userId": DEFAULT_USER_ID})
+    # Also reset the balance to 0 and set updatedAt to old date to trigger monthly refill
+    old_date = datetime.now(timezone.utc) - timedelta(days=35)  # More than a month ago
+    await UserBalance.prisma().upsert(
+        where={"userId": DEFAULT_USER_ID},
+        data={
+            "create": {"userId": DEFAULT_USER_ID, "balance": 0},
+            "update": {"balance": 0, "updatedAt": old_date},
+        },
+    )
 
 
 async def top_up(amount: int):
-    await user_credit._add_transaction(
+    balance, _ = await user_credit._add_transaction(
         DEFAULT_USER_ID,
         amount,
         CreditTransactionType.TOP_UP,
     )
+    return balance
 
 
 async def spend_credits(entry: NodeExecutionEntry) -> int:
@@ -111,29 +121,90 @@ async def test_block_credit_top_up(server: SpinTestServer):
 
 @pytest.mark.asyncio(loop_scope="session")
 async def test_block_credit_reset(server: SpinTestServer):
+    """Test that BetaUserCredit provides monthly refills correctly."""
     await disable_test_user_transactions()
-    month1 = 1
-    month2 = 2
 
-    # set the calendar to month 2 but use current time from now
-    user_credit.time_now = lambda: datetime.now(timezone.utc).replace(
-        month=month2, day=1
-    )
-    month2credit = await user_credit.get_credits(DEFAULT_USER_ID)
+    # Save original time_now function for restoration
+    original_time_now = user_credit.time_now
 
-    # Month 1 result should only affect month 1
-    user_credit.time_now = lambda: datetime.now(timezone.utc).replace(
-        month=month1, day=1
-    )
-    month1credit = await user_credit.get_credits(DEFAULT_USER_ID)
-    await top_up(100)
-    assert await user_credit.get_credits(DEFAULT_USER_ID) == month1credit + 100
+    try:
+        # Test month 1 behavior
+        month1 = datetime.now(timezone.utc).replace(month=1, day=1)
+        user_credit.time_now = lambda: month1
 
-    # Month 2 balance is unaffected
-    user_credit.time_now = lambda: datetime.now(timezone.utc).replace(
-        month=month2, day=1
-    )
-    assert await user_credit.get_credits(DEFAULT_USER_ID) == month2credit
+        # First call in month 1 should trigger refill
+        balance = await user_credit.get_credits(DEFAULT_USER_ID)
+        assert balance == REFILL_VALUE  # Should get 1000 credits
+
+        # Manually create a transaction with month 1 timestamp to establish history
+        await CreditTransaction.prisma().create(
+            data={
+                "userId": DEFAULT_USER_ID,
+                "amount": 100,
+                "type": CreditTransactionType.TOP_UP,
+                "runningBalance": 1100,
+                "isActive": True,
+                "createdAt": month1,  # Set specific timestamp
+            }
+        )
+
+        # Update user balance to match
+        await UserBalance.prisma().upsert(
+            where={"userId": DEFAULT_USER_ID},
+            data={
+                "create": {"userId": DEFAULT_USER_ID, "balance": 1100},
+                "update": {"balance": 1100},
+            },
+        )
+
+        # Now test month 2 behavior
+        month2 = datetime.now(timezone.utc).replace(month=2, day=1)
+        user_credit.time_now = lambda: month2
+
+        # In month 2, since balance (1100) > refill (1000), no refill should happen
+        month2_balance = await user_credit.get_credits(DEFAULT_USER_ID)
+        assert month2_balance == 1100  # Balance persists, no reset
+
+        # Now test the refill behavior when balance is low
+        # Set balance below refill threshold
+        await UserBalance.prisma().update(
+            where={"userId": DEFAULT_USER_ID}, data={"balance": 400}
+        )
+
+        # Create a month 2 transaction to update the last transaction time
+        await CreditTransaction.prisma().create(
+            data={
+                "userId": DEFAULT_USER_ID,
+                "amount": -700,  # Spent 700 to get to 400
+                "type": CreditTransactionType.USAGE,
+                "runningBalance": 400,
+                "isActive": True,
+                "createdAt": month2,
+            }
+        )
+
+        # Move to month 3
+        month3 = datetime.now(timezone.utc).replace(month=3, day=1)
+        user_credit.time_now = lambda: month3
+
+        # Should get refilled since balance (400) < refill value (1000)
+        month3_balance = await user_credit.get_credits(DEFAULT_USER_ID)
+        assert month3_balance == REFILL_VALUE  # Should be refilled to 1000
+
+        # Verify the refill transaction was created
+        refill_tx = await CreditTransaction.prisma().find_first(
+            where={
+                "userId": DEFAULT_USER_ID,
+                "type": CreditTransactionType.GRANT,
+                "transactionKey": {"contains": "MONTHLY-CREDIT-TOP-UP"},
+            },
+            order={"createdAt": "desc"},
+        )
+        assert refill_tx is not None, "Monthly refill transaction should be created"
+        assert refill_tx.amount == 600, "Refill should be 600 (1000 - 400)"
+    finally:
+        # Restore original time_now function
+        user_credit.time_now = original_time_now
 
 
 @pytest.mark.asyncio(loop_scope="session")

autogpt_platform/backend/backend/data/credit_underflow_test.py

@@ -0,0 +1,361 @@
+"""
+Test underflow protection for cumulative refunds and negative transactions.
+
+This test ensures that when multiple large refunds are processed, the user balance
+doesn't underflow below POSTGRES_INT_MIN, which could cause integer wraparound issues.
+"""
+
+import asyncio
+from uuid import uuid4
+
+import pytest
+from prisma.enums import CreditTransactionType
+from prisma.errors import UniqueViolationError
+from prisma.models import CreditTransaction, User, UserBalance
+
+from backend.data.credit import POSTGRES_INT_MIN, UserCredit
+from backend.util.test import SpinTestServer
+
+
+async def create_test_user(user_id: str) -> None:
+    """Create a test user for underflow tests."""
+    try:
+        await User.prisma().create(
+            data={
+                "id": user_id,
+                "email": f"test-{user_id}@example.com",
+                "name": f"Test User {user_id[:8]}",
+            }
+        )
+    except UniqueViolationError:
+        # User already exists, continue
+        pass
+
+    await UserBalance.prisma().upsert(
+        where={"userId": user_id},
+        data={"create": {"userId": user_id, "balance": 0}, "update": {"balance": 0}},
+    )
+
+
+async def cleanup_test_user(user_id: str) -> None:
+    """Clean up test user and their transactions."""
+    try:
+        await CreditTransaction.prisma().delete_many(where={"userId": user_id})
+        await UserBalance.prisma().delete_many(where={"userId": user_id})
+        await User.prisma().delete_many(where={"id": user_id})
+    except Exception as e:
+        # Log cleanup failures but don't fail the test
+        print(f"Warning: Failed to cleanup test user {user_id}: {e}")
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_debug_underflow_step_by_step(server: SpinTestServer):
+    """Debug underflow behavior step by step."""
+    credit_system = UserCredit()
+    user_id = f"debug-underflow-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        print(f"POSTGRES_INT_MIN: {POSTGRES_INT_MIN}")
+
+        # Test 1: Set up balance close to underflow threshold
+        print("\n=== Test 1: Setting up balance close to underflow threshold ===")
+        # First, manually set balance to a value very close to POSTGRES_INT_MIN
+        # We'll set it to POSTGRES_INT_MIN + 100, then try to subtract 200
+        # This should trigger underflow protection: (POSTGRES_INT_MIN + 100) + (-200) = POSTGRES_INT_MIN - 100
+        initial_balance_target = POSTGRES_INT_MIN + 100
+
+        # Use direct database update to set the balance close to underflow
+        from prisma.models import UserBalance
+
+        await UserBalance.prisma().upsert(
+            where={"userId": user_id},
+            data={
+                "create": {"userId": user_id, "balance": initial_balance_target},
+                "update": {"balance": initial_balance_target},
+            },
+        )
+
+        current_balance = await credit_system.get_credits(user_id)
+        print(f"Set balance to: {current_balance}")
+        assert current_balance == initial_balance_target
+
+        # Test 2: Apply amount that should cause underflow
+        print("\n=== Test 2: Testing underflow protection ===")
+        test_amount = (
+            -200
+        )  # This should cause underflow: (POSTGRES_INT_MIN + 100) + (-200) = POSTGRES_INT_MIN - 100
+        expected_without_protection = current_balance + test_amount
+        print(f"Current balance: {current_balance}")
+        print(f"Test amount: {test_amount}")
+        print(f"Without protection would be: {expected_without_protection}")
+        print(f"Should be clamped to POSTGRES_INT_MIN: {POSTGRES_INT_MIN}")
+
+        # Apply the amount that should trigger underflow protection
+        balance_result, _ = await credit_system._add_transaction(
+            user_id=user_id,
+            amount=test_amount,
+            transaction_type=CreditTransactionType.REFUND,
+            fail_insufficient_credits=False,
+        )
+        print(f"Actual result: {balance_result}")
+
+        # Check if underflow protection worked
+        assert (
+            balance_result == POSTGRES_INT_MIN
+        ), f"Expected underflow protection to clamp balance to {POSTGRES_INT_MIN}, got {balance_result}"
+
+        # Test 3: Edge case - exactly at POSTGRES_INT_MIN
+        print("\n=== Test 3: Testing exact POSTGRES_INT_MIN boundary ===")
+        # Set balance to exactly POSTGRES_INT_MIN
+        await UserBalance.prisma().upsert(
+            where={"userId": user_id},
+            data={
+                "create": {"userId": user_id, "balance": POSTGRES_INT_MIN},
+                "update": {"balance": POSTGRES_INT_MIN},
+            },
+        )
+
+        edge_balance = await credit_system.get_credits(user_id)
+        print(f"Balance set to exactly POSTGRES_INT_MIN: {edge_balance}")
+
+        # Try to subtract 1 - should stay at POSTGRES_INT_MIN
+        edge_result, _ = await credit_system._add_transaction(
+            user_id=user_id,
+            amount=-1,
+            transaction_type=CreditTransactionType.REFUND,
+            fail_insufficient_credits=False,
+        )
+        print(f"After subtracting 1: {edge_result}")
+
+        assert (
+            edge_result == POSTGRES_INT_MIN
+        ), f"Expected balance to remain clamped at {POSTGRES_INT_MIN}, got {edge_result}"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_underflow_protection_large_refunds(server: SpinTestServer):
+    """Test that large cumulative refunds don't cause integer underflow."""
+    credit_system = UserCredit()
+    user_id = f"underflow-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Set up balance close to underflow threshold to test the protection
+        # Set balance to POSTGRES_INT_MIN + 1000, then try to subtract 2000
+        # This should trigger underflow protection
+        from prisma.models import UserBalance
+
+        test_balance = POSTGRES_INT_MIN + 1000
+        await UserBalance.prisma().upsert(
+            where={"userId": user_id},
+            data={
+                "create": {"userId": user_id, "balance": test_balance},
+                "update": {"balance": test_balance},
+            },
+        )
+
+        current_balance = await credit_system.get_credits(user_id)
+        assert current_balance == test_balance
+
+        # Try to deduct amount that would cause underflow: test_balance + (-2000) = POSTGRES_INT_MIN - 1000
+        underflow_amount = -2000
+        expected_without_protection = (
+            current_balance + underflow_amount
+        )  # Should be POSTGRES_INT_MIN - 1000
+
+        # Use _add_transaction directly with amount that would cause underflow
+        final_balance, _ = await credit_system._add_transaction(
+            user_id=user_id,
+            amount=underflow_amount,
+            transaction_type=CreditTransactionType.REFUND,
+            fail_insufficient_credits=False,  # Allow going negative for refunds
+        )
+
+        # Balance should be clamped to POSTGRES_INT_MIN, not the calculated underflow value
+        assert (
+            final_balance == POSTGRES_INT_MIN
+        ), f"Balance should be clamped to {POSTGRES_INT_MIN}, got {final_balance}"
+        assert (
+            final_balance > expected_without_protection
+        ), f"Balance should be greater than underflow result {expected_without_protection}, got {final_balance}"
+
+        # Verify with get_credits too
+        stored_balance = await credit_system.get_credits(user_id)
+        assert (
+            stored_balance == POSTGRES_INT_MIN
+        ), f"Stored balance should be {POSTGRES_INT_MIN}, got {stored_balance}"
+
+        # Verify transaction was created with the underflow-protected balance
+        transactions = await CreditTransaction.prisma().find_many(
+            where={"userId": user_id, "type": CreditTransactionType.REFUND},
+            order={"createdAt": "desc"},
+        )
+        assert len(transactions) > 0, "Refund transaction should be created"
+        assert (
+            transactions[0].runningBalance == POSTGRES_INT_MIN
+        ), f"Transaction should show clamped balance {POSTGRES_INT_MIN}, got {transactions[0].runningBalance}"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_multiple_large_refunds_cumulative_underflow(server: SpinTestServer):
+    """Test that multiple large refunds applied sequentially don't cause underflow."""
+    credit_system = UserCredit()
+    user_id = f"cumulative-underflow-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Set up balance close to underflow threshold
+        from prisma.models import UserBalance
+
+        initial_balance = POSTGRES_INT_MIN + 500  # Close to minimum but with some room
+        await UserBalance.prisma().upsert(
+            where={"userId": user_id},
+            data={
+                "create": {"userId": user_id, "balance": initial_balance},
+                "update": {"balance": initial_balance},
+            },
+        )
+
+        # Apply multiple refunds that would cumulatively underflow
+        refund_amount = -300  # Each refund that would cause underflow when cumulative
+
+        # First refund: (POSTGRES_INT_MIN + 500) + (-300) = POSTGRES_INT_MIN + 200 (still above minimum)
+        balance_1, _ = await credit_system._add_transaction(
+            user_id=user_id,
+            amount=refund_amount,
+            transaction_type=CreditTransactionType.REFUND,
+            fail_insufficient_credits=False,
+        )
+
+        # Should be above minimum for first refund
+        expected_balance_1 = (
+            initial_balance + refund_amount
+        )  # Should be POSTGRES_INT_MIN + 200
+        assert (
+            balance_1 == expected_balance_1
+        ), f"First refund should result in {expected_balance_1}, got {balance_1}"
+        assert (
+            balance_1 >= POSTGRES_INT_MIN
+        ), f"First refund should not go below {POSTGRES_INT_MIN}, got {balance_1}"
+
+        # Second refund: (POSTGRES_INT_MIN + 200) + (-300) = POSTGRES_INT_MIN - 100 (would underflow)
+        balance_2, _ = await credit_system._add_transaction(
+            user_id=user_id,
+            amount=refund_amount,
+            transaction_type=CreditTransactionType.REFUND,
+            fail_insufficient_credits=False,
+        )
+
+        # Should be clamped to minimum due to underflow protection
+        assert (
+            balance_2 == POSTGRES_INT_MIN
+        ), f"Second refund should be clamped to {POSTGRES_INT_MIN}, got {balance_2}"
+
+        # Third refund: Should stay at minimum
+        balance_3, _ = await credit_system._add_transaction(
+            user_id=user_id,
+            amount=refund_amount,
+            transaction_type=CreditTransactionType.REFUND,
+            fail_insufficient_credits=False,
+        )
+
+        # Should still be at minimum
+        assert (
+            balance_3 == POSTGRES_INT_MIN
+        ), f"Third refund should stay at {POSTGRES_INT_MIN}, got {balance_3}"
+
+        # Final balance check
+        final_balance = await credit_system.get_credits(user_id)
+        assert (
+            final_balance == POSTGRES_INT_MIN
+        ), f"Final balance should be {POSTGRES_INT_MIN}, got {final_balance}"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_concurrent_large_refunds_no_underflow(server: SpinTestServer):
+    """Test that concurrent large refunds don't cause race condition underflow."""
+    credit_system = UserCredit()
+    user_id = f"concurrent-underflow-test-{uuid4()}"
+    await create_test_user(user_id)
+
+    try:
+        # Set up balance close to underflow threshold
+        from prisma.models import UserBalance
+
+        initial_balance = POSTGRES_INT_MIN + 1000  # Close to minimum
+        await UserBalance.prisma().upsert(
+            where={"userId": user_id},
+            data={
+                "create": {"userId": user_id, "balance": initial_balance},
+                "update": {"balance": initial_balance},
+            },
+        )
+
+        async def large_refund(amount: int, label: str):
+            try:
+                return await credit_system._add_transaction(
+                    user_id=user_id,
+                    amount=-amount,
+                    transaction_type=CreditTransactionType.REFUND,
+                    fail_insufficient_credits=False,
+                )
+            except Exception as e:
+                return f"FAILED-{label}: {e}"
+
+        # Run concurrent refunds that would cause underflow if not protected
+        # Each refund of 500 would cause underflow: initial_balance + (-500) could go below POSTGRES_INT_MIN
+        refund_amount = 500
+        results = await asyncio.gather(
+            large_refund(refund_amount, "A"),
+            large_refund(refund_amount, "B"),
+            large_refund(refund_amount, "C"),
+            return_exceptions=True,
+        )
+
+        # Check all results are valid and no underflow occurred
+        valid_results = []
+        for i, result in enumerate(results):
+            if isinstance(result, tuple):
+                balance, _ = result
+                assert (
+                    balance >= POSTGRES_INT_MIN
+                ), f"Result {i} balance {balance} underflowed below {POSTGRES_INT_MIN}"
+                valid_results.append(balance)
+            elif isinstance(result, str) and "FAILED" in result:
+                # Some operations might fail due to validation, that's okay
+                pass
+            else:
+                # Unexpected exception
+                assert not isinstance(
+                    result, Exception
+                ), f"Unexpected exception in result {i}: {result}"
+
+        # At least one operation should succeed
+        assert (
+            len(valid_results) > 0
+        ), f"At least one refund should succeed, got results: {results}"
+
+        # All successful results should be >= POSTGRES_INT_MIN
+        for balance in valid_results:
+            assert (
+                balance >= POSTGRES_INT_MIN
+            ), f"Balance {balance} should not be below {POSTGRES_INT_MIN}"
+
+        # Final balance should be valid and at or above POSTGRES_INT_MIN
+        final_balance = await credit_system.get_credits(user_id)
+        assert (
+            final_balance >= POSTGRES_INT_MIN
+        ), f"Final balance {final_balance} should not underflow below {POSTGRES_INT_MIN}"
+
+    finally:
+        await cleanup_test_user(user_id)

autogpt_platform/backend/backend/data/credit_user_balance_migration_test.py

@@ -0,0 +1,217 @@
+"""
+Integration test to verify complete migration from User.balance to UserBalance table.
+
+This test ensures that:
+1. No User.balance queries exist in the system
+2. All balance operations go through UserBalance table
+3. User and UserBalance stay synchronized properly
+"""
+
+import asyncio
+from datetime import datetime
+
+import pytest
+from prisma.enums import CreditTransactionType
+from prisma.errors import UniqueViolationError
+from prisma.models import CreditTransaction, User, UserBalance
+
+from backend.data.credit import UsageTransactionMetadata, UserCredit
+from backend.util.json import SafeJson
+from backend.util.test import SpinTestServer
+
+
+async def create_test_user(user_id: str) -> None:
+    """Create a test user for migration tests."""
+    try:
+        await User.prisma().create(
+            data={
+                "id": user_id,
+                "email": f"test-{user_id}@example.com",
+                "name": f"Test User {user_id[:8]}",
+            }
+        )
+    except UniqueViolationError:
+        # User already exists, continue
+        pass
+
+
+async def cleanup_test_user(user_id: str) -> None:
+    """Clean up test user and their data."""
+    try:
+        await CreditTransaction.prisma().delete_many(where={"userId": user_id})
+        await UserBalance.prisma().delete_many(where={"userId": user_id})
+        await User.prisma().delete_many(where={"id": user_id})
+    except Exception as e:
+        # Log cleanup failures but don't fail the test
+        print(f"Warning: Failed to cleanup test user {user_id}: {e}")
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_user_balance_migration_complete(server: SpinTestServer):
+    """Test that User table balance is never used and UserBalance is source of truth."""
+    credit_system = UserCredit()
+    user_id = f"migration-test-{datetime.now().timestamp()}"
+    await create_test_user(user_id)
+
+    try:
+        # 1. Verify User table does NOT have balance set initially
+        user = await User.prisma().find_unique(where={"id": user_id})
+        assert user is not None
+        # User.balance should not exist or should be None/0 if it exists
+        user_balance_attr = getattr(user, "balance", None)
+        if user_balance_attr is not None:
+            assert (
+                user_balance_attr == 0 or user_balance_attr is None
+            ), f"User.balance should be 0 or None, got {user_balance_attr}"
+
+        # 2. Perform various credit operations using internal method (bypasses Stripe)
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=1000,
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "migration_test"}),
+        )
+        balance1 = await credit_system.get_credits(user_id)
+        assert balance1 == 1000
+
+        await credit_system.spend_credits(
+            user_id,
+            300,
+            UsageTransactionMetadata(
+                graph_exec_id="test", reason="Migration test spend"
+            ),
+        )
+        balance2 = await credit_system.get_credits(user_id)
+        assert balance2 == 700
+
+        # 3. Verify UserBalance table has correct values
+        user_balance = await UserBalance.prisma().find_unique(where={"userId": user_id})
+        assert user_balance is not None
+        assert (
+            user_balance.balance == 700
+        ), f"UserBalance should be 700, got {user_balance.balance}"
+
+        # 4. CRITICAL: Verify User.balance is NEVER updated during operations
+        user_after = await User.prisma().find_unique(where={"id": user_id})
+        assert user_after is not None
+        user_balance_after = getattr(user_after, "balance", None)
+        if user_balance_after is not None:
+            # If User.balance exists, it should still be 0 (never updated)
+            assert (
+                user_balance_after == 0 or user_balance_after is None
+            ), f"User.balance should remain 0/None after operations, got {user_balance_after}. This indicates User.balance is still being used!"
+
+        # 5. Verify get_credits always returns UserBalance value, not User.balance
+        final_balance = await credit_system.get_credits(user_id)
+        assert (
+            final_balance == user_balance.balance
+        ), f"get_credits should return UserBalance value {user_balance.balance}, got {final_balance}"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_detect_stale_user_balance_queries(server: SpinTestServer):
+    """Test to detect if any operations are still using User.balance instead of UserBalance."""
+    credit_system = UserCredit()
+    user_id = f"stale-query-test-{datetime.now().timestamp()}"
+    await create_test_user(user_id)
+
+    try:
+        # Create UserBalance with specific value
+        await UserBalance.prisma().create(
+            data={"userId": user_id, "balance": 5000}  # $50
+        )
+
+        # Verify that get_credits returns UserBalance value (5000), not any stale User.balance value
+        balance = await credit_system.get_credits(user_id)
+        assert (
+            balance == 5000
+        ), f"Expected get_credits to return 5000 from UserBalance, got {balance}"
+
+        # Verify all operations use UserBalance using internal method (bypasses Stripe)
+        await credit_system._add_transaction(
+            user_id=user_id,
+            amount=1000,
+            transaction_type=CreditTransactionType.TOP_UP,
+            metadata=SafeJson({"test": "final_verification"}),
+        )
+        final_balance = await credit_system.get_credits(user_id)
+        assert final_balance == 6000, f"Expected 6000, got {final_balance}"
+
+        # Verify UserBalance table has the correct value
+        user_balance = await UserBalance.prisma().find_unique(where={"userId": user_id})
+        assert user_balance is not None
+        assert (
+            user_balance.balance == 6000
+        ), f"UserBalance should be 6000, got {user_balance.balance}"
+
+    finally:
+        await cleanup_test_user(user_id)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_concurrent_operations_use_userbalance_only(server: SpinTestServer):
+    """Test that concurrent operations all use UserBalance locking, not User.balance."""
+    credit_system = UserCredit()
+    user_id = f"concurrent-userbalance-test-{datetime.now().timestamp()}"
+    await create_test_user(user_id)
+
+    try:
+        # Set initial balance in UserBalance
+        await UserBalance.prisma().create(data={"userId": user_id, "balance": 1000})
+
+        # Run concurrent operations to ensure they all use UserBalance atomic operations
+        async def concurrent_spend(amount: int, label: str):
+            try:
+                await credit_system.spend_credits(
+                    user_id,
+                    amount,
+                    UsageTransactionMetadata(
+                        graph_exec_id=f"concurrent-{label}",
+                        reason=f"Concurrent test {label}",
+                    ),
+                )
+                return f"{label}-SUCCESS"
+            except Exception as e:
+                return f"{label}-FAILED: {e}"
+
+        # Run concurrent operations
+        results = await asyncio.gather(
+            concurrent_spend(100, "A"),
+            concurrent_spend(200, "B"),
+            concurrent_spend(300, "C"),
+            return_exceptions=True,
+        )
+
+        # All should succeed (1000 >= 100+200+300)
+        successful = [r for r in results if "SUCCESS" in str(r)]
+        assert len(successful) == 3, f"All operations should succeed, got {results}"
+
+        # Final balance should be 1000 - 600 = 400
+        final_balance = await credit_system.get_credits(user_id)
+        assert final_balance == 400, f"Expected final balance 400, got {final_balance}"
+
+        # Verify UserBalance has correct value
+        user_balance = await UserBalance.prisma().find_unique(where={"userId": user_id})
+        assert user_balance is not None
+        assert (
+            user_balance.balance == 400
+        ), f"UserBalance should be 400, got {user_balance.balance}"
+
+        # Critical: If User.balance exists and was used, it might have wrong value
+        try:
+            user = await User.prisma().find_unique(where={"id": user_id})
+            user_balance_attr = getattr(user, "balance", None)
+            if user_balance_attr is not None:
+                # If User.balance exists, it should NOT be used for operations
+                # The fact that our final balance is correct from UserBalance proves the system is working
+                print(
+                    f"✅ User.balance exists ({user_balance_attr}) but UserBalance ({user_balance.balance}) is being used correctly"
+                )
+        except Exception:
+            print("✅ User.balance column doesn't exist - migration is complete")
+
+    finally:
+        await cleanup_test_user(user_id)

autogpt_platform/backend/backend/data/db.py

@@ -98,42 +98,6 @@ async def transaction(timeout: int = TRANSACTION_TIMEOUT):
         yield tx
 
 
-@asynccontextmanager
-async def locked_transaction(key: str, timeout: int = TRANSACTION_TIMEOUT):
-    """
-    Create a transaction and take a per-key advisory *transaction* lock.
-
-    - Uses a 64-bit lock id via hashtextextended(key, 0) to avoid 32-bit collisions.
-    - Bound by lock_timeout and statement_timeout so it won't block indefinitely.
-    - Lock is held for the duration of the transaction and auto-released on commit/rollback.
-
-    Args:
-        key: String lock key (e.g., "usr_trx_<uuid>").
-        timeout: Transaction/lock/statement timeout in milliseconds.
-    """
-    async with transaction(timeout=timeout) as tx:
-        # Ensure we don't wait longer than desired
-        # Note: SET LOCAL doesn't support parameterized queries, must use string interpolation
-        await tx.execute_raw(f"SET LOCAL statement_timeout = '{int(timeout)}ms'")  # type: ignore[arg-type]
-        await tx.execute_raw(f"SET LOCAL lock_timeout = '{int(timeout)}ms'")  # type: ignore[arg-type]
-
-        # Block until acquired or lock_timeout hits
-        try:
-            await tx.execute_raw(
-                "SELECT pg_advisory_xact_lock(hashtextextended($1, 0))",
-                key,
-            )
-        except Exception as e:
-            # Normalize PG's lock timeout error to TimeoutError for callers
-            if "lock timeout" in str(e).lower():
-                raise TimeoutError(
-                    f"Could not acquire lock for key={key!r} within {timeout}ms"
-                ) from e
-            raise
-
-        yield tx
-
-
 def get_database_schema() -> str:
     """Extract database schema from DATABASE_URL."""
     parsed_url = urlparse(DATABASE_URL)

autogpt_platform/backend/backend/data/dynamic_fields.py

@@ -0,0 +1,284 @@
+"""
+Utilities for handling dynamic field names with special delimiters.
+
+Dynamic fields allow graphs to connect complex data structures using special delimiters:
+- _#_ for dictionary keys (e.g., "values_#_name" → values["name"])
+- _$_ for list indices (e.g., "items_$_0" → items[0])
+- _@_ for object attributes (e.g., "obj_@_attr" → obj.attr)
+"""
+
+from typing import Any
+
+from backend.util.mock import MockObject
+
+# Dynamic field delimiters
+LIST_SPLIT = "_$_"
+DICT_SPLIT = "_#_"
+OBJC_SPLIT = "_@_"
+
+DYNAMIC_DELIMITERS = (LIST_SPLIT, DICT_SPLIT, OBJC_SPLIT)
+
+
+def extract_base_field_name(field_name: str) -> str:
+    """
+    Extract the base field name from a dynamic field name by removing all dynamic suffixes.
+
+    Examples:
+        extract_base_field_name("values_#_name") → "values"
+        extract_base_field_name("items_$_0") → "items"
+        extract_base_field_name("obj_@_attr") → "obj"
+        extract_base_field_name("regular_field") → "regular_field"
+
+    Args:
+        field_name: The field name that may contain dynamic delimiters
+
+    Returns:
+        The base field name without any dynamic suffixes
+    """
+    base_name = field_name
+    for delimiter in DYNAMIC_DELIMITERS:
+        if delimiter in base_name:
+            base_name = base_name.split(delimiter)[0]
+    return base_name
+
+
+def is_dynamic_field(field_name: str) -> bool:
+    """
+    Check if a field name contains dynamic delimiters.
+
+    Args:
+        field_name: The field name to check
+
+    Returns:
+        True if the field contains any dynamic delimiters, False otherwise
+    """
+    return any(delimiter in field_name for delimiter in DYNAMIC_DELIMITERS)
+
+
+def get_dynamic_field_description(field_name: str) -> str:
+    """
+    Generate a description for a dynamic field based on its structure.
+
+    Args:
+        field_name: The full dynamic field name (e.g., "values_#_name")
+
+    Returns:
+        A descriptive string explaining what this dynamic field represents
+    """
+    base_name = extract_base_field_name(field_name)
+
+    if DICT_SPLIT in field_name:
+        # Extract the key part after _#_
+        parts = field_name.split(DICT_SPLIT)
+        if len(parts) > 1:
+            key = parts[1].split("_")[0] if "_" in parts[1] else parts[1]
+            return f"Dictionary field '{key}' for base field '{base_name}' ({base_name}['{key}'])"
+    elif LIST_SPLIT in field_name:
+        # Extract the index part after _$_
+        parts = field_name.split(LIST_SPLIT)
+        if len(parts) > 1:
+            index = parts[1].split("_")[0] if "_" in parts[1] else parts[1]
+            return (
+                f"List item {index} for base field '{base_name}' ({base_name}[{index}])"
+            )
+    elif OBJC_SPLIT in field_name:
+        # Extract the attribute part after _@_
+        parts = field_name.split(OBJC_SPLIT)
+        if len(parts) > 1:
+            # Get the full attribute name (everything after _@_)
+            attr = parts[1]
+            return f"Object attribute '{attr}' for base field '{base_name}' ({base_name}.{attr})"
+
+    return f"Value for {field_name}"
+
+
+# --------------------------------------------------------------------------- #
+#  Dynamic field parsing and merging utilities
+# --------------------------------------------------------------------------- #
+
+
+def _next_delim(s: str) -> tuple[str | None, int]:
+    """
+    Return the *earliest* delimiter appearing in `s` and its index.
+
+    If none present → (None, -1).
+    """
+    first: str | None = None
+    pos = len(s)  # sentinel: larger than any real index
+    for d in DYNAMIC_DELIMITERS:
+        i = s.find(d)
+        if 0 <= i < pos:
+            first, pos = d, i
+    return first, (pos if first else -1)
+
+
+def _tokenise(path: str) -> list[tuple[str, str]] | None:
+    """
+    Convert the raw path string (starting with a delimiter) into
+    [ (delimiter, identifier), … ] or None if the syntax is malformed.
+    """
+    tokens: list[tuple[str, str]] = []
+    while path:
+        # 1. Which delimiter starts this chunk?
+        delim = next((d for d in DYNAMIC_DELIMITERS if path.startswith(d)), None)
+        if delim is None:
+            return None  # invalid syntax
+
+        # 2. Slice off the delimiter, then up to the next delimiter (or EOS)
+        path = path[len(delim) :]
+        nxt_delim, pos = _next_delim(path)
+        token, path = (
+            path[: pos if pos != -1 else len(path)],
+            path[pos if pos != -1 else len(path) :],
+        )
+        if token == "":
+            return None  # empty identifier is invalid
+        tokens.append((delim, token))
+    return tokens
+
+
+def parse_execution_output(output: tuple[str, Any], name: str) -> Any:
+    """
+    Retrieve a nested value out of `output` using the flattened *name*.
+
+    On any failure (wrong name, wrong type, out-of-range, bad path)
+    returns **None**.
+
+    Args:
+        output: Tuple of (base_name, data) representing a block output entry
+        name: The flattened field name to extract from the output data
+
+    Returns:
+        The value at the specified path, or None if not found/invalid
+    """
+    base_name, data = output
+
+    # Exact match → whole object
+    if name == base_name:
+        return data
+
+    # Must start with the expected name
+    if not name.startswith(base_name):
+        return None
+    path = name[len(base_name) :]
+    if not path:
+        return None  # nothing left to parse
+
+    tokens = _tokenise(path)
+    if tokens is None:
+        return None
+
+    cur: Any = data
+    for delim, ident in tokens:
+        if delim == LIST_SPLIT:
+            # list[index]
+            try:
+                idx = int(ident)
+            except ValueError:
+                return None
+            if not isinstance(cur, list) or idx >= len(cur):
+                return None
+            cur = cur[idx]
+
+        elif delim == DICT_SPLIT:
+            if not isinstance(cur, dict) or ident not in cur:
+                return None
+            cur = cur[ident]
+
+        elif delim == OBJC_SPLIT:
+            if not hasattr(cur, ident):
+                return None
+            cur = getattr(cur, ident)
+
+        else:
+            return None  # unreachable
+
+    return cur
+
+
+def _assign(container: Any, tokens: list[tuple[str, str]], value: Any) -> Any:
+    """
+    Recursive helper that *returns* the (possibly new) container with
+    `value` assigned along the remaining `tokens` path.
+    """
+    if not tokens:
+        return value  # leaf reached
+
+    delim, ident = tokens[0]
+    rest = tokens[1:]
+
+    # ---------- list ----------
+    if delim == LIST_SPLIT:
+        try:
+            idx = int(ident)
+        except ValueError:
+            raise ValueError("index must be an integer")
+
+        if container is None:
+            container = []
+        elif not isinstance(container, list):
+            container = list(container) if hasattr(container, "__iter__") else []
+
+        while len(container) <= idx:
+            container.append(None)
+        container[idx] = _assign(container[idx], rest, value)
+        return container
+
+    # ---------- dict ----------
+    if delim == DICT_SPLIT:
+        if container is None:
+            container = {}
+        elif not isinstance(container, dict):
+            container = dict(container) if hasattr(container, "items") else {}
+        container[ident] = _assign(container.get(ident), rest, value)
+        return container
+
+    # ---------- object ----------
+    if delim == OBJC_SPLIT:
+        if container is None:
+            container = MockObject()
+        elif not hasattr(container, "__dict__"):
+            # If it's not an object, create a new one
+            container = MockObject()
+        setattr(
+            container,
+            ident,
+            _assign(getattr(container, ident, None), rest, value),
+        )
+        return container
+
+    return value  # unreachable
+
+
+def merge_execution_input(data: dict[str, Any]) -> dict[str, Any]:
+    """
+    Reconstruct nested objects from a *flattened* dict of key → value.
+
+    Raises ValueError on syntactically invalid list indices.
+
+    Args:
+        data: Dictionary with potentially flattened dynamic field keys
+
+    Returns:
+        Dictionary with nested objects reconstructed from flattened keys
+    """
+    merged: dict[str, Any] = {}
+
+    for key, value in data.items():
+        # Split off the base name (before the first delimiter, if any)
+        delim, pos = _next_delim(key)
+        if delim is None:
+            merged[key] = value
+            continue
+
+        base, path = key[:pos], key[pos:]
+        tokens = _tokenise(path)
+        if tokens is None:
+            # Invalid key; treat as scalar under the raw name
+            merged[key] = value
+            continue
+
+        merged[base] = _assign(merged.get(base), tokens, value)
+
+    data.update(merged)
+    return data

autogpt_platform/backend/backend/data/execution.py

@@ -38,8 +38,8 @@ from prisma.types import (
 from pydantic import BaseModel, ConfigDict, JsonValue, ValidationError
 from pydantic.fields import Field
 
-from backend.server.v2.store.exceptions import DatabaseError
 from backend.util import type as type_utils
+from backend.util.exceptions import DatabaseError
 from backend.util.json import SafeJson
 from backend.util.models import Pagination
 from backend.util.retry import func_retry
@@ -478,6 +478,48 @@ async def get_graph_executions(
     return [GraphExecutionMeta.from_db(execution) for execution in executions]
 
 
+async def get_graph_executions_count(
+    user_id: Optional[str] = None,
+    graph_id: Optional[str] = None,
+    statuses: Optional[list[ExecutionStatus]] = None,
+    created_time_gte: Optional[datetime] = None,
+    created_time_lte: Optional[datetime] = None,
+) -> int:
+    """
+    Get count of graph executions with optional filters.
+
+    Args:
+        user_id: Optional user ID to filter by
+        graph_id: Optional graph ID to filter by
+        statuses: Optional list of execution statuses to filter by
+        created_time_gte: Optional minimum creation time
+        created_time_lte: Optional maximum creation time
+
+    Returns:
+        Count of matching graph executions
+    """
+    where_filter: AgentGraphExecutionWhereInput = {
+        "isDeleted": False,
+    }
+
+    if user_id:
+        where_filter["userId"] = user_id
+
+    if graph_id:
+        where_filter["agentGraphId"] = graph_id
+
+    if created_time_gte or created_time_lte:
+        where_filter["createdAt"] = {
+            "gte": created_time_gte or datetime.min.replace(tzinfo=timezone.utc),
+            "lte": created_time_lte or datetime.max.replace(tzinfo=timezone.utc),
+        }
+    if statuses:
+        where_filter["OR"] = [{"executionStatus": status} for status in statuses]
+
+    count = await AgentGraphExecution.prisma().count(where=where_filter)
+    return count
+
+
 class GraphExecutionsPaginated(BaseModel):
     """Response schema for paginated graph executions."""
 

autogpt_platform/backend/backend/data/generate_data.py

@@ -7,7 +7,7 @@ from prisma.enums import AgentExecutionStatus
 from backend.data.execution import get_graph_executions
 from backend.data.graph import get_graph_metadata
 from backend.data.model import UserExecutionSummaryStats
-from backend.server.v2.store.exceptions import DatabaseError
+from backend.util.exceptions import DatabaseError
 from backend.util.logging import TruncatedLogger
 
 logger = TruncatedLogger(logging.getLogger(__name__), prefix="[SummaryData]")

autogpt_platform/backend/backend/data/graph.py

@@ -20,6 +20,8 @@ from backend.blocks.agent import AgentExecutorBlock
 from backend.blocks.io import AgentInputBlock, AgentOutputBlock
 from backend.blocks.llm import LlmModel
 from backend.data.db import prisma as db
+from backend.data.dynamic_fields import extract_base_field_name
+from backend.data.includes import MAX_GRAPH_VERSIONS_FETCH
 from backend.data.model import (
     CredentialsField,
     CredentialsFieldInfo,
@@ -31,7 +33,15 @@ from backend.util import type as type_utils
 from backend.util.json import SafeJson
 from backend.util.models import Pagination
 
-from .block import Block, BlockInput, BlockSchema, BlockType, get_block, get_blocks
+from .block import (
+    Block,
+    BlockInput,
+    BlockSchema,
+    BlockType,
+    EmptySchema,
+    get_block,
+    get_blocks,
+)
 from .db import BaseDbModel, query_raw_with_schema, transaction
 from .includes import AGENT_GRAPH_INCLUDE, AGENT_NODE_INCLUDE
 
@@ -72,12 +82,15 @@ class Node(BaseDbModel):
     output_links: list[Link] = []
 
     @property
-    def block(self) -> Block[BlockSchema, BlockSchema]:
+    def block(self) -> "Block[BlockSchema, BlockSchema] | _UnknownBlockBase":
+        """Get the block for this node. Returns UnknownBlock if block is deleted/missing."""
         block = get_block(self.block_id)
         if not block:
-            raise ValueError(
-                f"Block #{self.block_id} does not exist -> Node #{self.id} is invalid"
+            # Log warning but don't raise exception - return a placeholder block for deleted blocks
+            logger.warning(
+                f"Block #{self.block_id} does not exist for Node #{self.id} (deleted/missing block), using UnknownBlock"
             )
+            return _UnknownBlockBase(self.block_id)
         return block
 
 
@@ -116,17 +129,20 @@ class NodeModel(Node):
         Returns a copy of the node model, stripped of any non-transferable properties
         """
         stripped_node = self.model_copy(deep=True)
-        # Remove credentials from node input
+
+        # Remove credentials and other (possible) secrets from node input
         if stripped_node.input_default:
             stripped_node.input_default = NodeModel._filter_secrets_from_node_input(
                 stripped_node.input_default, self.block.input_schema.jsonschema()
             )
 
+        # Remove default secret value from secret input nodes
         if (
             stripped_node.block.block_type == BlockType.INPUT
+            and stripped_node.input_default.get("secret", False) is True
             and "value" in stripped_node.input_default
         ):
-            stripped_node.input_default["value"] = ""
+            del stripped_node.input_default["value"]
 
         # Remove webhook info
         stripped_node.webhook_id = None
@@ -143,8 +159,10 @@ class NodeModel(Node):
         result = {}
         for key, value in input_data.items():
             field_schema: dict | None = field_schemas.get(key)
-            if (field_schema and field_schema.get("secret", False)) or any(
-                sensitive_key in key.lower() for sensitive_key in sensitive_keys
+            if (field_schema and field_schema.get("secret", False)) or (
+                any(sensitive_key in key.lower() for sensitive_key in sensitive_keys)
+                # Prevent removing `secret` flag on input nodes
+                and type(value) is not bool
             ):
                 # This is a secret value -> filter this key-value pair out
                 continue
@@ -729,7 +747,7 @@ def _is_tool_pin(name: str) -> bool:
 
 
 def _sanitize_pin_name(name: str) -> str:
-    sanitized_name = name.split("_#_")[0].split("_@_")[0].split("_$_")[0]
+    sanitized_name = extract_base_field_name(name)
     if _is_tool_pin(sanitized_name):
         return "tools"
     return sanitized_name
@@ -1059,11 +1077,14 @@ async def set_graph_active_version(graph_id: str, version: int, user_id: str) ->
     )
 
 
-async def get_graph_all_versions(graph_id: str, user_id: str) -> list[GraphModel]:
+async def get_graph_all_versions(
+    graph_id: str, user_id: str, limit: int = MAX_GRAPH_VERSIONS_FETCH
+) -> list[GraphModel]:
     graph_versions = await AgentGraph.prisma().find_many(
         where={"id": graph_id, "userId": user_id},
         order={"version": "desc"},
         include=AGENT_GRAPH_INCLUDE,
+        take=limit,
     )
 
     if not graph_versions:
@@ -1312,3 +1333,34 @@ async def migrate_llm_models(migrate_to: LlmModel):
             id,
             path,
         )
+
+
+# Simple placeholder class for deleted/missing blocks
+class _UnknownBlockBase(Block):
+    """
+    Placeholder for deleted/missing blocks that inherits from Block
+    but uses a name that doesn't end with 'Block' to avoid auto-discovery.
+    """
+
+    def __init__(self, block_id: str = "00000000-0000-0000-0000-000000000000"):
+        # Initialize with minimal valid Block parameters
+        super().__init__(
+            id=block_id,
+            description=f"Unknown or deleted block (original ID: {block_id})",
+            disabled=True,
+            input_schema=EmptySchema,
+            output_schema=EmptySchema,
+            categories=set(),
+            contributors=[],
+            static_output=False,
+            block_type=BlockType.STANDARD,
+            webhook_config=None,
+        )
+
+    @property
+    def name(self):
+        return "UnknownBlock"
+
+    async def run(self, input_data, **kwargs):
+        """Always yield an error for missing blocks."""
+        yield "error", f"Block {self.id} no longer exists"

autogpt_platform/backend/backend/data/graph_test.py

@@ -201,25 +201,56 @@ async def test_get_input_schema(server: SpinTestServer, snapshot: Snapshot):
 @pytest.mark.asyncio(loop_scope="session")
 async def test_clean_graph(server: SpinTestServer):
     """
-    Test the clean_graph function that:
-    1. Clears input block values
-    2. Removes credentials from nodes
+    Test the stripped_for_export function that:
+    1. Removes sensitive/secret fields from node inputs
+    2. Removes webhook information
+    3. Preserves non-sensitive data including input block values
     """
-    # Create a graph with input blocks and credentials
+    # Create a graph with input blocks containing both sensitive and normal data
     graph = Graph(
         id="test_clean_graph",
         name="Test Clean Graph",
         description="Test graph cleaning",
         nodes=[
             Node(
-                id="input_node",
                 block_id=AgentInputBlock().id,
                 input_default={
+                    "_test_id": "input_node",
                     "name": "test_input",
-                    "value": "test value",
+                    "value": "test value",  # This should be preserved
                     "description": "Test input description",
                 },
             ),
+            Node(
+                block_id=AgentInputBlock().id,
+                input_default={
+                    "_test_id": "input_node_secret",
+                    "name": "secret_input",
+                    "value": "another value",
+                    "secret": True,  # This makes the input secret
+                },
+            ),
+            Node(
+                block_id=StoreValueBlock().id,
+                input_default={
+                    "_test_id": "node_with_secrets",
+                    "input": "normal_value",
+                    "control_test_input": "should be preserved",
+                    "api_key": "secret_api_key_123",  # Should be filtered
+                    "password": "secret_password_456",  # Should be filtered
+                    "token": "secret_token_789",  # Should be filtered
+                    "credentials": {  # Should be filtered
+                        "id": "fake-github-credentials-id",
+                        "provider": "github",
+                        "type": "api_key",
+                    },
+                    "anthropic_credentials": {  # Should be filtered
+                        "id": "fake-anthropic-credentials-id",
+                        "provider": "anthropic",
+                        "type": "api_key",
+                    },
+                },
+            ),
         ],
         links=[],
     )
@@ -231,15 +262,54 @@ async def test_clean_graph(server: SpinTestServer):
     )
 
     # Clean the graph
-    created_graph = await server.agent_server.test_get_graph(
+    cleaned_graph = await server.agent_server.test_get_graph(
         created_graph.id, created_graph.version, DEFAULT_USER_ID, for_export=True
     )
 
-    # # Verify input block value is cleared
+    # Verify sensitive fields are removed but normal fields are preserved
     input_node = next(
-        n for n in created_graph.nodes if n.block_id == AgentInputBlock().id
+        n for n in cleaned_graph.nodes if n.input_default["_test_id"] == "input_node"
     )
-    assert input_node.input_default["value"] == ""
+
+    # Non-sensitive fields should be preserved
+    assert input_node.input_default["name"] == "test_input"
+    assert input_node.input_default["value"] == "test value"  # Should be preserved now
+    assert input_node.input_default["description"] == "Test input description"
+
+    # Sensitive fields should be filtered out
+    assert "api_key" not in input_node.input_default
+    assert "password" not in input_node.input_default
+
+    # Verify secret input node preserves non-sensitive fields but removes secret value
+    secret_node = next(
+        n
+        for n in cleaned_graph.nodes
+        if n.input_default["_test_id"] == "input_node_secret"
+    )
+    assert secret_node.input_default["name"] == "secret_input"
+    assert "value" not in secret_node.input_default  # Secret default should be removed
+    assert secret_node.input_default["secret"] is True
+
+    # Verify sensitive fields are filtered from nodes with secrets
+    secrets_node = next(
+        n
+        for n in cleaned_graph.nodes
+        if n.input_default["_test_id"] == "node_with_secrets"
+    )
+    # Normal fields should be preserved
+    assert secrets_node.input_default["input"] == "normal_value"
+    assert secrets_node.input_default["control_test_input"] == "should be preserved"
+    # Sensitive fields should be filtered out
+    assert "api_key" not in secrets_node.input_default
+    assert "password" not in secrets_node.input_default
+    assert "token" not in secrets_node.input_default
+    assert "credentials" not in secrets_node.input_default
+    assert "anthropic_credentials" not in secrets_node.input_default
+
+    # Verify webhook info is removed (if any nodes had it)
+    for node in cleaned_graph.nodes:
+        assert node.webhook_id is None
+        assert node.webhook is None
 
 
 @pytest.mark.asyncio(loop_scope="session")

autogpt_platform/backend/backend/data/includes.py

@@ -14,6 +14,7 @@ AGENT_GRAPH_INCLUDE: prisma.types.AgentGraphInclude = {
     "Nodes": {"include": AGENT_NODE_INCLUDE}
 }
 
+
 EXECUTION_RESULT_ORDER: list[prisma.types.AgentNodeExecutionOrderByInput] = [
     {"queuedTime": "desc"},
     # Fallback: Incomplete execs has no queuedTime.
@@ -28,6 +29,13 @@ EXECUTION_RESULT_INCLUDE: prisma.types.AgentNodeExecutionInclude = {
 }
 
 MAX_NODE_EXECUTIONS_FETCH = 1000
+MAX_LIBRARY_AGENT_EXECUTIONS_FETCH = 10
+
+# Default limits for potentially large result sets
+MAX_CREDIT_REFUND_REQUESTS_FETCH = 100
+MAX_INTEGRATION_WEBHOOKS_FETCH = 100
+MAX_USER_API_KEYS_FETCH = 500
+MAX_GRAPH_VERSIONS_FETCH = 50
 
 GRAPH_EXECUTION_INCLUDE_WITH_NODES: prisma.types.AgentGraphExecutionInclude = {
     "NodeExecutions": {
@@ -71,13 +79,56 @@ INTEGRATION_WEBHOOK_INCLUDE: prisma.types.IntegrationWebhookInclude = {
 }
 
 
-def library_agent_include(user_id: str) -> prisma.types.LibraryAgentInclude:
-    return {
-        "AgentGraph": {
-            "include": {
-                **AGENT_GRAPH_INCLUDE,
-                "Executions": {"where": {"userId": user_id}},
-            }
-        },
-        "Creator": True,
+def library_agent_include(
+    user_id: str,
+    include_nodes: bool = True,
+    include_executions: bool = True,
+    execution_limit: int = MAX_LIBRARY_AGENT_EXECUTIONS_FETCH,
+) -> prisma.types.LibraryAgentInclude:
+    """
+    Fully configurable includes for library agent queries with performance optimization.
+
+    Args:
+        user_id: User ID for filtering user-specific data
+        include_nodes: Whether to include graph nodes (default: True, needed for get_sub_graphs)
+        include_executions: Whether to include executions (default: True, safe with execution_limit)
+        execution_limit: Limit on executions to fetch (default: MAX_LIBRARY_AGENT_EXECUTIONS_FETCH)
+
+    Defaults maintain backward compatibility and safety - includes everything needed for all functionality.
+    For performance optimization, explicitly set include_nodes=False and include_executions=False
+    for listing views where frontend fetches data separately.
+
+    Performance impact:
+    - Default (full nodes + limited executions): Original performance, works everywhere
+    - Listing optimization (no nodes/executions): ~2s for 15 agents vs potential timeouts
+    - Unlimited executions: varies by user (thousands of executions = timeouts)
+    """
+    result: prisma.types.LibraryAgentInclude = {
+        "Creator": True,  # Always needed for creator info
     }
+
+    # Build AgentGraph include based on requested options
+    if include_nodes or include_executions:
+        agent_graph_include = {}
+
+        # Add nodes if requested (always full nodes)
+        if include_nodes:
+            agent_graph_include.update(AGENT_GRAPH_INCLUDE)  # Full nodes
+
+        # Add executions if requested
+        if include_executions:
+            agent_graph_include["Executions"] = {
+                "where": {"userId": user_id},
+                "order_by": {"createdAt": "desc"},
+                "take": execution_limit,
+            }
+
+        result["AgentGraph"] = cast(
+            prisma.types.AgentGraphArgsFromLibraryAgent,
+            {"include": agent_graph_include},
+        )
+    else:
+        # Default: Basic metadata only (fast - recommended for most use cases)
+        result["AgentGraph"] = True  # Basic graph metadata (name, description, id)
+
+    return result

autogpt_platform/backend/backend/data/integrations.py

@@ -11,7 +11,10 @@ from prisma.types import (
 from pydantic import Field, computed_field
 
 from backend.data.event_bus import AsyncRedisEventBus
-from backend.data.includes import INTEGRATION_WEBHOOK_INCLUDE
+from backend.data.includes import (
+    INTEGRATION_WEBHOOK_INCLUDE,
+    MAX_INTEGRATION_WEBHOOKS_FETCH,
+)
 from backend.integrations.providers import ProviderName
 from backend.integrations.webhooks.utils import webhook_ingress_url
 from backend.server.v2.library.model import LibraryAgentPreset
@@ -128,22 +131,36 @@ async def get_webhook(
 
 @overload
 async def get_all_webhooks_by_creds(
-    user_id: str, credentials_id: str, *, include_relations: Literal[True]
+    user_id: str,
+    credentials_id: str,
+    *,
+    include_relations: Literal[True],
+    limit: int = MAX_INTEGRATION_WEBHOOKS_FETCH,
 ) -> list[WebhookWithRelations]: ...
 @overload
 async def get_all_webhooks_by_creds(
-    user_id: str, credentials_id: str, *, include_relations: Literal[False] = False
+    user_id: str,
+    credentials_id: str,
+    *,
+    include_relations: Literal[False] = False,
+    limit: int = MAX_INTEGRATION_WEBHOOKS_FETCH,
 ) -> list[Webhook]: ...
 
 
 async def get_all_webhooks_by_creds(
-    user_id: str, credentials_id: str, *, include_relations: bool = False
+    user_id: str,
+    credentials_id: str,
+    *,
+    include_relations: bool = False,
+    limit: int = MAX_INTEGRATION_WEBHOOKS_FETCH,
 ) -> list[Webhook] | list[WebhookWithRelations]:
     if not credentials_id:
         raise ValueError("credentials_id must not be empty")
     webhooks = await IntegrationWebhook.prisma().find_many(
         where={"userId": user_id, "credentialsId": credentials_id},
         include=INTEGRATION_WEBHOOK_INCLUDE if include_relations else None,
+        order={"createdAt": "desc"},
+        take=limit,
     )
     return [
         (WebhookWithRelations if include_relations else Webhook).from_db(webhook)

autogpt_platform/backend/backend/data/model.py

@@ -270,6 +270,7 @@ def SchemaField(
     min_length: Optional[int] = None,
     max_length: Optional[int] = None,
     discriminator: Optional[str] = None,
+    format: Optional[str] = None,
     json_schema_extra: Optional[dict[str, Any]] = None,
 ) -> T:
     if default is PydanticUndefined and default_factory is None:
@@ -285,6 +286,7 @@ def SchemaField(
             "advanced": advanced,
             "hidden": hidden,
             "depends_on": depends_on,
+            "format": format,
             **(json_schema_extra or {}),
         }.items()
         if v is not None
@@ -345,6 +347,9 @@ class APIKeyCredentials(_BaseCredentials):
     """Unix timestamp (seconds) indicating when the API key expires (if at all)"""
 
     def auth_header(self) -> str:
+        # Linear API keys should not have Bearer prefix
+        if self.provider == "linear":
+            return self.api_key.get_secret_value()
         return f"Bearer {self.api_key.get_secret_value()}"
 
 

autogpt_platform/backend/backend/data/notifications.py

@@ -15,7 +15,7 @@ from prisma.types import (
 # from backend.notifications.models import NotificationEvent
 from pydantic import BaseModel, ConfigDict, EmailStr, Field, field_validator
 
-from backend.server.v2.store.exceptions import DatabaseError
+from backend.util.exceptions import DatabaseError
 from backend.util.json import SafeJson
 from backend.util.logging import TruncatedLogger
 
@@ -235,6 +235,7 @@ class BaseEventModel(BaseModel):
 
 
 class NotificationEventModel(BaseEventModel, Generic[NotificationDataType_co]):
+    id: Optional[str] = None  # None when creating, populated when reading from DB
     data: NotificationDataType_co
 
     @property
@@ -378,6 +379,7 @@ class NotificationPreference(BaseModel):
 
 
 class UserNotificationEventDTO(BaseModel):
+    id: str  # Added to track notifications for removal
     type: NotificationType
     data: dict
     created_at: datetime
@@ -386,6 +388,7 @@ class UserNotificationEventDTO(BaseModel):
     @staticmethod
     def from_db(model: NotificationEvent) -> "UserNotificationEventDTO":
         return UserNotificationEventDTO(
+            id=model.id,
             type=model.type,
             data=dict(model.data),
             created_at=model.createdAt,
@@ -541,6 +544,79 @@ async def empty_user_notification_batch(
         ) from e
 
 
+async def clear_all_user_notification_batches(user_id: str) -> None:
+    """Clear ALL notification batches for a user across all types.
+
+    Used when user's email is bounced/inactive and we should stop
+    trying to send them ANY emails.
+    """
+    try:
+        # Delete all notification events for this user
+        await NotificationEvent.prisma().delete_many(
+            where={"UserNotificationBatch": {"is": {"userId": user_id}}}
+        )
+
+        # Delete all batches for this user
+        await UserNotificationBatch.prisma().delete_many(where={"userId": user_id})
+
+        logger.info(f"Cleared all notification batches for user {user_id}")
+    except Exception as e:
+        raise DatabaseError(
+            f"Failed to clear all notification batches for user {user_id}: {e}"
+        ) from e
+
+
+async def remove_notifications_from_batch(
+    user_id: str, notification_type: NotificationType, notification_ids: list[str]
+) -> None:
+    """Remove specific notifications from a user's batch by their IDs.
+
+    This is used after successful sending to remove only the
+    sent notifications, preventing duplicates on retry.
+    """
+    if not notification_ids:
+        return
+
+    try:
+        # Delete the specific notification events
+        deleted_count = await NotificationEvent.prisma().delete_many(
+            where={
+                "id": {"in": notification_ids},
+                "UserNotificationBatch": {
+                    "is": {"userId": user_id, "type": notification_type}
+                },
+            }
+        )
+
+        logger.info(
+            f"Removed {deleted_count} notifications from batch for user {user_id}"
+        )
+
+        # Check if batch is now empty and delete it if so
+        remaining = await NotificationEvent.prisma().count(
+            where={
+                "UserNotificationBatch": {
+                    "is": {"userId": user_id, "type": notification_type}
+                }
+            }
+        )
+
+        if remaining == 0:
+            await UserNotificationBatch.prisma().delete_many(
+                where=UserNotificationBatchWhereInput(
+                    userId=user_id,
+                    type=notification_type,
+                )
+            )
+            logger.info(
+                f"Deleted empty batch for user {user_id} and type {notification_type}"
+            )
+    except Exception as e:
+        raise DatabaseError(
+            f"Failed to remove notifications from batch for user {user_id} and type {notification_type}: {e}"
+        ) from e
+
+
 async def get_user_notification_batch(
     user_id: str,
     notification_type: NotificationType,

autogpt_platform/backend/backend/data/onboarding.py

@@ -1,4 +1,5 @@
 import re
+from datetime import datetime
 from typing import Any, Optional
 
 import prisma
@@ -9,9 +10,9 @@ from prisma.types import UserOnboardingCreateInput, UserOnboardingUpdateInput
 
 from backend.data.block import get_blocks
 from backend.data.credit import get_user_credit_model
-from backend.data.graph import GraphModel
 from backend.data.model import CredentialsMetaInput
 from backend.server.v2.store.model import StoreAgentDetails
+from backend.util.cache import cached
 from backend.util.json import SafeJson
 
 # Mapping from user reason id to categories to search for when choosing agent to show
@@ -25,12 +26,10 @@ REASON_MAPPING: dict[str, list[str]] = {
 POINTS_AGENT_COUNT = 50  # Number of agents to calculate points for
 MIN_AGENT_COUNT = 2  # Minimum number of marketplace agents to enable onboarding
 
-user_credit = get_user_credit_model()
-
 
 class UserOnboardingUpdate(pydantic.BaseModel):
     completedSteps: Optional[list[OnboardingStep]] = None
-    notificationDot: Optional[bool] = None
+    walletShown: Optional[bool] = None
     notified: Optional[list[OnboardingStep]] = None
     usageReason: Optional[str] = None
     integrations: Optional[list[str]] = None
@@ -39,6 +38,8 @@ class UserOnboardingUpdate(pydantic.BaseModel):
     agentInput: Optional[dict[str, Any]] = None
     onboardingAgentExecutionId: Optional[str] = None
     agentRuns: Optional[int] = None
+    lastRunAt: Optional[datetime] = None
+    consecutiveRunDays: Optional[int] = None
 
 
 async def get_user_onboarding(user_id: str):
@@ -57,16 +58,22 @@ async def update_user_onboarding(user_id: str, data: UserOnboardingUpdate):
         update["completedSteps"] = list(set(data.completedSteps))
         for step in (
             OnboardingStep.AGENT_NEW_RUN,
-            OnboardingStep.RUN_AGENTS,
+            OnboardingStep.MARKETPLACE_VISIT,
             OnboardingStep.MARKETPLACE_ADD_AGENT,
             OnboardingStep.MARKETPLACE_RUN_AGENT,
             OnboardingStep.BUILDER_SAVE_AGENT,
-            OnboardingStep.BUILDER_RUN_AGENT,
+            OnboardingStep.RE_RUN_AGENT,
+            OnboardingStep.SCHEDULE_AGENT,
+            OnboardingStep.RUN_AGENTS,
+            OnboardingStep.RUN_3_DAYS,
+            OnboardingStep.TRIGGER_WEBHOOK,
+            OnboardingStep.RUN_14_DAYS,
+            OnboardingStep.RUN_AGENTS_100,
         ):
             if step in data.completedSteps:
                 await reward_user(user_id, step)
-    if data.notificationDot is not None:
-        update["notificationDot"] = data.notificationDot
+    if data.walletShown is not None:
+        update["walletShown"] = data.walletShown
     if data.notified is not None:
         update["notified"] = list(set(data.notified))
     if data.usageReason is not None:
@@ -83,6 +90,10 @@ async def update_user_onboarding(user_id: str, data: UserOnboardingUpdate):
         update["onboardingAgentExecutionId"] = data.onboardingAgentExecutionId
     if data.agentRuns is not None:
         update["agentRuns"] = data.agentRuns
+    if data.lastRunAt is not None:
+        update["lastRunAt"] = data.lastRunAt
+    if data.consecutiveRunDays is not None:
+        update["consecutiveRunDays"] = data.consecutiveRunDays
 
     return await UserOnboarding.prisma().upsert(
         where={"userId": user_id},
@@ -101,16 +112,28 @@ async def reward_user(user_id: str, step: OnboardingStep):
         # This is seen as a reward for the GET_RESULTS step in the wallet
         case OnboardingStep.AGENT_NEW_RUN:
             reward = 300
-        case OnboardingStep.RUN_AGENTS:
-            reward = 300
+        case OnboardingStep.MARKETPLACE_VISIT:
+            reward = 100
         case OnboardingStep.MARKETPLACE_ADD_AGENT:
             reward = 100
         case OnboardingStep.MARKETPLACE_RUN_AGENT:
             reward = 100
         case OnboardingStep.BUILDER_SAVE_AGENT:
             reward = 100
-        case OnboardingStep.BUILDER_RUN_AGENT:
+        case OnboardingStep.RE_RUN_AGENT:
             reward = 100
+        case OnboardingStep.SCHEDULE_AGENT:
+            reward = 100
+        case OnboardingStep.RUN_AGENTS:
+            reward = 300
+        case OnboardingStep.RUN_3_DAYS:
+            reward = 100
+        case OnboardingStep.TRIGGER_WEBHOOK:
+            reward = 100
+        case OnboardingStep.RUN_14_DAYS:
+            reward = 300
+        case OnboardingStep.RUN_AGENTS_100:
+            reward = 300
 
     if reward == 0:
         return
@@ -122,7 +145,8 @@ async def reward_user(user_id: str, step: OnboardingStep):
         return
 
     onboarding.rewardedFor.append(step)
-    await user_credit.onboarding_reward(user_id, reward, step)
+    user_credit_model = await get_user_credit_model(user_id)
+    await user_credit_model.onboarding_reward(user_id, reward, step)
     await UserOnboarding.prisma().update(
         where={"userId": user_id},
         data={
@@ -132,6 +156,22 @@ async def reward_user(user_id: str, step: OnboardingStep):
     )
 
 
+async def complete_webhook_trigger_step(user_id: str):
+    """
+    Completes the TRIGGER_WEBHOOK onboarding step for the user if not already completed.
+    """
+
+    onboarding = await get_user_onboarding(user_id)
+    if OnboardingStep.TRIGGER_WEBHOOK not in onboarding.completedSteps:
+        await update_user_onboarding(
+            user_id,
+            UserOnboardingUpdate(
+                completedSteps=onboarding.completedSteps
+                + [OnboardingStep.TRIGGER_WEBHOOK]
+            ),
+        )
+
+
 def clean_and_split(text: str) -> list[str]:
     """
     Removes all special characters from a string, truncates it to 100 characters,
@@ -236,8 +276,14 @@ async def get_recommended_agents(user_id: str) -> list[StoreAgentDetails]:
         for word in user_onboarding.integrations
     ]
 
+    where_clause["is_available"] = True
+
+    # Try to take only agents that are available and allowed for onboarding
     storeAgents = await prisma.models.StoreAgent.prisma().find_many(
-        where=prisma.types.StoreAgentWhereInput(**where_clause),
+        where={
+            "is_available": True,
+            "useForOnboarding": True,
+        },
         order=[
             {"featured": "desc"},
             {"runs": "desc"},
@@ -246,59 +292,16 @@ async def get_recommended_agents(user_id: str) -> list[StoreAgentDetails]:
         take=100,
     )
 
-    agentListings = await prisma.models.StoreListingVersion.prisma().find_many(
-        where={
-            "id": {"in": [agent.storeListingVersionId for agent in storeAgents]},
-        },
-        include={"AgentGraph": True},
-    )
-
-    for listing in agentListings:
-        agent = listing.AgentGraph
-        if agent is None:
-            continue
-        graph = GraphModel.from_db(agent)
-        # Remove agents with empty input schema
-        if not graph.input_schema:
-            storeAgents = [
-                a for a in storeAgents if a.storeListingVersionId != listing.id
-            ]
-            continue
-
-        # Remove agents with empty credentials
-        # Get nodes from this agent that have credentials
-        nodes = await prisma.models.AgentNode.prisma().find_many(
-            where={
-                "agentGraphId": agent.id,
-                "agentBlockId": {"in": list(CREDENTIALS_FIELDS.keys())},
-            },
-        )
-        for node in nodes:
-            block_id = node.agentBlockId
-            field_name = CREDENTIALS_FIELDS[block_id]
-            # If there are no credentials or they are empty, remove the agent
-            # FIXME ignores default values
-            if (
-                field_name not in node.constantInput
-                or node.constantInput[field_name] is None
-            ):
-                storeAgents = [
-                    a for a in storeAgents if a.storeListingVersionId != listing.id
-                ]
-                break
-
-    # If there are less than 2 agents, add more agents to the list
+    # If not enough agents found, relax the useForOnboarding filter
     if len(storeAgents) < 2:
-        storeAgents += await prisma.models.StoreAgent.prisma().find_many(
-            where={
-                "listing_id": {"not_in": [agent.listing_id for agent in storeAgents]},
-            },
+        storeAgents = await prisma.models.StoreAgent.prisma().find_many(
+            where=prisma.types.StoreAgentWhereInput(**where_clause),
             order=[
                 {"featured": "desc"},
                 {"runs": "desc"},
                 {"rating": "desc"},
             ],
-            take=2 - len(storeAgents),
+            take=100,
         )
 
     # Calculate points for the first X agents and choose the top 2
@@ -333,8 +336,13 @@ async def get_recommended_agents(user_id: str) -> list[StoreAgentDetails]:
     ]
 
 
+@cached(maxsize=1, ttl_seconds=300)  # Cache for 5 minutes since this rarely changes
 async def onboarding_enabled() -> bool:
+    """
+    Check if onboarding should be enabled based on store agent count.
+    Cached to prevent repeated slow database queries.
+    """
+    # Use a more efficient query that stops counting after finding enough agents
     count = await prisma.models.StoreAgent.prisma().count(take=MIN_AGENT_COUNT + 1)
-
-    # Onboading is enabled if there are at least 2 agents in the store
+    # Onboarding is enabled if there are at least 2 agents in the store
     return count >= MIN_AGENT_COUNT

autogpt_platform/backend/backend/data/partial_types.py

@@ -0,0 +1,5 @@
+import prisma.models
+
+
+class StoreAgentWithRank(prisma.models.StoreAgent):
+    rank: float

autogpt_platform/backend/backend/data/redis_client.py

@@ -1,11 +1,11 @@
 import logging
 import os
 
-from autogpt_libs.utils.cache import cached, thread_cached
 from dotenv import load_dotenv
 from redis import Redis
 from redis.asyncio import Redis as AsyncRedis
 
+from backend.util.cache import cached, thread_cached
 from backend.util.retry import conn_retry
 
 load_dotenv()
@@ -34,7 +34,7 @@ def disconnect():
     get_redis().close()
 
 
-@cached()
+@cached(ttl_seconds=3600)
 def get_redis() -> Redis:
     return connect()
 

autogpt_platform/backend/backend/data/user.py

@@ -7,7 +7,6 @@ from typing import Optional, cast
 from urllib.parse import quote_plus
 
 from autogpt_libs.auth.models import DEFAULT_USER_ID
-from autogpt_libs.utils.cache import cached
 from fastapi import HTTPException
 from prisma.enums import NotificationType
 from prisma.models import User as PrismaUser
@@ -16,8 +15,9 @@ from prisma.types import JsonFilter, UserCreateInput, UserUpdateInput
 from backend.data.db import prisma
 from backend.data.model import User, UserIntegrations, UserMetadata
 from backend.data.notifications import NotificationPreference, NotificationPreferenceDTO
-from backend.server.v2.store.exceptions import DatabaseError
+from backend.util.cache import cached
 from backend.util.encryption import JSONCryptor
+from backend.util.exceptions import DatabaseError
 from backend.util.json import SafeJson
 from backend.util.settings import Settings
 
@@ -354,6 +354,36 @@ async def set_user_email_verification(user_id: str, verified: bool) -> None:
         ) from e
 
 
+async def disable_all_user_notifications(user_id: str) -> None:
+    """Disable all notification preferences for a user.
+
+    Used when user's email bounces/is inactive to prevent any future notifications.
+    """
+    try:
+        await PrismaUser.prisma().update(
+            where={"id": user_id},
+            data={
+                "notifyOnAgentRun": False,
+                "notifyOnZeroBalance": False,
+                "notifyOnLowBalance": False,
+                "notifyOnBlockExecutionFailed": False,
+                "notifyOnContinuousAgentError": False,
+                "notifyOnDailySummary": False,
+                "notifyOnWeeklySummary": False,
+                "notifyOnMonthlySummary": False,
+                "notifyOnAgentApproved": False,
+                "notifyOnAgentRejected": False,
+            },
+        )
+        # Invalidate cache for this user
+        get_user_by_id.cache_delete(user_id)
+        logger.info(f"Disabled all notification preferences for user {user_id}")
+    except Exception as e:
+        raise DatabaseError(
+            f"Failed to disable notifications for user {user_id}: {e}"
+        ) from e
+
+
 async def get_user_email_verification(user_id: str) -> bool:
     """Get the email verification status for a user."""
     try:

autogpt_platform/backend/backend/executor/activity_status_generator.py

@@ -4,7 +4,12 @@ Module for generating AI-based activity status for graph executions.
 
 import json
 import logging
-from typing import TYPE_CHECKING, Any, NotRequired, TypedDict
+from typing import TYPE_CHECKING, Any, TypedDict
+
+try:
+    from typing import NotRequired
+except ImportError:
+    from typing_extensions import NotRequired
 
 from pydantic import SecretStr
 
@@ -146,17 +151,35 @@ async def generate_activity_status_for_execution(
                     "Focus on the ACTUAL TASK the user wanted done, not the internal workflow steps. "
                     "Avoid technical terms like 'workflow', 'execution', 'components', 'nodes', 'processing', etc. "
                     "Keep it to 3 sentences maximum. Be conversational and human-friendly.\n\n"
+                    "UNDERSTAND THE INTENDED PURPOSE:\n"
+                    "- FIRST: Read the graph description carefully to understand what the user wanted to accomplish\n"
+                    "- The graph name and description tell you the main goal/intention of this automation\n"
+                    "- Use this intended purpose as your PRIMARY criteria for success/failure evaluation\n"
+                    "- Ask yourself: 'Did this execution actually accomplish what the graph was designed to do?'\n\n"
+                    "CRITICAL OUTPUT ANALYSIS:\n"
+                    "- Check if blocks that should produce user-facing results actually produced outputs\n"
+                    "- Blocks with names containing 'Output', 'Post', 'Create', 'Send', 'Publish', 'Generate' are usually meant to produce final results\n"
+                    "- If these critical blocks have NO outputs (empty recent_outputs), the task likely FAILED even if status shows 'completed'\n"
+                    "- Sub-agents (AgentExecutorBlock) that produce no outputs usually indicate failed sub-tasks\n"
+                    "- Most importantly: Does the execution result match what the graph description promised to deliver?\n\n"
+                    "SUCCESS EVALUATION BASED ON INTENTION:\n"
+                    "- If the graph is meant to 'create blog posts' → check if blog content was actually created\n"
+                    "- If the graph is meant to 'send emails' → check if emails were actually sent\n"
+                    "- If the graph is meant to 'analyze data' → check if analysis results were produced\n"
+                    "- If the graph is meant to 'generate reports' → check if reports were generated\n"
+                    "- Technical completion ≠ goal achievement. Focus on whether the USER'S INTENDED OUTCOME was delivered\n\n"
                     "IMPORTANT: Be HONEST about what actually happened:\n"
                     "- If the input was invalid/nonsensical, say so directly\n"
                     "- If the task failed, explain what went wrong in simple terms\n"
                     "- If errors occurred, focus on what the user needs to know\n"
-                    "- Only claim success if the task was genuinely completed\n"
-                    "- Don't sugar-coat failures or present them as helpful feedback\n\n"
+                    "- Only claim success if the INTENDED PURPOSE was genuinely accomplished AND produced expected outputs\n"
+                    "- Don't sugar-coat failures or present them as helpful feedback\n"
+                    "- ESPECIALLY: If the graph's main purpose wasn't achieved, this is a failure regardless of 'completed' status\n\n"
                     "Understanding Errors:\n"
                     "- Node errors: Individual steps may fail but the overall task might still complete (e.g., one data source fails but others work)\n"
                     "- Graph error (in overall_status.graph_error): This means the entire execution failed and nothing was accomplished\n"
-                    "- Even if execution shows 'completed', check if critical nodes failed that would prevent the desired outcome\n"
-                    "- Focus on the end result the user wanted, not whether technical steps completed"
+                    "- Missing outputs from critical blocks: Even if no errors, this means the task failed to produce expected results\n"
+                    "- Focus on whether the graph's intended purpose was fulfilled, not whether technical steps completed"
                 ),
             },
             {
@@ -165,15 +188,28 @@ async def generate_activity_status_for_execution(
                     f"A user ran '{graph_name}' to accomplish something. Based on this execution data, "
                     f"write what they achieved in simple, user-friendly terms:\n\n"
                     f"{json.dumps(execution_data, indent=2)}\n\n"
-                    "CRITICAL: Check overall_status.graph_error FIRST - if present, the entire execution failed.\n"
-                    "Then check individual node errors to understand partial failures.\n\n"
+                    "ANALYSIS CHECKLIST:\n"
+                    "1. READ graph_info.description FIRST - this tells you what the user intended to accomplish\n"
+                    "2. Check overall_status.graph_error - if present, the entire execution failed\n"
+                    "3. Look for nodes with 'Output', 'Post', 'Create', 'Send', 'Publish', 'Generate' in their block_name\n"
+                    "4. Check if these critical blocks have empty recent_outputs arrays - this indicates failure\n"
+                    "5. Look for AgentExecutorBlock (sub-agents) with no outputs - this suggests sub-task failures\n"
+                    "6. Count how many nodes produced outputs vs total nodes - low ratio suggests problems\n"
+                    "7. MOST IMPORTANT: Does the execution outcome match what graph_info.description promised?\n\n"
+                    "INTENTION-BASED EVALUATION:\n"
+                    "- If description mentions 'blog writing' → did it create blog content?\n"
+                    "- If description mentions 'email automation' → were emails actually sent?\n"
+                    "- If description mentions 'data analysis' → were analysis results produced?\n"
+                    "- If description mentions 'content generation' → was content actually generated?\n"
+                    "- If description mentions 'social media posting' → were posts actually made?\n"
+                    "- Match the outputs to the stated intention, not just technical completion\n\n"
                     "Write 1-3 sentences about what the user accomplished, such as:\n"
                     "- 'I analyzed your resume and provided detailed feedback for the IT industry.'\n"
-                    "- 'I couldn't analyze your resume because the input was just nonsensical text.'\n"
-                    "- 'I failed to complete the task due to missing API access.'\n"
+                    "- 'I couldn't complete the task because critical steps failed to produce any results.'\n"
+                    "- 'I failed to generate the content you requested due to missing API access.'\n"
                     "- 'I extracted key information from your documents and organized it into a summary.'\n"
-                    "- 'The task failed to run due to system configuration issues.'\n\n"
-                    "Focus on what ACTUALLY happened, not what was attempted."
+                    "- 'The task failed because the blog post creation step didn't produce any output.'\n\n"
+                    "BE CRITICAL: If the graph's intended purpose (from description) wasn't achieved, report this as a failure even if status is 'completed'."
                 ),
             },
         ]
@@ -197,6 +233,7 @@ async def generate_activity_status_for_execution(
         logger.debug(
             f"Generated activity status for {graph_exec_id}: {activity_status}"
         )
+
         return activity_status
 
     except Exception as e:

autogpt_platform/backend/backend/executor/cluster_lock.py

@@ -0,0 +1,115 @@
+"""Redis-based distributed locking for cluster coordination."""
+
+import logging
+import time
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from redis import Redis
+
+logger = logging.getLogger(__name__)
+
+
+class ClusterLock:
+    """Simple Redis-based distributed lock for preventing duplicate execution."""
+
+    def __init__(self, redis: "Redis", key: str, owner_id: str, timeout: int = 300):
+        self.redis = redis
+        self.key = key
+        self.owner_id = owner_id
+        self.timeout = timeout
+        self._last_refresh = 0.0
+
+    def try_acquire(self) -> str | None:
+        """Try to acquire the lock.
+
+        Returns:
+            - owner_id (self.owner_id) if successfully acquired
+            - different owner_id if someone else holds the lock
+            - None if Redis is unavailable or other error
+        """
+        try:
+            success = self.redis.set(self.key, self.owner_id, nx=True, ex=self.timeout)
+            if success:
+                self._last_refresh = time.time()
+                return self.owner_id  # Successfully acquired
+
+            # Failed to acquire, get current owner
+            current_value = self.redis.get(self.key)
+            if current_value:
+                current_owner = (
+                    current_value.decode("utf-8")
+                    if isinstance(current_value, bytes)
+                    else str(current_value)
+                )
+                return current_owner
+
+            # Key doesn't exist but we failed to set it - race condition or Redis issue
+            return None
+
+        except Exception as e:
+            logger.error(f"ClusterLock.try_acquire failed for key {self.key}: {e}")
+            return None
+
+    def refresh(self) -> bool:
+        """Refresh lock TTL if we still own it.
+
+        Rate limited to at most once every timeout/10 seconds (minimum 1 second).
+        During rate limiting, still verifies lock existence but skips TTL extension.
+        Setting _last_refresh to 0 bypasses rate limiting for testing.
+        """
+        # Calculate refresh interval: max(timeout // 10, 1)
+        refresh_interval = max(self.timeout // 10, 1)
+        current_time = time.time()
+
+        # Check if we're within the rate limit period
+        # _last_refresh == 0 forces a refresh (bypasses rate limiting for testing)
+        is_rate_limited = (
+            self._last_refresh > 0
+            and (current_time - self._last_refresh) < refresh_interval
+        )
+
+        try:
+            # Always verify lock existence, even during rate limiting
+            current_value = self.redis.get(self.key)
+            if not current_value:
+                self._last_refresh = 0
+                return False
+
+            stored_owner = (
+                current_value.decode("utf-8")
+                if isinstance(current_value, bytes)
+                else str(current_value)
+            )
+            if stored_owner != self.owner_id:
+                self._last_refresh = 0
+                return False
+
+            # If rate limited, return True but don't update TTL or timestamp
+            if is_rate_limited:
+                return True
+
+            # Perform actual refresh
+            if self.redis.expire(self.key, self.timeout):
+                self._last_refresh = current_time
+                return True
+
+            self._last_refresh = 0
+            return False
+
+        except Exception as e:
+            logger.error(f"ClusterLock.refresh failed for key {self.key}: {e}")
+            self._last_refresh = 0
+            return False
+
+    def release(self):
+        """Release the lock."""
+        if self._last_refresh == 0:
+            return
+
+        try:
+            self.redis.delete(self.key)
+        except Exception:
+            pass
+
+        self._last_refresh = 0.0

autogpt_platform/backend/backend/executor/cluster_lock_test.py

@@ -0,0 +1,507 @@
+"""
+Integration tests for ClusterLock - Redis-based distributed locking.
+
+Tests the complete lock lifecycle without mocking Redis to ensure
+real-world behavior is correct. Covers acquisition, refresh, expiry,
+contention, and error scenarios.
+"""
+
+import logging
+import time
+import uuid
+from threading import Thread
+
+import pytest
+import redis
+
+from .cluster_lock import ClusterLock
+
+logger = logging.getLogger(__name__)
+
+
+@pytest.fixture
+def redis_client():
+    """Get Redis client for testing using same config as backend."""
+    from backend.data.redis_client import HOST, PASSWORD, PORT
+
+    # Use same config as backend but without decode_responses since ClusterLock needs raw bytes
+    client = redis.Redis(
+        host=HOST,
+        port=PORT,
+        password=PASSWORD,
+        decode_responses=False,  # ClusterLock needs raw bytes for ownership verification
+    )
+
+    # Clean up any existing test keys
+    try:
+        for key in client.scan_iter(match="test_lock:*"):
+            client.delete(key)
+    except Exception:
+        pass  # Ignore cleanup errors
+
+    return client
+
+
+@pytest.fixture
+def lock_key():
+    """Generate unique lock key for each test."""
+    return f"test_lock:{uuid.uuid4()}"
+
+
+@pytest.fixture
+def owner_id():
+    """Generate unique owner ID for each test."""
+    return str(uuid.uuid4())
+
+
+class TestClusterLockBasic:
+    """Basic lock acquisition and release functionality."""
+
+    def test_lock_acquisition_success(self, redis_client, lock_key, owner_id):
+        """Test basic lock acquisition succeeds."""
+        lock = ClusterLock(redis_client, lock_key, owner_id, timeout=60)
+
+        # Lock should be acquired successfully
+        result = lock.try_acquire()
+        assert result == owner_id  # Returns our owner_id when successfully acquired
+        assert lock._last_refresh > 0
+
+        # Lock key should exist in Redis
+        assert redis_client.exists(lock_key) == 1
+        assert redis_client.get(lock_key).decode("utf-8") == owner_id
+
+    def test_lock_acquisition_contention(self, redis_client, lock_key):
+        """Test second acquisition fails when lock is held."""
+        owner1 = str(uuid.uuid4())
+        owner2 = str(uuid.uuid4())
+
+        lock1 = ClusterLock(redis_client, lock_key, owner1, timeout=60)
+        lock2 = ClusterLock(redis_client, lock_key, owner2, timeout=60)
+
+        # First lock should succeed
+        result1 = lock1.try_acquire()
+        assert result1 == owner1  # Successfully acquired, returns our owner_id
+
+        # Second lock should fail and return the first owner
+        result2 = lock2.try_acquire()
+        assert result2 == owner1  # Returns the current owner (first owner)
+        assert lock2._last_refresh == 0
+
+    def test_lock_release_deletes_redis_key(self, redis_client, lock_key, owner_id):
+        """Test lock release deletes Redis key and marks locally as released."""
+        lock = ClusterLock(redis_client, lock_key, owner_id, timeout=60)
+
+        lock.try_acquire()
+        assert lock._last_refresh > 0
+        assert redis_client.exists(lock_key) == 1
+
+        # Release should delete Redis key and mark locally as released
+        lock.release()
+        assert lock._last_refresh == 0
+        assert lock._last_refresh == 0.0
+
+        # Redis key should be deleted for immediate release
+        assert redis_client.exists(lock_key) == 0
+
+        # Another lock should be able to acquire immediately
+        new_owner_id = str(uuid.uuid4())
+        new_lock = ClusterLock(redis_client, lock_key, new_owner_id, timeout=60)
+        assert new_lock.try_acquire() == new_owner_id
+
+
+class TestClusterLockRefresh:
+    """Lock refresh and TTL management."""
+
+    def test_lock_refresh_success(self, redis_client, lock_key, owner_id):
+        """Test lock refresh extends TTL."""
+        lock = ClusterLock(redis_client, lock_key, owner_id, timeout=60)
+
+        lock.try_acquire()
+        original_ttl = redis_client.ttl(lock_key)
+
+        # Wait a bit then refresh
+        time.sleep(1)
+        lock._last_refresh = 0  # Force refresh past rate limit
+        assert lock.refresh() is True
+
+        # TTL should be reset to full timeout (allow for small timing differences)
+        new_ttl = redis_client.ttl(lock_key)
+        assert new_ttl >= original_ttl or new_ttl >= 58  # Allow for timing variance
+
+    def test_lock_refresh_rate_limiting(self, redis_client, lock_key, owner_id):
+        """Test refresh is rate-limited to timeout/10."""
+        lock = ClusterLock(
+            redis_client, lock_key, owner_id, timeout=100
+        )  # 100s timeout
+
+        lock.try_acquire()
+
+        # First refresh should work
+        assert lock.refresh() is True
+        first_refresh_time = lock._last_refresh
+
+        # Immediate second refresh should be skipped (rate limited) but verify key exists
+        assert lock.refresh() is True  # Returns True but skips actual refresh
+        assert lock._last_refresh == first_refresh_time  # Time unchanged
+
+    def test_lock_refresh_verifies_existence_during_rate_limit(
+        self, redis_client, lock_key, owner_id
+    ):
+        """Test refresh verifies lock existence even during rate limiting."""
+        lock = ClusterLock(redis_client, lock_key, owner_id, timeout=100)
+
+        lock.try_acquire()
+
+        # Manually delete the key (simulates expiry or external deletion)
+        redis_client.delete(lock_key)
+
+        # Refresh should detect missing key even during rate limit period
+        assert lock.refresh() is False
+        assert lock._last_refresh == 0
+
+    def test_lock_refresh_ownership_lost(self, redis_client, lock_key, owner_id):
+        """Test refresh fails when ownership is lost."""
+        lock = ClusterLock(redis_client, lock_key, owner_id, timeout=60)
+
+        lock.try_acquire()
+
+        # Simulate another process taking the lock
+        different_owner = str(uuid.uuid4())
+        redis_client.set(lock_key, different_owner, ex=60)
+
+        # Force refresh past rate limit and verify it fails
+        lock._last_refresh = 0  # Force refresh past rate limit
+        assert lock.refresh() is False
+        assert lock._last_refresh == 0
+
+    def test_lock_refresh_when_not_acquired(self, redis_client, lock_key, owner_id):
+        """Test refresh fails when lock was never acquired."""
+        lock = ClusterLock(redis_client, lock_key, owner_id, timeout=60)
+
+        # Refresh without acquiring should fail
+        assert lock.refresh() is False
+
+
+class TestClusterLockExpiry:
+    """Lock expiry and timeout behavior."""
+
+    def test_lock_natural_expiry(self, redis_client, lock_key, owner_id):
+        """Test lock expires naturally via Redis TTL."""
+        lock = ClusterLock(
+            redis_client, lock_key, owner_id, timeout=2
+        )  # 2 second timeout
+
+        lock.try_acquire()
+        assert redis_client.exists(lock_key) == 1
+
+        # Wait for expiry
+        time.sleep(3)
+        assert redis_client.exists(lock_key) == 0
+
+        # New lock with same key should succeed
+        new_lock = ClusterLock(redis_client, lock_key, owner_id, timeout=60)
+        assert new_lock.try_acquire() == owner_id
+
+    def test_lock_refresh_prevents_expiry(self, redis_client, lock_key, owner_id):
+        """Test refreshing prevents lock from expiring."""
+        lock = ClusterLock(
+            redis_client, lock_key, owner_id, timeout=3
+        )  # 3 second timeout
+
+        lock.try_acquire()
+
+        # Wait and refresh before expiry
+        time.sleep(1)
+        lock._last_refresh = 0  # Force refresh past rate limit
+        assert lock.refresh() is True
+
+        # Wait beyond original timeout
+        time.sleep(2.5)
+        assert redis_client.exists(lock_key) == 1  # Should still exist
+
+
+class TestClusterLockConcurrency:
+    """Concurrent access patterns."""
+
+    def test_multiple_threads_contention(self, redis_client, lock_key):
+        """Test multiple threads competing for same lock."""
+        num_threads = 5
+        successful_acquisitions = []
+
+        def try_acquire_lock(thread_id):
+            owner_id = f"thread_{thread_id}"
+            lock = ClusterLock(redis_client, lock_key, owner_id, timeout=60)
+            if lock.try_acquire() == owner_id:
+                successful_acquisitions.append(thread_id)
+                time.sleep(0.1)  # Hold lock briefly
+                lock.release()
+
+        threads = []
+        for i in range(num_threads):
+            thread = Thread(target=try_acquire_lock, args=(i,))
+            threads.append(thread)
+            thread.start()
+
+        for thread in threads:
+            thread.join()
+
+        # Only one thread should have acquired the lock
+        assert len(successful_acquisitions) == 1
+
+    def test_sequential_lock_reuse(self, redis_client, lock_key):
+        """Test lock can be reused after natural expiry."""
+        owners = [str(uuid.uuid4()) for _ in range(3)]
+
+        for i, owner_id in enumerate(owners):
+            lock = ClusterLock(redis_client, lock_key, owner_id, timeout=1)  # 1 second
+
+            assert lock.try_acquire() == owner_id
+            time.sleep(1.5)  # Wait for expiry
+
+            # Verify lock expired
+            assert redis_client.exists(lock_key) == 0
+
+    def test_refresh_during_concurrent_access(self, redis_client, lock_key):
+        """Test lock refresh works correctly during concurrent access attempts."""
+        owner1 = str(uuid.uuid4())
+        owner2 = str(uuid.uuid4())
+
+        lock1 = ClusterLock(redis_client, lock_key, owner1, timeout=5)
+        lock2 = ClusterLock(redis_client, lock_key, owner2, timeout=5)
+
+        # Thread 1 holds lock and refreshes
+        assert lock1.try_acquire() == owner1
+
+        def refresh_continuously():
+            for _ in range(10):
+                lock1._last_refresh = 0  # Force refresh
+                lock1.refresh()
+                time.sleep(0.1)
+
+        def try_acquire_continuously():
+            attempts = 0
+            while attempts < 20:
+                if lock2.try_acquire() == owner2:
+                    return True
+                time.sleep(0.1)
+                attempts += 1
+            return False
+
+        refresh_thread = Thread(target=refresh_continuously)
+        acquire_thread = Thread(target=try_acquire_continuously)
+
+        refresh_thread.start()
+        acquire_thread.start()
+
+        refresh_thread.join()
+        acquire_thread.join()
+
+        # Lock1 should still own the lock due to refreshes
+        assert lock1._last_refresh > 0
+        assert lock2._last_refresh == 0
+
+
+class TestClusterLockErrorHandling:
+    """Error handling and edge cases."""
+
+    def test_redis_connection_failure_on_acquire(self, lock_key, owner_id):
+        """Test graceful handling when Redis is unavailable during acquisition."""
+        # Use invalid Redis connection
+        bad_redis = redis.Redis(
+            host="invalid_host", port=1234, socket_connect_timeout=1
+        )
+        lock = ClusterLock(bad_redis, lock_key, owner_id, timeout=60)
+
+        # Should return None for Redis connection failures
+        result = lock.try_acquire()
+        assert result is None  # Returns None when Redis fails
+        assert lock._last_refresh == 0
+
+    def test_redis_connection_failure_on_refresh(
+        self, redis_client, lock_key, owner_id
+    ):
+        """Test graceful handling when Redis fails during refresh."""
+        lock = ClusterLock(redis_client, lock_key, owner_id, timeout=60)
+
+        # Acquire normally
+        assert lock.try_acquire() == owner_id
+
+        # Replace Redis client with failing one
+        lock.redis = redis.Redis(
+            host="invalid_host", port=1234, socket_connect_timeout=1
+        )
+
+        # Refresh should fail gracefully
+        lock._last_refresh = 0  # Force refresh
+        assert lock.refresh() is False
+        assert lock._last_refresh == 0
+
+    def test_invalid_lock_parameters(self, redis_client):
+        """Test validation of lock parameters."""
+        owner_id = str(uuid.uuid4())
+
+        # All parameters are now simple - no validation needed
+        # Just test basic construction works
+        lock = ClusterLock(redis_client, "test_key", owner_id, timeout=60)
+        assert lock.key == "test_key"
+        assert lock.owner_id == owner_id
+        assert lock.timeout == 60
+
+    def test_refresh_after_redis_key_deleted(self, redis_client, lock_key, owner_id):
+        """Test refresh behavior when Redis key is manually deleted."""
+        lock = ClusterLock(redis_client, lock_key, owner_id, timeout=60)
+
+        lock.try_acquire()
+
+        # Manually delete the key (simulates external deletion)
+        redis_client.delete(lock_key)
+
+        # Refresh should fail and mark as not acquired
+        lock._last_refresh = 0  # Force refresh
+        assert lock.refresh() is False
+        assert lock._last_refresh == 0
+
+
+class TestClusterLockDynamicRefreshInterval:
+    """Dynamic refresh interval based on timeout."""
+
+    def test_refresh_interval_calculation(self, redis_client, lock_key, owner_id):
+        """Test refresh interval is calculated as max(timeout/10, 1)."""
+        test_cases = [
+            (5, 1),  # 5/10 = 0, but minimum is 1
+            (10, 1),  # 10/10 = 1
+            (30, 3),  # 30/10 = 3
+            (100, 10),  # 100/10 = 10
+            (200, 20),  # 200/10 = 20
+            (1000, 100),  # 1000/10 = 100
+        ]
+
+        for timeout, expected_interval in test_cases:
+            lock = ClusterLock(
+                redis_client, f"{lock_key}_{timeout}", owner_id, timeout=timeout
+            )
+            lock.try_acquire()
+
+            # Calculate expected interval using same logic as implementation
+            refresh_interval = max(timeout // 10, 1)
+            assert refresh_interval == expected_interval
+
+            # Test rate limiting works with calculated interval
+            assert lock.refresh() is True
+            first_refresh_time = lock._last_refresh
+
+            # Sleep less than interval - should be rate limited
+            time.sleep(0.1)
+            assert lock.refresh() is True
+            assert lock._last_refresh == first_refresh_time  # No actual refresh
+
+
+class TestClusterLockRealWorldScenarios:
+    """Real-world usage patterns."""
+
+    def test_execution_coordination_simulation(self, redis_client):
+        """Simulate graph execution coordination across multiple pods."""
+        graph_exec_id = str(uuid.uuid4())
+        lock_key = f"execution:{graph_exec_id}"
+
+        # Simulate 3 pods trying to execute same graph
+        pods = [f"pod_{i}" for i in range(3)]
+        execution_results = {}
+
+        def execute_graph(pod_id):
+            """Simulate graph execution with cluster lock."""
+            lock = ClusterLock(redis_client, lock_key, pod_id, timeout=300)
+
+            if lock.try_acquire() == pod_id:
+                # Simulate execution work
+                execution_results[pod_id] = "executed"
+                time.sleep(0.1)
+                lock.release()
+            else:
+                execution_results[pod_id] = "rejected"
+
+        threads = []
+        for pod_id in pods:
+            thread = Thread(target=execute_graph, args=(pod_id,))
+            threads.append(thread)
+            thread.start()
+
+        for thread in threads:
+            thread.join()
+
+        # Only one pod should have executed
+        executed_count = sum(
+            1 for result in execution_results.values() if result == "executed"
+        )
+        rejected_count = sum(
+            1 for result in execution_results.values() if result == "rejected"
+        )
+
+        assert executed_count == 1
+        assert rejected_count == 2
+
+    def test_long_running_execution_with_refresh(
+        self, redis_client, lock_key, owner_id
+    ):
+        """Test lock maintains ownership during long execution with periodic refresh."""
+        lock = ClusterLock(
+            redis_client, lock_key, owner_id, timeout=30
+        )  # 30 second timeout, refresh interval = max(30//10, 1) = 3 seconds
+
+        def long_execution_with_refresh():
+            """Simulate long-running execution with periodic refresh."""
+            assert lock.try_acquire() == owner_id
+
+            # Simulate 10 seconds of work with refreshes every 2 seconds
+            # This respects rate limiting - actual refreshes will happen at 0s, 3s, 6s, 9s
+            try:
+                for i in range(5):  # 5 iterations * 2 seconds = 10 seconds total
+                    time.sleep(2)
+                    refresh_success = lock.refresh()
+                    assert refresh_success is True, f"Refresh failed at iteration {i}"
+                return "completed"
+            finally:
+                lock.release()
+
+        # Should complete successfully without losing lock
+        result = long_execution_with_refresh()
+        assert result == "completed"
+
+    def test_graceful_degradation_pattern(self, redis_client, lock_key):
+        """Test graceful degradation when Redis becomes unavailable."""
+        owner_id = str(uuid.uuid4())
+        lock = ClusterLock(
+            redis_client, lock_key, owner_id, timeout=3
+        )  # Use shorter timeout
+
+        # Normal operation
+        assert lock.try_acquire() == owner_id
+        lock._last_refresh = 0  # Force refresh past rate limit
+        assert lock.refresh() is True
+
+        # Simulate Redis becoming unavailable
+        original_redis = lock.redis
+        lock.redis = redis.Redis(
+            host="invalid_host",
+            port=1234,
+            socket_connect_timeout=1,
+            decode_responses=False,
+        )
+
+        # Should degrade gracefully
+        lock._last_refresh = 0  # Force refresh past rate limit
+        assert lock.refresh() is False
+        assert lock._last_refresh == 0
+
+        # Restore Redis and verify can acquire again
+        lock.redis = original_redis
+        # Wait for original lock to expire (use longer wait for 3s timeout)
+        time.sleep(4)
+
+        new_lock = ClusterLock(redis_client, lock_key, owner_id, timeout=60)
+        assert new_lock.try_acquire() == owner_id
+
+
+if __name__ == "__main__":
+    # Run specific test for quick validation
+    pytest.main([__file__, "-v"])

autogpt_platform/backend/backend/executor/database.py

@@ -1,5 +1,6 @@
 import logging
-from typing import Callable, Concatenate, ParamSpec, TypeVar, cast
+from contextlib import asynccontextmanager
+from typing import TYPE_CHECKING, Callable, Concatenate, ParamSpec, TypeVar, cast
 
 from backend.data import db
 from backend.data.credit import UsageTransactionMetadata, get_user_credit_model
@@ -9,6 +10,7 @@ from backend.data.execution import (
     get_execution_kv_data,
     get_graph_execution_meta,
     get_graph_executions,
+    get_graph_executions_count,
     get_latest_node_execution,
     get_node_execution,
     get_node_executions,
@@ -28,14 +30,17 @@ from backend.data.graph import (
     get_node,
 )
 from backend.data.notifications import (
+    clear_all_user_notification_batches,
     create_or_add_to_user_notification_batch,
     empty_user_notification_batch,
     get_all_batches_by_type,
     get_user_notification_batch,
     get_user_notification_oldest_message_in_batch,
+    remove_notifications_from_batch,
 )
 from backend.data.user import (
     get_active_user_ids_in_timerange,
+    get_user_by_id,
     get_user_email_by_id,
     get_user_email_verification,
     get_user_integrations,
@@ -53,8 +58,10 @@ from backend.util.service import (
 )
 from backend.util.settings import Config
 
+if TYPE_CHECKING:
+    from fastapi import FastAPI
+
 config = Config()
-_user_credit_model = get_user_credit_model()
 logger = logging.getLogger(__name__)
 P = ParamSpec("P")
 R = TypeVar("R")
@@ -63,24 +70,27 @@ R = TypeVar("R")
 async def _spend_credits(
     user_id: str, cost: int, metadata: UsageTransactionMetadata
 ) -> int:
-    return await _user_credit_model.spend_credits(user_id, cost, metadata)
+    user_credit_model = await get_user_credit_model(user_id)
+    return await user_credit_model.spend_credits(user_id, cost, metadata)
 
 
 async def _get_credits(user_id: str) -> int:
-    return await _user_credit_model.get_credits(user_id)
+    user_credit_model = await get_user_credit_model(user_id)
+    return await user_credit_model.get_credits(user_id)
 
 
 class DatabaseManager(AppService):
+    @asynccontextmanager
+    async def lifespan(self, app: "FastAPI"):
+        async with super().lifespan(app):
+            logger.info(f"[{self.service_name}] ⏳ Connecting to Database...")
+            await db.connect()
 
-    def run_service(self) -> None:
-        logger.info(f"[{self.service_name}] ⏳ Connecting to Database...")
-        self.run_and_wait(db.connect())
-        super().run_service()
+            logger.info(f"[{self.service_name}] ✅ Ready")
+            yield
 
-    def cleanup(self):
-        super().cleanup()
-        logger.info(f"[{self.service_name}] ⏳ Disconnecting Database...")
-        self.run_and_wait(db.disconnect())
+            logger.info(f"[{self.service_name}] ⏳ Disconnecting Database...")
+            await db.disconnect()
 
     async def health_check(self) -> str:
         if not db.is_connected():
@@ -111,6 +121,7 @@ class DatabaseManager(AppService):
 
     # Executions
     get_graph_executions = _(get_graph_executions)
+    get_graph_executions_count = _(get_graph_executions_count)
     get_graph_execution_meta = _(get_graph_execution_meta)
     create_graph_execution = _(create_graph_execution)
     get_node_execution = _(get_node_execution)
@@ -142,15 +153,18 @@ class DatabaseManager(AppService):
 
     # User Comms - async
     get_active_user_ids_in_timerange = _(get_active_user_ids_in_timerange)
+    get_user_by_id = _(get_user_by_id)
     get_user_email_by_id = _(get_user_email_by_id)
     get_user_email_verification = _(get_user_email_verification)
     get_user_notification_preference = _(get_user_notification_preference)
 
     # Notifications - async
+    clear_all_user_notification_batches = _(clear_all_user_notification_batches)
     create_or_add_to_user_notification_batch = _(
         create_or_add_to_user_notification_batch
     )
     empty_user_notification_batch = _(empty_user_notification_batch)
+    remove_notifications_from_batch = _(remove_notifications_from_batch)
     get_all_batches_by_type = _(get_all_batches_by_type)
     get_user_notification_batch = _(get_user_notification_batch)
     get_user_notification_oldest_message_in_batch = _(
@@ -179,6 +193,7 @@ class DatabaseManagerClient(AppServiceClient):
 
     # Executions
     get_graph_executions = _(d.get_graph_executions)
+    get_graph_executions_count = _(d.get_graph_executions_count)
     get_graph_execution_meta = _(d.get_graph_execution_meta)
     get_node_executions = _(d.get_node_executions)
     update_node_execution_status = _(d.update_node_execution_status)
@@ -224,6 +239,7 @@ class DatabaseManagerAsyncClient(AppServiceClient):
     get_node = d.get_node
     get_node_execution = d.get_node_execution
     get_node_executions = d.get_node_executions
+    get_user_by_id = d.get_user_by_id
     get_user_integrations = d.get_user_integrations
     upsert_execution_input = d.upsert_execution_input
     upsert_execution_output = d.upsert_execution_output
@@ -241,10 +257,12 @@ class DatabaseManagerAsyncClient(AppServiceClient):
     get_user_notification_preference = d.get_user_notification_preference
 
     # Notifications
+    clear_all_user_notification_batches = d.clear_all_user_notification_batches
     create_or_add_to_user_notification_batch = (
         d.create_or_add_to_user_notification_batch
     )
     empty_user_notification_batch = d.empty_user_notification_batch
+    remove_notifications_from_batch = d.remove_notifications_from_batch
     get_all_batches_by_type = d.get_all_batches_by_type
     get_user_notification_batch = d.get_user_notification_batch
     get_user_notification_oldest_message_in_batch = (

autogpt_platform/backend/backend/executor/manager.py

@@ -3,16 +3,42 @@ import logging
 import os
 import threading
 import time
+import uuid
 from collections import defaultdict
 from concurrent.futures import Future, ThreadPoolExecutor
 from contextlib import asynccontextmanager
+from datetime import datetime, timedelta, timezone
 from typing import TYPE_CHECKING, Any, Optional, TypeVar, cast
 
+import sentry_sdk
 from pika.adapters.blocking_connection import BlockingChannel
 from pika.spec import Basic, BasicProperties
-from redis.asyncio.lock import Lock as RedisLock
+from prometheus_client import Gauge, start_http_server
+from redis.asyncio.lock import Lock as AsyncRedisLock
 
+from backend.blocks.agent import AgentExecutorBlock
 from backend.blocks.io import AgentOutputBlock
+from backend.data import redis_client as redis
+from backend.data.block import (
+    BlockInput,
+    BlockOutput,
+    BlockOutputEntry,
+    BlockSchema,
+    get_block,
+)
+from backend.data.credit import UsageTransactionMetadata
+from backend.data.dynamic_fields import parse_execution_output
+from backend.data.execution import (
+    ExecutionQueue,
+    ExecutionStatus,
+    GraphExecution,
+    GraphExecutionEntry,
+    NodeExecutionEntry,
+    NodeExecutionResult,
+    NodesInputMasks,
+    UserContext,
+)
+from backend.data.graph import Link, Node
 from backend.data.model import GraphExecutionStats, NodeExecutionStats
 from backend.data.notifications import (
     AgentRunData,
@@ -25,50 +51,21 @@ from backend.data.rabbitmq import SyncRabbitMQ
 from backend.executor.activity_status_generator import (
     generate_activity_status_for_execution,
 )
-from backend.executor.utils import LogMetadata
-from backend.notifications.notifications import queue_notification
-from backend.util.exceptions import InsufficientBalanceError, ModerationError
-
-if TYPE_CHECKING:
-    from backend.executor import DatabaseManagerClient, DatabaseManagerAsyncClient
-
-from prometheus_client import Gauge, start_http_server
-
-from backend.blocks.agent import AgentExecutorBlock
-from backend.data import redis_client as redis
-from backend.data.block import (
-    BlockInput,
-    BlockOutput,
-    BlockOutputEntry,
-    BlockSchema,
-    get_block,
-)
-from backend.data.credit import UsageTransactionMetadata
-from backend.data.execution import (
-    ExecutionQueue,
-    ExecutionStatus,
-    GraphExecution,
-    GraphExecutionEntry,
-    NodeExecutionEntry,
-    NodeExecutionResult,
-    NodesInputMasks,
-    UserContext,
-)
-from backend.data.graph import Link, Node
 from backend.executor.utils import (
     GRACEFUL_SHUTDOWN_TIMEOUT_SECONDS,
     GRAPH_EXECUTION_CANCEL_QUEUE_NAME,
     GRAPH_EXECUTION_QUEUE_NAME,
     CancelExecutionEvent,
     ExecutionOutputEntry,
+    LogMetadata,
     NodeExecutionProgress,
     block_usage_cost,
     create_execution_queue_config,
     execution_usage_cost,
-    parse_execution_output,
     validate_exec,
 )
 from backend.integrations.creds_manager import IntegrationCredentialsManager
+from backend.notifications.notifications import queue_notification
 from backend.server.v2.AutoMod.manager import automod_manager
 from backend.util import json
 from backend.util.clients import (
@@ -84,13 +81,24 @@ from backend.util.decorator import (
     error_logged,
     time_measured,
 )
+from backend.util.exceptions import InsufficientBalanceError, ModerationError
 from backend.util.file import clean_exec_files
 from backend.util.logging import TruncatedLogger, configure_logging
 from backend.util.metrics import DiscordChannel
 from backend.util.process import AppProcess, set_service_name
-from backend.util.retry import continuous_retry, func_retry
+from backend.util.retry import (
+    continuous_retry,
+    func_retry,
+    send_rate_limited_discord_alert,
+)
 from backend.util.settings import Settings
 
+from .cluster_lock import ClusterLock
+
+if TYPE_CHECKING:
+    from backend.executor import DatabaseManagerAsyncClient, DatabaseManagerClient
+
+
 _logger = logging.getLogger(__name__)
 logger = TruncatedLogger(_logger, prefix="[GraphExecutor]")
 settings = Settings()
@@ -106,6 +114,7 @@ utilization_gauge = Gauge(
     "Ratio of active graph runs to max graph workers",
 )
 
+
 # Thread-local storage for ExecutionProcessor instances
 _tls = threading.local()
 
@@ -117,10 +126,14 @@ def init_worker():
 
 
 def execute_graph(
-    graph_exec_entry: "GraphExecutionEntry", cancel_event: threading.Event
+    graph_exec_entry: "GraphExecutionEntry",
+    cancel_event: threading.Event,
+    cluster_lock: ClusterLock,
 ):
     """Execute graph using thread-local ExecutionProcessor instance"""
-    return _tls.processor.on_graph_execution(graph_exec_entry, cancel_event)
+    return _tls.processor.on_graph_execution(
+        graph_exec_entry, cancel_event, cluster_lock
+    )
 
 
 T = TypeVar("T")
@@ -177,6 +190,7 @@ async def execute_node(
         _input_data.inputs = input_data
         if nodes_input_masks:
             _input_data.nodes_input_masks = nodes_input_masks
+        _input_data.user_id = user_id
         input_data = _input_data.model_dump()
     data.inputs = input_data
 
@@ -211,14 +225,37 @@ async def execute_node(
         extra_exec_kwargs[field_name] = credentials
 
     output_size = 0
+
+    # sentry tracking nonsense to get user counts for blocks because isolation scopes don't work :(
+    scope = sentry_sdk.get_current_scope()
+
+    # save the tags
+    original_user = scope._user
+    original_tags = dict(scope._tags) if scope._tags else {}
+    # Set user ID for error tracking
+    scope.set_user({"id": user_id})
+
+    scope.set_tag("graph_id", graph_id)
+    scope.set_tag("node_id", node_id)
+    scope.set_tag("block_name", node_block.name)
+    scope.set_tag("block_id", node_block.id)
+    for k, v in (data.user_context or UserContext(timezone="UTC")).model_dump().items():
+        scope.set_tag(f"user_context.{k}", v)
+
     try:
         async for output_name, output_data in node_block.execute(
             input_data, **extra_exec_kwargs
         ):
-            output_data = json.convert_pydantic_to_json(output_data)
+            output_data = json.to_dict(output_data)
             output_size += len(json.dumps(output_data))
             log_metadata.debug("Node produced output", **{output_name: output_data})
             yield output_name, output_data
+    except Exception:
+        # Capture exception WITH context still set before restoring scope
+        sentry_sdk.capture_exception(scope=scope)
+        sentry_sdk.flush()  # Ensure it's sent before we restore scope
+        # Re-raise to maintain normal error flow
+        raise
     finally:
         # Ensure credentials are released even if execution fails
         if creds_lock and (await creds_lock.locked()) and (await creds_lock.owned()):
@@ -233,6 +270,10 @@ async def execute_node(
             execution_stats.input_size = input_size
             execution_stats.output_size = output_size
 
+        # Restore scope AFTER error has been captured
+        scope._user = original_user
+        scope._tags = original_tags
+
 
 async def _enqueue_next_nodes(
     db_client: "DatabaseManagerAsyncClient",
@@ -429,7 +470,7 @@ class ExecutionProcessor:
             graph_id=node_exec.graph_id,
             node_eid=node_exec.node_exec_id,
             node_id=node_exec.node_id,
-            block_name="-",
+            block_name=b.name if (b := get_block(node_exec.block_id)) else "-",
         )
         db_client = get_db_async_client()
         node = await db_client.get_node(node_exec.node_id)
@@ -557,7 +598,6 @@ class ExecutionProcessor:
                 await persist_output(
                     "error", str(stats.error) or type(stats.error).__name__
                 )
-
         return status
 
     @func_retry
@@ -583,6 +623,7 @@ class ExecutionProcessor:
         self,
         graph_exec: GraphExecutionEntry,
         cancel: threading.Event,
+        cluster_lock: ClusterLock,
     ):
         log_metadata = LogMetadata(
             logger=_logger,
@@ -641,6 +682,7 @@ class ExecutionProcessor:
             cancel=cancel,
             log_metadata=log_metadata,
             execution_stats=exec_stats,
+            cluster_lock=cluster_lock,
         )
         exec_stats.walltime += timing_info.wall_time
         exec_stats.cputime += timing_info.cpu_time
@@ -742,6 +784,7 @@ class ExecutionProcessor:
         cancel: threading.Event,
         log_metadata: LogMetadata,
         execution_stats: GraphExecutionStats,
+        cluster_lock: ClusterLock,
     ) -> ExecutionStatus:
         """
         Returns:
@@ -927,7 +970,7 @@ class ExecutionProcessor:
                         and execution_queue.empty()
                         and (running_node_execution or running_node_evaluation)
                     ):
-                        # There is nothing to execute, and no output to process, let's relax for a while.
+                        cluster_lock.refresh()
                         time.sleep(0.1)
 
             # loop done --------------------------------------------------
@@ -969,16 +1012,31 @@ class ExecutionProcessor:
                 if isinstance(e, Exception)
                 else Exception(f"{e.__class__.__name__}: {e}")
             )
+            if not execution_stats.error:
+                execution_stats.error = str(error)
 
             known_errors = (InsufficientBalanceError, ModerationError)
             if isinstance(error, known_errors):
-                execution_stats.error = str(error)
                 return ExecutionStatus.FAILED
 
             execution_status = ExecutionStatus.FAILED
             log_metadata.exception(
                 f"Failed graph execution {graph_exec.graph_exec_id}: {error}"
             )
+
+            # Send rate-limited Discord alert for unknown/unexpected errors
+            send_rate_limited_discord_alert(
+                "graph_execution",
+                error,
+                "unknown_error",
+                f"🚨 **Unknown Graph Execution Error**\n"
+                f"User: {graph_exec.user_id}\n"
+                f"Graph ID: {graph_exec.graph_id}\n"
+                f"Execution ID: {graph_exec.graph_exec_id}\n"
+                f"Error Type: {type(error).__name__}\n"
+                f"Error: {str(error)[:200]}{'...' if len(str(error)) > 200 else ''}\n",
+            )
+
             raise
 
         finally:
@@ -1153,9 +1211,9 @@ class ExecutionProcessor:
                 f"❌ **Insufficient Funds Alert**\n"
                 f"User: {user_email or user_id}\n"
                 f"Agent: {metadata.name if metadata else 'Unknown Agent'}\n"
-                f"Current balance: ${e.balance/100:.2f}\n"
-                f"Attempted cost: ${abs(e.amount)/100:.2f}\n"
-                f"Shortfall: ${abs(shortfall)/100:.2f}\n"
+                f"Current balance: ${e.balance / 100:.2f}\n"
+                f"Attempted cost: ${abs(e.amount) / 100:.2f}\n"
+                f"Shortfall: ${abs(shortfall) / 100:.2f}\n"
                 f"[View User Details]({base_url}/admin/spending?search={user_email})"
             )
 
@@ -1202,9 +1260,9 @@ class ExecutionProcessor:
                 alert_message = (
                     f"⚠️ **Low Balance Alert**\n"
                     f"User: {user_email or user_id}\n"
-                    f"Balance dropped below ${LOW_BALANCE_THRESHOLD/100:.2f}\n"
-                    f"Current balance: ${current_balance/100:.2f}\n"
-                    f"Transaction cost: ${transaction_cost/100:.2f}\n"
+                    f"Balance dropped below ${LOW_BALANCE_THRESHOLD / 100:.2f}\n"
+                    f"Current balance: ${current_balance / 100:.2f}\n"
+                    f"Transaction cost: ${transaction_cost / 100:.2f}\n"
                     f"[View User Details]({base_url}/admin/spending?search={user_email})"
                 )
                 get_notification_manager_client().discord_system_alert(
@@ -1219,6 +1277,7 @@ class ExecutionManager(AppProcess):
         super().__init__()
         self.pool_size = settings.config.num_graph_workers
         self.active_graph_runs: dict[str, tuple[Future, threading.Event]] = {}
+        self.executor_id = str(uuid.uuid4())
 
         self._executor = None
         self._stop_consuming = None
@@ -1228,6 +1287,8 @@ class ExecutionManager(AppProcess):
         self._run_thread = None
         self._run_client = None
 
+        self._execution_locks = {}
+
     @property
     def cancel_thread(self) -> threading.Thread:
         if self._cancel_thread is None:
@@ -1432,20 +1493,79 @@ class ExecutionManager(AppProcess):
             return
 
         graph_exec_id = graph_exec_entry.graph_exec_id
+        user_id = graph_exec_entry.user_id
+        graph_id = graph_exec_entry.graph_id
         logger.info(
-            f"[{self.service_name}] Received RUN for graph_exec_id={graph_exec_id}"
+            f"[{self.service_name}] Received RUN for graph_exec_id={graph_exec_id}, user_id={user_id}"
         )
-        if graph_exec_id in self.active_graph_runs:
-            # TODO: Make this check cluster-wide, prevent duplicate runs across executor pods.
-            logger.error(
-                f"[{self.service_name}] Graph {graph_exec_id} already running; rejecting duplicate run."
+
+        # Check user rate limit before processing
+        try:
+            # Only check executions from the last 24 hours for performance
+            current_running_count = get_db_client().get_graph_executions_count(
+                user_id=user_id,
+                graph_id=graph_id,
+                statuses=[ExecutionStatus.RUNNING],
+                created_time_gte=datetime.now(timezone.utc) - timedelta(hours=24),
             )
-            _ack_message(reject=True, requeue=False)
+
+            if (
+                current_running_count
+                >= settings.config.max_concurrent_graph_executions_per_user
+            ):
+                logger.warning(
+                    f"[{self.service_name}] Rate limit exceeded for user {user_id} on graph {graph_id}: "
+                    f"{current_running_count}/{settings.config.max_concurrent_graph_executions_per_user} running executions"
+                )
+                _ack_message(reject=True, requeue=True)
+                return
+
+        except Exception as e:
+            logger.error(
+                f"[{self.service_name}] Failed to check rate limit for user {user_id}: {e}, proceeding with execution"
+            )
+            # If rate limit check fails, proceed to avoid blocking executions
+
+        # Check for local duplicate execution first
+        if graph_exec_id in self.active_graph_runs:
+            logger.warning(
+                f"[{self.service_name}] Graph {graph_exec_id} already running locally; rejecting duplicate."
+            )
+            _ack_message(reject=True, requeue=True)
             return
 
+        # Try to acquire cluster-wide execution lock
+        cluster_lock = ClusterLock(
+            redis=redis.get_redis(),
+            key=f"exec_lock:{graph_exec_id}",
+            owner_id=self.executor_id,
+            timeout=settings.config.cluster_lock_timeout,
+        )
+        current_owner = cluster_lock.try_acquire()
+        if current_owner != self.executor_id:
+            # Either someone else has it or Redis is unavailable
+            if current_owner is not None:
+                logger.warning(
+                    f"[{self.service_name}] Graph {graph_exec_id} already running on pod {current_owner}"
+                )
+                _ack_message(reject=True, requeue=False)
+            else:
+                logger.warning(
+                    f"[{self.service_name}] Could not acquire lock for {graph_exec_id} - Redis unavailable"
+                )
+                _ack_message(reject=True, requeue=True)
+            return
+        self._execution_locks[graph_exec_id] = cluster_lock
+
+        logger.info(
+            f"[{self.service_name}] Acquired cluster lock for {graph_exec_id} with executor {self.executor_id}"
+        )
+
         cancel_event = threading.Event()
 
-        future = self.executor.submit(execute_graph, graph_exec_entry, cancel_event)
+        future = self.executor.submit(
+            execute_graph, graph_exec_entry, cancel_event, cluster_lock
+        )
         self.active_graph_runs[graph_exec_id] = (future, cancel_event)
         self._update_prompt_metrics()
 
@@ -1464,6 +1584,10 @@ class ExecutionManager(AppProcess):
                     f"[{self.service_name}] Error in run completion callback: {e}"
                 )
             finally:
+                # Release the cluster-wide execution lock
+                if graph_exec_id in self._execution_locks:
+                    self._execution_locks[graph_exec_id].release()
+                    del self._execution_locks[graph_exec_id]
                 self._cleanup_completed_runs()
 
         future.add_done_callback(_on_run_done)
@@ -1546,6 +1670,10 @@ class ExecutionManager(AppProcess):
                         f"{prefix} ⏳ Still waiting for {len(self.active_graph_runs)} executions: {ids}"
                     )
 
+                    for graph_exec_id in self.active_graph_runs:
+                        if lock := self._execution_locks.get(graph_exec_id):
+                            lock.refresh()
+
                 time.sleep(wait_interval)
                 waited += wait_interval
 
@@ -1563,6 +1691,15 @@ class ExecutionManager(AppProcess):
         except Exception as e:
             logger.error(f"{prefix} ⚠️ Error during executor shutdown: {type(e)} {e}")
 
+        # Release remaining execution locks
+        try:
+            for lock in self._execution_locks.values():
+                lock.release()
+            self._execution_locks.clear()
+            logger.info(f"{prefix} ✅ Released execution locks")
+        except Exception as e:
+            logger.warning(f"{prefix} ⚠️ Failed to release all locks: {e}")
+
         # Disconnect the run execution consumer
         self._stop_message_consumers(
             self.run_thread,
@@ -1577,6 +1714,8 @@ class ExecutionManager(AppProcess):
 
         logger.info(f"{prefix} ✅ Finished GraphExec cleanup")
 
+        super().cleanup()
+
 
 # ------- UTILITIES ------- #
 
@@ -1668,15 +1807,18 @@ def update_graph_execution_state(
 
 
 @asynccontextmanager
-async def synchronized(key: str, timeout: int = 60):
+async def synchronized(key: str, timeout: int = settings.config.cluster_lock_timeout):
     r = await redis.get_redis_async()
-    lock: RedisLock = r.lock(f"lock:{key}", timeout=timeout)
+    lock: AsyncRedisLock = r.lock(f"lock:{key}", timeout=timeout)
     try:
         await lock.acquire()
         yield
     finally:
         if await lock.locked() and await lock.owned():
-            await lock.release()
+            try:
+                await lock.release()
+            except Exception as e:
+                logger.warning(f"Failed to release lock for key {key}: {e}")
 
 
 def increment_execution_count(user_id: str) -> int:

autogpt_platform/backend/backend/executor/scheduler.py

@@ -248,7 +248,7 @@ class Scheduler(AppService):
             raise UnhealthyServiceError("Scheduler is still initializing")
 
         # Check if we're in the middle of cleanup
-        if self.cleaned_up:
+        if self._shutting_down:
             return await super().health_check()
 
         # Normal operation - check if scheduler is running
@@ -375,7 +375,6 @@ class Scheduler(AppService):
         super().run_service()
 
     def cleanup(self):
-        super().cleanup()
         if self.scheduler:
             logger.info("⏳ Shutting down scheduler...")
             self.scheduler.shutdown(wait=True)
@@ -390,7 +389,7 @@ class Scheduler(AppService):
             logger.info("⏳ Waiting for event loop thread to finish...")
             _event_loop_thread.join(timeout=SCHEDULER_OPERATION_TIMEOUT_SECONDS)
 
-        logger.info("Scheduler cleanup complete.")
+        super().cleanup()
 
     @expose
     def add_graph_execution_schedule(

autogpt_platform/backend/backend/executor/utils.py

@@ -4,7 +4,7 @@ import threading
 import time
 from collections import defaultdict
 from concurrent.futures import Future
-from typing import Any, Mapping, Optional, cast
+from typing import Mapping, Optional, cast
 
 from pydantic import BaseModel, JsonValue, ValidationError
 
@@ -20,6 +20,9 @@ from backend.data.block import (
 )
 from backend.data.block_cost_config import BLOCK_COSTS
 from backend.data.db import prisma
+
+# Import dynamic field utilities from centralized location
+from backend.data.dynamic_fields import merge_execution_input
 from backend.data.execution import (
     ExecutionStatus,
     GraphExecutionStats,
@@ -31,6 +34,7 @@ from backend.data.graph import GraphModel, Node
 from backend.data.model import CredentialsMetaInput
 from backend.data.rabbitmq import Exchange, ExchangeType, Queue, RabbitMQConfig
 from backend.data.user import get_user_by_id
+from backend.util.cache import cached
 from backend.util.clients import (
     get_async_execution_event_bus,
     get_async_execution_queue,
@@ -38,12 +42,12 @@ from backend.util.clients import (
     get_integration_credentials_store,
 )
 from backend.util.exceptions import GraphValidationError, NotFoundError
-from backend.util.logging import TruncatedLogger
-from backend.util.mock import MockObject
+from backend.util.logging import TruncatedLogger, is_structured_logging_enabled
 from backend.util.settings import Config
 from backend.util.type import convert
 
 
+@cached(maxsize=1000, ttl_seconds=3600)
 async def get_user_context(user_id: str) -> UserContext:
     """
     Get UserContext for a user, always returns a valid context with timezone.
@@ -51,7 +55,11 @@ async def get_user_context(user_id: str) -> UserContext:
     """
     user_context = UserContext(timezone="UTC")  # Default to UTC
     try:
-        user = await get_user_by_id(user_id)
+        if prisma.is_connected():
+            user = await get_user_by_id(user_id)
+        else:
+            user = await get_database_manager_async_client().get_user_by_id(user_id)
+
         if user and user.timezone and user.timezone != "not-set":
             user_context.timezone = user.timezone
             logger.debug(f"Retrieved user context: timezone={user.timezone}")
@@ -91,7 +99,11 @@ class LogMetadata(TruncatedLogger):
             "node_id": node_id,
             "block_name": block_name,
         }
-        prefix = f"[ExecutionManager|uid:{user_id}|gid:{graph_id}|nid:{node_id}]|geid:{graph_eid}|neid:{node_eid}|{block_name}]"
+        prefix = (
+            "[ExecutionManager]"
+            if is_structured_logging_enabled()
+            else f"[ExecutionManager|uid:{user_id}|gid:{graph_id}|nid:{node_id}]|geid:{graph_eid}|neid:{node_eid}|{block_name}]"  # noqa
+        )
         super().__init__(
             logger,
             max_length=max_length,
@@ -186,195 +198,7 @@ def _is_cost_filter_match(cost_filter: BlockInput, input_data: BlockInput) -> bo
 
 # ============ Execution Input Helpers ============ #
 
-# --------------------------------------------------------------------------- #
-#  Delimiters
-# --------------------------------------------------------------------------- #
-
-LIST_SPLIT = "_$_"
-DICT_SPLIT = "_#_"
-OBJC_SPLIT = "_@_"
-
-_DELIMS = (LIST_SPLIT, DICT_SPLIT, OBJC_SPLIT)
-
-# --------------------------------------------------------------------------- #
-#  Tokenisation utilities
-# --------------------------------------------------------------------------- #
-
-
-def _next_delim(s: str) -> tuple[str | None, int]:
-    """
-    Return the *earliest* delimiter appearing in `s` and its index.
-
-    If none present → (None, -1).
-    """
-    first: str | None = None
-    pos = len(s)  # sentinel: larger than any real index
-    for d in _DELIMS:
-        i = s.find(d)
-        if 0 <= i < pos:
-            first, pos = d, i
-    return first, (pos if first else -1)
-
-
-def _tokenise(path: str) -> list[tuple[str, str]] | None:
-    """
-    Convert the raw path string (starting with a delimiter) into
-    [ (delimiter, identifier), … ] or None if the syntax is malformed.
-    """
-    tokens: list[tuple[str, str]] = []
-    while path:
-        # 1. Which delimiter starts this chunk?
-        delim = next((d for d in _DELIMS if path.startswith(d)), None)
-        if delim is None:
-            return None  # invalid syntax
-
-        # 2. Slice off the delimiter, then up to the next delimiter (or EOS)
-        path = path[len(delim) :]
-        nxt_delim, pos = _next_delim(path)
-        token, path = (
-            path[: pos if pos != -1 else len(path)],
-            path[pos if pos != -1 else len(path) :],
-        )
-        if token == "":
-            return None  # empty identifier is invalid
-        tokens.append((delim, token))
-    return tokens
-
-
-# --------------------------------------------------------------------------- #
-#  Public API – parsing (flattened ➜ concrete)
-# --------------------------------------------------------------------------- #
-
-
-def parse_execution_output(output: BlockOutputEntry, name: str) -> JsonValue | None:
-    """
-    Retrieve a nested value out of `output` using the flattened *name*.
-
-    On any failure (wrong name, wrong type, out-of-range, bad path)
-    returns **None**.
-    """
-    base_name, data = output
-
-    # Exact match → whole object
-    if name == base_name:
-        return data
-
-    # Must start with the expected name
-    if not name.startswith(base_name):
-        return None
-    path = name[len(base_name) :]
-    if not path:
-        return None  # nothing left to parse
-
-    tokens = _tokenise(path)
-    if tokens is None:
-        return None
-
-    cur: JsonValue = data
-    for delim, ident in tokens:
-        if delim == LIST_SPLIT:
-            # list[index]
-            try:
-                idx = int(ident)
-            except ValueError:
-                return None
-            if not isinstance(cur, list) or idx >= len(cur):
-                return None
-            cur = cur[idx]
-
-        elif delim == DICT_SPLIT:
-            if not isinstance(cur, dict) or ident not in cur:
-                return None
-            cur = cur[ident]
-
-        elif delim == OBJC_SPLIT:
-            if not hasattr(cur, ident):
-                return None
-            cur = getattr(cur, ident)
-
-        else:
-            return None  # unreachable
-
-    return cur
-
-
-def _assign(container: Any, tokens: list[tuple[str, str]], value: Any) -> Any:
-    """
-    Recursive helper that *returns* the (possibly new) container with
-    `value` assigned along the remaining `tokens` path.
-    """
-    if not tokens:
-        return value  # leaf reached
-
-    delim, ident = tokens[0]
-    rest = tokens[1:]
-
-    # ---------- list ----------
-    if delim == LIST_SPLIT:
-        try:
-            idx = int(ident)
-        except ValueError:
-            raise ValueError("index must be an integer")
-
-        if container is None:
-            container = []
-        elif not isinstance(container, list):
-            container = list(container) if hasattr(container, "__iter__") else []
-
-        while len(container) <= idx:
-            container.append(None)
-        container[idx] = _assign(container[idx], rest, value)
-        return container
-
-    # ---------- dict ----------
-    if delim == DICT_SPLIT:
-        if container is None:
-            container = {}
-        elif not isinstance(container, dict):
-            container = dict(container) if hasattr(container, "items") else {}
-        container[ident] = _assign(container.get(ident), rest, value)
-        return container
-
-    # ---------- object ----------
-    if delim == OBJC_SPLIT:
-        if container is None or not isinstance(container, MockObject):
-            container = MockObject()
-        setattr(
-            container,
-            ident,
-            _assign(getattr(container, ident, None), rest, value),
-        )
-        return container
-
-    return value  # unreachable
-
-
-def merge_execution_input(data: BlockInput) -> BlockInput:
-    """
-    Reconstruct nested objects from a *flattened* dict of key → value.
-
-    Raises ValueError on syntactically invalid list indices.
-    """
-    merged: BlockInput = {}
-
-    for key, value in data.items():
-        # Split off the base name (before the first delimiter, if any)
-        delim, pos = _next_delim(key)
-        if delim is None:
-            merged[key] = value
-            continue
-
-        base, path = key[:pos], key[pos:]
-        tokens = _tokenise(path)
-        if tokens is None:
-            # Invalid key; treat as scalar under the raw name
-            merged[key] = value
-            continue
-
-        merged[base] = _assign(merged.get(base), tokens, value)
-
-    data.update(merged)
-    return data
+# Dynamic field utilities are now imported from backend.data.dynamic_fields
 
 
 def validate_exec(

autogpt_platform/backend/backend/executor/utils_test.py

@@ -3,7 +3,7 @@ from typing import cast
 import pytest
 from pytest_mock import MockerFixture
 
-from backend.executor.utils import merge_execution_input, parse_execution_output
+from backend.data.dynamic_fields import merge_execution_input, parse_execution_output
 from backend.util.mock import MockObject
 
 

autogpt_platform/backend/backend/integrations/creds_manager.py

@@ -151,7 +151,10 @@ class IntegrationCredentialsManager:
                 fresh_credentials = await oauth_handler.refresh_tokens(credentials)
                 await self.store.update_creds(user_id, fresh_credentials)
                 if _lock and (await _lock.locked()) and (await _lock.owned()):
-                    await _lock.release()
+                    try:
+                        await _lock.release()
+                    except Exception as e:
+                        logger.warning(f"Failed to release OAuth refresh lock: {e}")
 
                 credentials = fresh_credentials
         return credentials
@@ -184,7 +187,10 @@ class IntegrationCredentialsManager:
             yield
         finally:
             if (await lock.locked()) and (await lock.owned()):
-                await lock.release()
+                try:
+                    await lock.release()
+                except Exception as e:
+                    logger.warning(f"Failed to release credentials lock: {e}")
 
     async def release_all_locks(self):
         """Call this on process termination to ensure all locks are released"""

autogpt_platform/backend/backend/integrations/webhooks/__init__.py

@@ -1,6 +1,6 @@
 from typing import TYPE_CHECKING
 
-from autogpt_libs.utils.cache import cached
+from backend.util.cache import cached
 
 if TYPE_CHECKING:
     from ..providers import ProviderName
@@ -8,7 +8,7 @@ if TYPE_CHECKING:
 
 
 # --8<-- [start:load_webhook_managers]
-@cached()
+@cached(ttl_seconds=3600)
 def load_webhook_managers() -> dict["ProviderName", type["BaseWebhooksManager"]]:
     webhook_managers = {}
 

autogpt_platform/backend/backend/notifications/notifications.py

@@ -25,7 +25,11 @@ from backend.data.notifications import (
     get_summary_params_type,
 )
 from backend.data.rabbitmq import Exchange, ExchangeType, Queue, RabbitMQConfig
-from backend.data.user import generate_unsubscribe_link
+from backend.data.user import (
+    disable_all_user_notifications,
+    generate_unsubscribe_link,
+    set_user_email_verification,
+)
 from backend.notifications.email import EmailSender
 from backend.util.clients import get_database_manager_async_client
 from backend.util.logging import TruncatedLogger
@@ -38,7 +42,7 @@ from backend.util.service import (
     endpoint_to_sync,
     expose,
 )
-from backend.util.settings import Settings
+from backend.util.settings import AppEnvironment, Settings
 
 logger = TruncatedLogger(logging.getLogger(__name__), "[NotificationManager]")
 settings = Settings()
@@ -124,6 +128,12 @@ def get_routing_key(event_type: NotificationType) -> str:
 
 def queue_notification(event: NotificationEventModel) -> NotificationResult:
     """Queue a notification - exposed method for other services to call"""
+    # Disable in production
+    if settings.config.app_env == AppEnvironment.PRODUCTION:
+        return NotificationResult(
+            success=True,
+            message="Queueing notifications is disabled in production",
+        )
     try:
         logger.debug(f"Received Request to queue {event=}")
 
@@ -151,6 +161,12 @@ def queue_notification(event: NotificationEventModel) -> NotificationResult:
 
 async def queue_notification_async(event: NotificationEventModel) -> NotificationResult:
     """Queue a notification - exposed method for other services to call"""
+    # Disable in production
+    if settings.config.app_env == AppEnvironment.PRODUCTION:
+        return NotificationResult(
+            success=True,
+            message="Queueing notifications is disabled in production",
+        )
     try:
         logger.debug(f"Received Request to queue {event=}")
 
@@ -213,6 +229,9 @@ class NotificationManager(AppService):
 
     @expose
     async def queue_weekly_summary(self):
+        # disable in prod
+        if settings.config.app_env == AppEnvironment.PRODUCTION:
+            return
         # Use the existing event loop instead of creating a new one with asyncio.run()
         asyncio.create_task(self._queue_weekly_summary())
 
@@ -226,7 +245,9 @@ class NotificationManager(AppService):
             logger.info(
                 f"Querying for active users between {start_time} and {current_time}"
             )
-            users = await get_database_manager_async_client().get_active_user_ids_in_timerange(
+            users = await get_database_manager_async_client(
+                should_retry=False
+            ).get_active_user_ids_in_timerange(
                 end_time=current_time.isoformat(),
                 start_time=start_time.isoformat(),
             )
@@ -253,6 +274,9 @@ class NotificationManager(AppService):
     async def process_existing_batches(
         self, notification_types: list[NotificationType]
     ):
+        # disable in prod
+        if settings.config.app_env == AppEnvironment.PRODUCTION:
+            return
         # Use the existing event loop instead of creating a new process
         asyncio.create_task(self._process_existing_batches(notification_types))
 
@@ -266,15 +290,15 @@ class NotificationManager(AppService):
 
             for notification_type in notification_types:
                 # Get all batches for this notification type
-                batches = (
-                    await get_database_manager_async_client().get_all_batches_by_type(
-                        notification_type
-                    )
-                )
+                batches = await get_database_manager_async_client(
+                    should_retry=False
+                ).get_all_batches_by_type(notification_type)
 
                 for batch in batches:
                     # Check if batch has aged out
-                    oldest_message = await get_database_manager_async_client().get_user_notification_oldest_message_in_batch(
+                    oldest_message = await get_database_manager_async_client(
+                        should_retry=False
+                    ).get_user_notification_oldest_message_in_batch(
                         batch.user_id, notification_type
                     )
 
@@ -289,9 +313,9 @@ class NotificationManager(AppService):
 
                     # If batch has aged out, process it
                     if oldest_message.created_at + max_delay < current_time:
-                        recipient_email = await get_database_manager_async_client().get_user_email_by_id(
-                            batch.user_id
-                        )
+                        recipient_email = await get_database_manager_async_client(
+                            should_retry=False
+                        ).get_user_email_by_id(batch.user_id)
 
                         if not recipient_email:
                             logger.error(
@@ -308,21 +332,25 @@ class NotificationManager(AppService):
                                 f"User {batch.user_id} does not want to receive {notification_type} notifications"
                             )
                             # Clear the batch
-                            await get_database_manager_async_client().empty_user_notification_batch(
+                            await get_database_manager_async_client(
+                                should_retry=False
+                            ).empty_user_notification_batch(
                                 batch.user_id, notification_type
                             )
                             continue
 
-                        batch_data = await get_database_manager_async_client().get_user_notification_batch(
-                            batch.user_id, notification_type
-                        )
+                        batch_data = await get_database_manager_async_client(
+                            should_retry=False
+                        ).get_user_notification_batch(batch.user_id, notification_type)
 
                         if not batch_data or not batch_data.notifications:
                             logger.error(
                                 f"Batch data not found for user {batch.user_id}"
                             )
                             # Clear the batch
-                            await get_database_manager_async_client().empty_user_notification_batch(
+                            await get_database_manager_async_client(
+                                should_retry=False
+                            ).empty_user_notification_batch(
                                 batch.user_id, notification_type
                             )
                             continue
@@ -358,7 +386,9 @@ class NotificationManager(AppService):
                         )
 
                         # Clear the batch
-                        await get_database_manager_async_client().empty_user_notification_batch(
+                        await get_database_manager_async_client(
+                            should_retry=False
+                        ).empty_user_notification_batch(
                             batch.user_id, notification_type
                         )
 
@@ -413,15 +443,13 @@ class NotificationManager(AppService):
         self, user_id: str, event_type: NotificationType
     ) -> bool:
         """Check if a user wants to receive a notification based on their preferences and email verification status"""
-        validated_email = (
-            await get_database_manager_async_client().get_user_email_verification(
-                user_id
-            )
-        )
+        validated_email = await get_database_manager_async_client(
+            should_retry=False
+        ).get_user_email_verification(user_id)
         preference = (
-            await get_database_manager_async_client().get_user_notification_preference(
-                user_id
-            )
+            await get_database_manager_async_client(
+                should_retry=False
+            ).get_user_notification_preference(user_id)
         ).preferences.get(event_type, True)
         # only if both are true, should we email this person
         return validated_email and preference
@@ -437,7 +465,9 @@ class NotificationManager(AppService):
 
         try:
             # Get summary data from the database
-            summary_data = await get_database_manager_async_client().get_user_execution_summary_data(
+            summary_data = await get_database_manager_async_client(
+                should_retry=False
+            ).get_user_execution_summary_data(
                 user_id=user_id,
                 start_time=params.start_date,
                 end_time=params.end_date,
@@ -524,13 +554,13 @@ class NotificationManager(AppService):
         self, user_id: str, event_type: NotificationType, event: NotificationEventModel
     ) -> bool:
 
-        await get_database_manager_async_client().create_or_add_to_user_notification_batch(
-            user_id, event_type, event
-        )
+        await get_database_manager_async_client(
+            should_retry=False
+        ).create_or_add_to_user_notification_batch(user_id, event_type, event)
 
-        oldest_message = await get_database_manager_async_client().get_user_notification_oldest_message_in_batch(
-            user_id, event_type
-        )
+        oldest_message = await get_database_manager_async_client(
+            should_retry=False
+        ).get_user_notification_oldest_message_in_batch(user_id, event_type)
         if not oldest_message:
             logger.error(
                 f"Batch for user {user_id} and type {event_type} has no oldest message whichshould never happen!!!!!!!!!!!!!!!!"
@@ -580,11 +610,9 @@ class NotificationManager(AppService):
                 return False
             logger.debug(f"Processing immediate notification: {event}")
 
-            recipient_email = (
-                await get_database_manager_async_client().get_user_email_by_id(
-                    event.user_id
-                )
-            )
+            recipient_email = await get_database_manager_async_client(
+                should_retry=False
+            ).get_user_email_by_id(event.user_id)
             if not recipient_email:
                 logger.error(f"User email not found for user {event.user_id}")
                 return False
@@ -619,11 +647,9 @@ class NotificationManager(AppService):
                 return False
             logger.info(f"Processing batch notification: {event}")
 
-            recipient_email = (
-                await get_database_manager_async_client().get_user_email_by_id(
-                    event.user_id
-                )
-            )
+            recipient_email = await get_database_manager_async_client(
+                should_retry=False
+            ).get_user_email_by_id(event.user_id)
             if not recipient_email:
                 logger.error(f"User email not found for user {event.user_id}")
                 return False
@@ -642,11 +668,9 @@ class NotificationManager(AppService):
             if not should_send:
                 logger.info("Batch not old enough to send")
                 return False
-            batch = (
-                await get_database_manager_async_client().get_user_notification_batch(
-                    event.user_id, event.type
-                )
-            )
+            batch = await get_database_manager_async_client(
+                should_retry=False
+            ).get_user_notification_batch(event.user_id, event.type)
             if not batch or not batch.notifications:
                 logger.error(f"Batch not found for user {event.user_id}")
                 return False
@@ -657,6 +681,7 @@ class NotificationManager(AppService):
                     get_notif_data_type(db_event.type)
                 ].model_validate(
                     {
+                        "id": db_event.id,  # Include ID from database
                         "user_id": event.user_id,
                         "type": db_event.type,
                         "data": db_event.data,
@@ -679,6 +704,9 @@ class NotificationManager(AppService):
                 chunk_sent = False
                 for attempt_size in [chunk_size, 50, 25, 10, 5, 1]:
                     chunk = batch_messages[i : i + attempt_size]
+                    chunk_ids = [
+                        msg.id for msg in chunk if msg.id
+                    ]  # Extract IDs for removal
 
                     try:
                         # Try to render the email to check its size
@@ -705,6 +733,23 @@ class NotificationManager(AppService):
                                 user_unsub_link=unsub_link,
                             )
 
+                            # Remove successfully sent notifications immediately
+                            if chunk_ids:
+                                try:
+                                    await get_database_manager_async_client(
+                                        should_retry=False
+                                    ).remove_notifications_from_batch(
+                                        event.user_id, event.type, chunk_ids
+                                    )
+                                    logger.info(
+                                        f"Removed {len(chunk_ids)} sent notifications from batch"
+                                    )
+                                except Exception as e:
+                                    logger.error(
+                                        f"Failed to remove sent notifications: {e}"
+                                    )
+                                    # Continue anyway - better to risk duplicates than lose emails
+
                             # Track successful sends
                             successfully_sent_count += len(chunk)
 
@@ -722,13 +767,137 @@ class NotificationManager(AppService):
                             i += len(chunk)
                             chunk_sent = True
                             break
+                        else:
+                            # Message is too large even after size reduction
+                            if attempt_size == 1:
+                                logger.error(
+                                    f"Failed to send notification at index {i}: "
+                                    f"Single notification exceeds email size limit "
+                                    f"({len(test_message):,} chars > {MAX_EMAIL_SIZE:,} chars). "
+                                    f"Removing permanently from batch - will not retry."
+                                )
+
+                                # Remove the oversized notification permanently - it will NEVER fit
+                                if chunk_ids:
+                                    try:
+                                        await get_database_manager_async_client(
+                                            should_retry=False
+                                        ).remove_notifications_from_batch(
+                                            event.user_id, event.type, chunk_ids
+                                        )
+                                        logger.info(
+                                            f"Removed oversized notification {chunk_ids[0]} from batch permanently"
+                                        )
+                                    except Exception as e:
+                                        logger.error(
+                                            f"Failed to remove oversized notification: {e}"
+                                        )
+
+                                failed_indices.append(i)
+                                i += 1
+                                chunk_sent = True
+                                break
+                            # Try smaller chunk size
+                            continue
                     except Exception as e:
+                        # Check if it's a Postmark API error
                         if attempt_size == 1:
-                            # Even single notification is too large
-                            logger.error(
-                                f"Single notification too large to send: {e}. "
-                                f"Skipping notification at index {i}"
-                            )
+                            # Single notification failed - determine the actual cause
+                            error_message = str(e).lower()
+                            error_type = type(e).__name__
+
+                            # Check for HTTP 406 - Inactive recipient (common in Postmark errors)
+                            if "406" in error_message or "inactive" in error_message:
+                                logger.warning(
+                                    f"Failed to send notification at index {i}: "
+                                    f"Recipient marked as inactive by Postmark. "
+                                    f"Error: {e}. Disabling ALL notifications for this user."
+                                )
+
+                                # 1. Mark email as unverified
+                                try:
+                                    await set_user_email_verification(
+                                        event.user_id, False
+                                    )
+                                    logger.info(
+                                        f"Set email verification to false for user {event.user_id}"
+                                    )
+                                except Exception as deactivation_error:
+                                    logger.error(
+                                        f"Failed to deactivate email for user {event.user_id}: "
+                                        f"{deactivation_error}"
+                                    )
+
+                                # 2. Disable all notification preferences
+                                try:
+                                    await disable_all_user_notifications(event.user_id)
+                                    logger.info(
+                                        f"Disabled all notification preferences for user {event.user_id}"
+                                    )
+                                except Exception as disable_error:
+                                    logger.error(
+                                        f"Failed to disable notification preferences: {disable_error}"
+                                    )
+
+                                # 3. Clear ALL notification batches for this user
+                                try:
+                                    await get_database_manager_async_client(
+                                        should_retry=False
+                                    ).clear_all_user_notification_batches(event.user_id)
+                                    logger.info(
+                                        f"Cleared ALL notification batches for user {event.user_id}"
+                                    )
+                                except Exception as remove_error:
+                                    logger.error(
+                                        f"Failed to clear batches for inactive recipient: {remove_error}"
+                                    )
+
+                                # Stop processing - we've nuked everything for this user
+                                return True
+                            # Check for HTTP 422 - Malformed data
+                            elif (
+                                "422" in error_message
+                                or "unprocessable" in error_message
+                            ):
+                                logger.error(
+                                    f"Failed to send notification at index {i}: "
+                                    f"Malformed notification data rejected by Postmark. "
+                                    f"Error: {e}. Removing from batch permanently."
+                                )
+
+                                # Remove from batch - 422 means bad data that won't fix itself
+                                if chunk_ids:
+                                    try:
+                                        await get_database_manager_async_client(
+                                            should_retry=False
+                                        ).remove_notifications_from_batch(
+                                            event.user_id, event.type, chunk_ids
+                                        )
+                                        logger.info(
+                                            "Removed malformed notification from batch permanently"
+                                        )
+                                    except Exception as remove_error:
+                                        logger.error(
+                                            f"Failed to remove malformed notification: {remove_error}"
+                                        )
+                            # Check if it's a ValueError for size limit
+                            elif (
+                                isinstance(e, ValueError)
+                                and "too large" in error_message
+                            ):
+                                logger.error(
+                                    f"Failed to send notification at index {i}: "
+                                    f"Notification size exceeds email limit. "
+                                    f"Error: {e}. Skipping this notification."
+                                )
+                            # Other API errors
+                            else:
+                                logger.error(
+                                    f"Failed to send notification at index {i}: "
+                                    f"Email API error ({error_type}): {e}. "
+                                    f"Skipping this notification."
+                                )
+
                             failed_indices.append(i)
                             i += 1
                             chunk_sent = True
@@ -742,18 +911,20 @@ class NotificationManager(AppService):
                     failed_indices.append(i)
                     i += 1
 
-            # Only empty the batch if ALL notifications were sent successfully
-            if successfully_sent_count == len(batch_messages):
+            # Check what remains in the batch (notifications are removed as sent)
+            remaining_batch = await get_database_manager_async_client(
+                should_retry=False
+            ).get_user_notification_batch(event.user_id, event.type)
+
+            if not remaining_batch or not remaining_batch.notifications:
                 logger.info(
-                    f"Successfully sent all {successfully_sent_count} notifications, clearing batch"
-                )
-                await get_database_manager_async_client().empty_user_notification_batch(
-                    event.user_id, event.type
+                    f"All {successfully_sent_count} notifications sent and removed from batch"
                 )
             else:
+                remaining_count = len(remaining_batch.notifications)
                 logger.warning(
-                    f"Only sent {successfully_sent_count} of {len(batch_messages)} notifications. "
-                    f"Failed indices: {failed_indices}. Batch will be retained for retry."
+                    f"Sent {successfully_sent_count} notifications. "
+                    f"{remaining_count} remain in batch for retry due to errors."
                 )
             return True
         except Exception as e:
@@ -771,11 +942,9 @@ class NotificationManager(AppService):
 
             logger.info(f"Processing summary notification: {model}")
 
-            recipient_email = (
-                await get_database_manager_async_client().get_user_email_by_id(
-                    event.user_id
-                )
-            )
+            recipient_email = await get_database_manager_async_client(
+                should_retry=False
+            ).get_user_email_by_id(event.user_id)
             if not recipient_email:
                 logger.error(f"User email not found for user {event.user_id}")
                 return False
@@ -848,10 +1017,14 @@ class NotificationManager(AppService):
             logger.exception(f"Fatal error in consumer for {queue_name}: {e}")
             raise
 
-    @continuous_retry()
     def run_service(self):
-        self.run_and_wait(self._run_service())
+        # Queue the main _run_service task
+        asyncio.run_coroutine_threadsafe(self._run_service(), self.shared_event_loop)
 
+        # Start the main event loop
+        super().run_service()
+
+    @continuous_retry()
     async def _run_service(self):
         logger.info(f"[{self.service_name}] ⏳ Configuring RabbitMQ...")
         self.rabbitmq_service = rabbitmq.AsyncRabbitMQ(self.rabbitmq_config)
@@ -917,10 +1090,11 @@ class NotificationManager(AppService):
     def cleanup(self):
         """Cleanup service resources"""
         self.running = False
-        super().cleanup()
-        logger.info(f"[{self.service_name}] ⏳ Disconnecting RabbitMQ...")
+        logger.info("⏳ Disconnecting RabbitMQ...")
         self.run_and_wait(self.rabbitmq_service.disconnect())
 
+        super().cleanup()
+
 
 class NotificationManagerClient(AppServiceClient):
     @classmethod

autogpt_platform/backend/backend/notifications/test_notifications.py

@@ -0,0 +1,598 @@
+"""Tests for notification error handling in NotificationManager."""
+
+from datetime import datetime, timezone
+from unittest.mock import AsyncMock, MagicMock, Mock, patch
+
+import pytest
+from prisma.enums import NotificationType
+
+from backend.data.notifications import AgentRunData, NotificationEventModel
+from backend.notifications.notifications import NotificationManager
+
+
+class TestNotificationErrorHandling:
+    """Test cases for notification error handling in NotificationManager."""
+
+    @pytest.fixture
+    def notification_manager(self):
+        """Create a NotificationManager instance for testing."""
+        with patch("backend.notifications.notifications.AppService.__init__"):
+            manager = NotificationManager()
+            manager.email_sender = MagicMock()
+            # Mock the _get_template method used by _process_batch
+            template_mock = Mock()
+            template_mock.base_template = "base"
+            template_mock.subject_template = "subject"
+            template_mock.body_template = "body"
+            manager.email_sender._get_template = Mock(return_value=template_mock)
+            # Mock the formatter
+            manager.email_sender.formatter = Mock()
+            manager.email_sender.formatter.format_email = Mock(
+                return_value=("subject", "body content")
+            )
+            manager.email_sender.formatter.env = Mock()
+            manager.email_sender.formatter.env.globals = {
+                "base_url": "http://example.com"
+            }
+            return manager
+
+    @pytest.fixture
+    def sample_batch_event(self):
+        """Create a sample batch event for testing."""
+        return NotificationEventModel(
+            type=NotificationType.AGENT_RUN,
+            user_id="user_1",
+            created_at=datetime.now(timezone.utc),
+            data=AgentRunData(
+                agent_name="Test Agent",
+                credits_used=10.0,
+                execution_time=5.0,
+                node_count=3,
+                graph_id="graph_1",
+                outputs=[],
+            ),
+        )
+
+    @pytest.fixture
+    def sample_batch_notifications(self):
+        """Create sample batch notifications for testing."""
+        notifications = []
+        for i in range(3):
+            notification = Mock()
+            notification.type = NotificationType.AGENT_RUN
+            notification.data = {
+                "agent_name": f"Test Agent {i}",
+                "credits_used": 10.0 * (i + 1),
+                "execution_time": 5.0 * (i + 1),
+                "node_count": 3 + i,
+                "graph_id": f"graph_{i}",
+                "outputs": [],
+            }
+            notification.created_at = datetime.now(timezone.utc)
+            notifications.append(notification)
+        return notifications
+
+    @pytest.mark.asyncio
+    async def test_406_stops_all_processing_for_user(
+        self, notification_manager, sample_batch_event
+    ):
+        """Test that 406 inactive recipient error stops ALL processing for that user."""
+        with patch("backend.notifications.notifications.logger"), patch(
+            "backend.notifications.notifications.set_user_email_verification",
+            new_callable=AsyncMock,
+        ) as mock_set_verification, patch(
+            "backend.notifications.notifications.disable_all_user_notifications",
+            new_callable=AsyncMock,
+        ) as mock_disable_all, patch(
+            "backend.notifications.notifications.get_database_manager_async_client"
+        ) as mock_db_client, patch(
+            "backend.notifications.notifications.generate_unsubscribe_link"
+        ) as mock_unsub_link:
+
+            # Create batch of 5 notifications
+            notifications = []
+            for i in range(5):
+                notification = Mock()
+                notification.id = f"notif_{i}"
+                notification.type = NotificationType.AGENT_RUN
+                notification.data = {
+                    "agent_name": f"Test Agent {i}",
+                    "credits_used": 10.0 * (i + 1),
+                    "execution_time": 5.0 * (i + 1),
+                    "node_count": 3 + i,
+                    "graph_id": f"graph_{i}",
+                    "outputs": [],
+                }
+                notification.created_at = datetime.now(timezone.utc)
+                notifications.append(notification)
+
+            # Setup mocks
+            mock_db = mock_db_client.return_value
+            mock_db.get_user_email_by_id = AsyncMock(return_value="test@example.com")
+            mock_db.get_user_notification_batch = AsyncMock(
+                return_value=Mock(notifications=notifications)
+            )
+            mock_db.clear_all_user_notification_batches = AsyncMock()
+            mock_db.remove_notifications_from_batch = AsyncMock()
+            mock_unsub_link.return_value = "http://example.com/unsub"
+
+            # Mock internal methods
+            notification_manager._should_email_user_based_on_preference = AsyncMock(
+                return_value=True
+            )
+            notification_manager._should_batch = AsyncMock(return_value=True)
+            notification_manager._parse_message = Mock(return_value=sample_batch_event)
+
+            # Track calls
+            call_count = [0]
+
+            def send_side_effect(*args, **kwargs):
+                data = kwargs.get("data", [])
+                if isinstance(data, list) and len(data) == 1:
+                    current_call = call_count[0]
+                    call_count[0] += 1
+
+                    # First two succeed, third hits 406
+                    if current_call < 2:
+                        return None
+                    else:
+                        raise Exception("Recipient marked as inactive (406)")
+                # Force single processing
+                raise Exception("Force single processing")
+
+            notification_manager.email_sender.send_templated.side_effect = (
+                send_side_effect
+            )
+
+            # Act
+            result = await notification_manager._process_batch(
+                sample_batch_event.model_dump_json()
+            )
+
+            # Assert
+            assert result is True
+
+            # Only 3 calls should have been made (2 successful, 1 failed with 406)
+            assert call_count[0] == 3
+
+            # User should be deactivated
+            mock_set_verification.assert_called_once_with("user_1", False)
+            mock_disable_all.assert_called_once_with("user_1")
+            mock_db.clear_all_user_notification_batches.assert_called_once_with(
+                "user_1"
+            )
+
+            # No further processing should occur after 406
+
+    @pytest.mark.asyncio
+    async def test_422_permanently_removes_malformed_notification(
+        self, notification_manager, sample_batch_event
+    ):
+        """Test that 422 error permanently removes the malformed notification from batch and continues with others."""
+        with patch("backend.notifications.notifications.logger") as mock_logger, patch(
+            "backend.notifications.notifications.get_database_manager_async_client"
+        ) as mock_db_client, patch(
+            "backend.notifications.notifications.generate_unsubscribe_link"
+        ) as mock_unsub_link:
+
+            # Create batch of 5 notifications
+            notifications = []
+            for i in range(5):
+                notification = Mock()
+                notification.id = f"notif_{i}"
+                notification.type = NotificationType.AGENT_RUN
+                notification.data = {
+                    "agent_name": f"Test Agent {i}",
+                    "credits_used": 10.0 * (i + 1),
+                    "execution_time": 5.0 * (i + 1),
+                    "node_count": 3 + i,
+                    "graph_id": f"graph_{i}",
+                    "outputs": [],
+                }
+                notification.created_at = datetime.now(timezone.utc)
+                notifications.append(notification)
+
+            # Setup mocks
+            mock_db = mock_db_client.return_value
+            mock_db.get_user_email_by_id = AsyncMock(return_value="test@example.com")
+            mock_db.get_user_notification_batch = AsyncMock(
+                side_effect=[
+                    Mock(notifications=notifications),
+                    Mock(notifications=[]),  # Empty after processing
+                ]
+            )
+            mock_db.remove_notifications_from_batch = AsyncMock()
+            mock_unsub_link.return_value = "http://example.com/unsub"
+
+            # Mock internal methods
+            notification_manager._should_email_user_based_on_preference = AsyncMock(
+                return_value=True
+            )
+            notification_manager._should_batch = AsyncMock(return_value=True)
+            notification_manager._parse_message = Mock(return_value=sample_batch_event)
+
+            # Track calls
+            call_count = [0]
+            successful_indices = []
+            removed_notification_ids = []
+
+            # Capture what gets removed
+            def remove_side_effect(user_id, notif_type, notif_ids):
+                removed_notification_ids.extend(notif_ids)
+                return None
+
+            mock_db.remove_notifications_from_batch.side_effect = remove_side_effect
+
+            def send_side_effect(*args, **kwargs):
+                data = kwargs.get("data", [])
+                if isinstance(data, list) and len(data) == 1:
+                    current_call = call_count[0]
+                    call_count[0] += 1
+
+                    # Index 2 has malformed data (422)
+                    if current_call == 2:
+                        raise Exception(
+                            "Unprocessable entity (422): Malformed email data"
+                        )
+                    else:
+                        successful_indices.append(current_call)
+                        return None
+                # Force single processing
+                raise Exception("Force single processing")
+
+            notification_manager.email_sender.send_templated.side_effect = (
+                send_side_effect
+            )
+
+            # Act
+            result = await notification_manager._process_batch(
+                sample_batch_event.model_dump_json()
+            )
+
+            # Assert
+            assert result is True
+            assert call_count[0] == 5  # All 5 attempted
+            assert len(successful_indices) == 4  # 4 succeeded (all except index 2)
+            assert 2 not in successful_indices  # Index 2 failed
+
+            # Verify 422 error was logged
+            error_calls = [call[0][0] for call in mock_logger.error.call_args_list]
+            assert any(
+                "422" in call or "malformed" in call.lower() for call in error_calls
+            )
+
+            # Verify all notifications were removed (4 successful + 1 malformed)
+            assert mock_db.remove_notifications_from_batch.call_count == 5
+            assert (
+                "notif_2" in removed_notification_ids
+            )  # Malformed one was removed permanently
+
+    @pytest.mark.asyncio
+    async def test_oversized_notification_permanently_removed(
+        self, notification_manager, sample_batch_event
+    ):
+        """Test that oversized notifications are permanently removed from batch but others continue."""
+        with patch("backend.notifications.notifications.logger") as mock_logger, patch(
+            "backend.notifications.notifications.get_database_manager_async_client"
+        ) as mock_db_client, patch(
+            "backend.notifications.notifications.generate_unsubscribe_link"
+        ) as mock_unsub_link:
+
+            # Create batch of 5 notifications
+            notifications = []
+            for i in range(5):
+                notification = Mock()
+                notification.id = f"notif_{i}"
+                notification.type = NotificationType.AGENT_RUN
+                notification.data = {
+                    "agent_name": f"Test Agent {i}",
+                    "credits_used": 10.0 * (i + 1),
+                    "execution_time": 5.0 * (i + 1),
+                    "node_count": 3 + i,
+                    "graph_id": f"graph_{i}",
+                    "outputs": [],
+                }
+                notification.created_at = datetime.now(timezone.utc)
+                notifications.append(notification)
+
+            # Setup mocks
+            mock_db = mock_db_client.return_value
+            mock_db.get_user_email_by_id = AsyncMock(return_value="test@example.com")
+            mock_db.get_user_notification_batch = AsyncMock(
+                side_effect=[
+                    Mock(notifications=notifications),
+                    Mock(notifications=[]),  # Empty after processing
+                ]
+            )
+            mock_db.remove_notifications_from_batch = AsyncMock()
+            mock_unsub_link.return_value = "http://example.com/unsub"
+
+            # Mock internal methods
+            notification_manager._should_email_user_based_on_preference = AsyncMock(
+                return_value=True
+            )
+            notification_manager._should_batch = AsyncMock(return_value=True)
+            notification_manager._parse_message = Mock(return_value=sample_batch_event)
+
+            # Override formatter to simulate oversized on index 3
+            # original_format = notification_manager.email_sender.formatter.format_email
+
+            def format_side_effect(*args, **kwargs):
+                # Check if we're formatting index 3
+                data = kwargs.get("data", {}).get("notifications", [])
+                if data and len(data) == 1:
+                    # Check notification content to identify index 3
+                    if any(
+                        "Test Agent 3" in str(n.data)
+                        for n in data
+                        if hasattr(n, "data")
+                    ):
+                        # Return oversized message for index 3
+                        return ("subject", "x" * 5_000_000)  # Over 4.5MB limit
+                return ("subject", "normal sized content")
+
+            notification_manager.email_sender.formatter.format_email = Mock(
+                side_effect=format_side_effect
+            )
+
+            # Track calls
+            successful_indices = []
+
+            def send_side_effect(*args, **kwargs):
+                data = kwargs.get("data", [])
+                if isinstance(data, list) and len(data) == 1:
+                    # Track which notification was sent based on content
+                    for i, notif in enumerate(notifications):
+                        if any(
+                            f"Test Agent {i}" in str(n.data)
+                            for n in data
+                            if hasattr(n, "data")
+                        ):
+                            successful_indices.append(i)
+                            return None
+                    return None
+                # Force single processing
+                raise Exception("Force single processing")
+
+            notification_manager.email_sender.send_templated.side_effect = (
+                send_side_effect
+            )
+
+            # Act
+            result = await notification_manager._process_batch(
+                sample_batch_event.model_dump_json()
+            )
+
+            # Assert
+            assert result is True
+            assert (
+                len(successful_indices) == 4
+            )  # Only 4 sent (index 3 skipped due to size)
+            assert 3 not in successful_indices  # Index 3 was not sent
+
+            # Verify oversized error was logged
+            error_calls = [call[0][0] for call in mock_logger.error.call_args_list]
+            assert any(
+                "exceeds email size limit" in call or "oversized" in call.lower()
+                for call in error_calls
+            )
+
+    @pytest.mark.asyncio
+    async def test_generic_api_error_keeps_notification_for_retry(
+        self, notification_manager, sample_batch_event
+    ):
+        """Test that generic API errors keep notifications in batch for retry while others continue."""
+        with patch("backend.notifications.notifications.logger") as mock_logger, patch(
+            "backend.notifications.notifications.get_database_manager_async_client"
+        ) as mock_db_client, patch(
+            "backend.notifications.notifications.generate_unsubscribe_link"
+        ) as mock_unsub_link:
+
+            # Create batch of 5 notifications
+            notifications = []
+            for i in range(5):
+                notification = Mock()
+                notification.id = f"notif_{i}"
+                notification.type = NotificationType.AGENT_RUN
+                notification.data = {
+                    "agent_name": f"Test Agent {i}",
+                    "credits_used": 10.0 * (i + 1),
+                    "execution_time": 5.0 * (i + 1),
+                    "node_count": 3 + i,
+                    "graph_id": f"graph_{i}",
+                    "outputs": [],
+                }
+                notification.created_at = datetime.now(timezone.utc)
+                notifications.append(notification)
+
+            # Notification that failed with generic error
+            failed_notifications = [notifications[1]]  # Only index 1 remains for retry
+
+            # Setup mocks
+            mock_db = mock_db_client.return_value
+            mock_db.get_user_email_by_id = AsyncMock(return_value="test@example.com")
+            mock_db.get_user_notification_batch = AsyncMock(
+                side_effect=[
+                    Mock(notifications=notifications),
+                    Mock(
+                        notifications=failed_notifications
+                    ),  # Failed ones remain for retry
+                ]
+            )
+            mock_db.remove_notifications_from_batch = AsyncMock()
+            mock_unsub_link.return_value = "http://example.com/unsub"
+
+            # Mock internal methods
+            notification_manager._should_email_user_based_on_preference = AsyncMock(
+                return_value=True
+            )
+            notification_manager._should_batch = AsyncMock(return_value=True)
+            notification_manager._parse_message = Mock(return_value=sample_batch_event)
+
+            # Track calls
+            successful_indices = []
+            failed_indices = []
+            removed_notification_ids = []
+
+            # Capture what gets removed
+            def remove_side_effect(user_id, notif_type, notif_ids):
+                removed_notification_ids.extend(notif_ids)
+                return None
+
+            mock_db.remove_notifications_from_batch.side_effect = remove_side_effect
+
+            def send_side_effect(*args, **kwargs):
+                data = kwargs.get("data", [])
+                if isinstance(data, list) and len(data) == 1:
+                    # Track which notification based on content
+                    for i, notif in enumerate(notifications):
+                        if any(
+                            f"Test Agent {i}" in str(n.data)
+                            for n in data
+                            if hasattr(n, "data")
+                        ):
+                            # Index 1 has generic API error
+                            if i == 1:
+                                failed_indices.append(i)
+                                raise Exception("Network timeout - temporary failure")
+                            else:
+                                successful_indices.append(i)
+                                return None
+                    return None
+                # Force single processing
+                raise Exception("Force single processing")
+
+            notification_manager.email_sender.send_templated.side_effect = (
+                send_side_effect
+            )
+
+            # Act
+            result = await notification_manager._process_batch(
+                sample_batch_event.model_dump_json()
+            )
+
+            # Assert
+            assert result is True
+            assert len(successful_indices) == 4  # 4 succeeded (0, 2, 3, 4)
+            assert len(failed_indices) == 1  # 1 failed
+            assert 1 in failed_indices  # Index 1 failed
+
+            # Verify generic error was logged
+            error_calls = [call[0][0] for call in mock_logger.error.call_args_list]
+            assert any(
+                "api error" in call.lower() or "skipping" in call.lower()
+                for call in error_calls
+            )
+
+            # Only successful ones should be removed from batch (failed one stays for retry)
+            assert mock_db.remove_notifications_from_batch.call_count == 4
+            assert (
+                "notif_1" not in removed_notification_ids
+            )  # Failed one NOT removed (stays for retry)
+            assert "notif_0" in removed_notification_ids  # Successful one removed
+            assert "notif_2" in removed_notification_ids  # Successful one removed
+            assert "notif_3" in removed_notification_ids  # Successful one removed
+            assert "notif_4" in removed_notification_ids  # Successful one removed
+
+    @pytest.mark.asyncio
+    async def test_batch_all_notifications_sent_successfully(
+        self, notification_manager, sample_batch_event
+    ):
+        """Test successful batch processing where all notifications are sent without errors."""
+        with patch("backend.notifications.notifications.logger") as mock_logger, patch(
+            "backend.notifications.notifications.get_database_manager_async_client"
+        ) as mock_db_client, patch(
+            "backend.notifications.notifications.generate_unsubscribe_link"
+        ) as mock_unsub_link:
+
+            # Create batch of 5 notifications
+            notifications = []
+            for i in range(5):
+                notification = Mock()
+                notification.id = f"notif_{i}"
+                notification.type = NotificationType.AGENT_RUN
+                notification.data = {
+                    "agent_name": f"Test Agent {i}",
+                    "credits_used": 10.0 * (i + 1),
+                    "execution_time": 5.0 * (i + 1),
+                    "node_count": 3 + i,
+                    "graph_id": f"graph_{i}",
+                    "outputs": [],
+                }
+                notification.created_at = datetime.now(timezone.utc)
+                notifications.append(notification)
+
+            # Setup mocks
+            mock_db = mock_db_client.return_value
+            mock_db.get_user_email_by_id = AsyncMock(return_value="test@example.com")
+            mock_db.get_user_notification_batch = AsyncMock(
+                side_effect=[
+                    Mock(notifications=notifications),
+                    Mock(notifications=[]),  # Empty after all sent successfully
+                ]
+            )
+            mock_db.remove_notifications_from_batch = AsyncMock()
+            mock_unsub_link.return_value = "http://example.com/unsub"
+
+            # Mock internal methods
+            notification_manager._should_email_user_based_on_preference = AsyncMock(
+                return_value=True
+            )
+            notification_manager._should_batch = AsyncMock(return_value=True)
+            notification_manager._parse_message = Mock(return_value=sample_batch_event)
+
+            # Track successful sends
+            successful_indices = []
+            removed_notification_ids = []
+
+            # Capture what gets removed
+            def remove_side_effect(user_id, notif_type, notif_ids):
+                removed_notification_ids.extend(notif_ids)
+                return None
+
+            mock_db.remove_notifications_from_batch.side_effect = remove_side_effect
+
+            def send_side_effect(*args, **kwargs):
+                data = kwargs.get("data", [])
+                if isinstance(data, list) and len(data) == 1:
+                    # Track which notification was sent
+                    for i, notif in enumerate(notifications):
+                        if any(
+                            f"Test Agent {i}" in str(n.data)
+                            for n in data
+                            if hasattr(n, "data")
+                        ):
+                            successful_indices.append(i)
+                            return None
+                    return None  # Success
+                # Force single processing
+                raise Exception("Force single processing")
+
+            notification_manager.email_sender.send_templated.side_effect = (
+                send_side_effect
+            )
+
+            # Act
+            result = await notification_manager._process_batch(
+                sample_batch_event.model_dump_json()
+            )
+
+            # Assert
+            assert result is True
+
+            # All 5 notifications should be sent successfully
+            assert len(successful_indices) == 5
+            assert successful_indices == [0, 1, 2, 3, 4]
+
+            # All notifications should be removed from batch
+            assert mock_db.remove_notifications_from_batch.call_count == 5
+            assert len(removed_notification_ids) == 5
+            for i in range(5):
+                assert f"notif_{i}" in removed_notification_ids
+
+            # No errors should be logged
+            assert mock_logger.error.call_count == 0
+
+            # Info message about successful sends should be logged
+            info_calls = [call[0][0] for call in mock_logger.info.call_args_list]
+            assert any("sent and removed" in call.lower() for call in info_calls)

autogpt_platform/backend/backend/server/conftest.py

@@ -14,19 +14,49 @@ def configured_snapshot(snapshot: Snapshot) -> Snapshot:
 @pytest.fixture
 def test_user_id() -> str:
     """Test user ID fixture."""
-    return "test-user-id"
+    return "3e53486c-cf57-477e-ba2a-cb02dc828e1a"
 
 
 @pytest.fixture
 def admin_user_id() -> str:
     """Admin user ID fixture."""
-    return "admin-user-id"
+    return "4e53486c-cf57-477e-ba2a-cb02dc828e1b"
 
 
 @pytest.fixture
 def target_user_id() -> str:
     """Target user ID fixture."""
-    return "target-user-id"
+    return "5e53486c-cf57-477e-ba2a-cb02dc828e1c"
+
+
+@pytest.fixture
+async def setup_test_user(test_user_id):
+    """Create test user in database before tests."""
+    from backend.data.user import get_or_create_user
+
+    # Create the test user in the database using JWT token format
+    user_data = {
+        "sub": test_user_id,
+        "email": "test@example.com",
+        "user_metadata": {"name": "Test User"},
+    }
+    await get_or_create_user(user_data)
+    return test_user_id
+
+
+@pytest.fixture
+async def setup_admin_user(admin_user_id):
+    """Create admin user in database before tests."""
+    from backend.data.user import get_or_create_user
+
+    # Create the admin user in the database using JWT token format
+    user_data = {
+        "sub": admin_user_id,
+        "email": "test-admin@example.com",
+        "user_metadata": {"name": "Test Admin"},
+    }
+    await get_or_create_user(user_data)
+    return admin_user_id
 
 
 @pytest.fixture

autogpt_platform/backend/backend/server/integrations/router.py

@@ -32,6 +32,7 @@ from backend.data.model import (
     OAuth2Credentials,
     UserIntegrations,
 )
+from backend.data.onboarding import complete_webhook_trigger_step
 from backend.data.user import get_user_integrations
 from backend.executor.utils import add_graph_execution
 from backend.integrations.ayrshare import AyrshareClient, SocialPlatform
@@ -63,7 +64,7 @@ class LoginResponse(BaseModel):
     state_token: str
 
 
-@router.get("/{provider}/login")
+@router.get("/{provider}/login", summary="Initiate OAuth flow")
 async def login(
     provider: Annotated[
         ProviderName, Path(title="The provider to initiate an OAuth flow for")
@@ -101,7 +102,7 @@ class CredentialsMetaResponse(BaseModel):
     )
 
 
-@router.post("/{provider}/callback")
+@router.post("/{provider}/callback", summary="Exchange OAuth code for tokens")
 async def callback(
     provider: Annotated[
         ProviderName, Path(title="The target provider for this OAuth exchange")
@@ -179,7 +180,7 @@ async def callback(
     )
 
 
-@router.get("/credentials")
+@router.get("/credentials", summary="List Credentials")
 async def list_credentials(
     user_id: Annotated[str, Security(get_user_id)],
 ) -> list[CredentialsMetaResponse]:
@@ -220,7 +221,9 @@ async def list_credentials_by_provider(
     ]
 
 
-@router.get("/{provider}/credentials/{cred_id}")
+@router.get(
+    "/{provider}/credentials/{cred_id}", summary="Get Specific Credential By ID"
+)
 async def get_credential(
     provider: Annotated[
         ProviderName, Path(title="The provider to retrieve credentials for")
@@ -241,7 +244,7 @@ async def get_credential(
     return credential
 
 
-@router.post("/{provider}/credentials", status_code=201)
+@router.post("/{provider}/credentials", status_code=201, summary="Create Credentials")
 async def create_credentials(
     user_id: Annotated[str, Security(get_user_id)],
     provider: Annotated[
@@ -367,6 +370,8 @@ async def webhook_ingress_generic(
         return
 
     executions: list[Awaitable] = []
+    await complete_webhook_trigger_step(user_id)
+
     for node in webhook.triggered_nodes:
         logger.debug(f"Webhook-attached node: {node}")
         if not node.is_triggered_by_event_type(event_type):

autogpt_platform/backend/backend/server/middleware/security.py

@@ -1,12 +1,10 @@
 import re
 from typing import Set
 
-from fastapi import Request, Response
-from starlette.middleware.base import BaseHTTPMiddleware
-from starlette.types import ASGIApp
+from starlette.types import ASGIApp, Message, Receive, Scope, Send
 
 
-class SecurityHeadersMiddleware(BaseHTTPMiddleware):
+class SecurityHeadersMiddleware:
     """
     Middleware to add security headers to responses, with cache control
     disabled by default for all endpoints except those explicitly allowed.
@@ -25,6 +23,8 @@ class SecurityHeadersMiddleware(BaseHTTPMiddleware):
         "/api/health",
         "/api/v1/health",
         "/api/status",
+        "/api/blocks",
+        "/api/v1/blocks",
         # Public store/marketplace pages (read-only)
         "/api/store/agents",
         "/api/v1/store/agents",
@@ -49,7 +49,7 @@ class SecurityHeadersMiddleware(BaseHTTPMiddleware):
     }
 
     def __init__(self, app: ASGIApp):
-        super().__init__(app)
+        self.app = app
         # Compile regex patterns for wildcard matching
         self.cacheable_patterns = [
             re.compile(pattern.replace("*", "[^/]+"))
@@ -72,26 +72,42 @@ class SecurityHeadersMiddleware(BaseHTTPMiddleware):
 
         return False
 
-    async def dispatch(self, request: Request, call_next):
-        response: Response = await call_next(request)
+    async def __call__(self, scope: Scope, receive: Receive, send: Send) -> None:
+        """Pure ASGI middleware implementation for better performance than BaseHTTPMiddleware."""
+        if scope["type"] != "http":
+            await self.app(scope, receive, send)
+            return
 
-        # Add general security headers
-        response.headers["X-Content-Type-Options"] = "nosniff"
-        response.headers["X-Frame-Options"] = "DENY"
-        response.headers["X-XSS-Protection"] = "1; mode=block"
-        response.headers["Referrer-Policy"] = "strict-origin-when-cross-origin"
+        # Extract path from scope
+        path = scope["path"]
 
-        # Add noindex header for shared execution pages
-        if "/public/shared" in request.url.path:
-            response.headers["X-Robots-Tag"] = "noindex, nofollow"
+        async def send_wrapper(message: Message) -> None:
+            if message["type"] == "http.response.start":
+                # Add security headers to the response
+                headers = dict(message.get("headers", []))
 
-        # Default: Disable caching for all endpoints
-        # Only allow caching for explicitly permitted paths
-        if not self.is_cacheable_path(request.url.path):
-            response.headers["Cache-Control"] = (
-                "no-store, no-cache, must-revalidate, private"
-            )
-            response.headers["Pragma"] = "no-cache"
-            response.headers["Expires"] = "0"
+                # Add general security headers (HTTP spec requires proper capitalization)
+                headers[b"X-Content-Type-Options"] = b"nosniff"
+                headers[b"X-Frame-Options"] = b"DENY"
+                headers[b"X-XSS-Protection"] = b"1; mode=block"
+                headers[b"Referrer-Policy"] = b"strict-origin-when-cross-origin"
 
-        return response
+                # Add noindex header for shared execution pages
+                if "/public/shared" in path:
+                    headers[b"X-Robots-Tag"] = b"noindex, nofollow"
+
+                # Default: Disable caching for all endpoints
+                # Only allow caching for explicitly permitted paths
+                if not self.is_cacheable_path(path):
+                    headers[b"Cache-Control"] = (
+                        b"no-store, no-cache, must-revalidate, private"
+                    )
+                    headers[b"Pragma"] = b"no-cache"
+                    headers[b"Expires"] = b"0"
+
+                # Convert headers back to list format
+                message["headers"] = list(headers.items())
+
+            await send(message)
+
+        await self.app(scope, receive, send_wrapper)

autogpt_platform/backend/backend/server/rest_api.py

@@ -1,5 +1,6 @@
 import contextlib
 import logging
+import platform
 from enum import Enum
 from typing import Any, Optional
 
@@ -11,6 +12,7 @@ import uvicorn
 from autogpt_libs.auth import add_auth_responses_to_openapi
 from autogpt_libs.auth import verify_settings as verify_auth_settings
 from fastapi.exceptions import RequestValidationError
+from fastapi.middleware.gzip import GZipMiddleware
 from fastapi.routing import APIRoute
 from prisma.errors import PrismaError
 
@@ -70,6 +72,26 @@ async def lifespan_context(app: fastapi.FastAPI):
 
     await backend.data.db.connect()
 
+    # Configure thread pool for FastAPI sync operation performance
+    # CRITICAL: FastAPI automatically runs ALL sync functions in this thread pool:
+    # - Any endpoint defined with 'def' (not async def)
+    # - Any dependency function defined with 'def' (not async def)
+    # - Manual run_in_threadpool() calls (like JWT decoding)
+    # Default pool size is only 40 threads, causing bottlenecks under high concurrency
+    config = backend.util.settings.Config()
+    try:
+        import anyio.to_thread
+
+        anyio.to_thread.current_default_thread_limiter().total_tokens = (
+            config.fastapi_thread_pool_size
+        )
+        logger.info(
+            f"Thread pool size set to {config.fastapi_thread_pool_size} for sync endpoint/dependency performance"
+        )
+    except (ImportError, AttributeError) as e:
+        logger.warning(f"Could not configure thread pool size: {e}")
+        # Continue without thread pool configuration
+
     # Ensure SDK auto-registration is patched before initializing blocks
     from backend.sdk.registry import AutoRegistry
 
@@ -140,6 +162,9 @@ app = fastapi.FastAPI(
 
 app.add_middleware(SecurityHeadersMiddleware)
 
+# Add GZip compression middleware for large responses (like /api/blocks)
+app.add_middleware(GZipMiddleware, minimum_size=50_000)  # 50KB threshold
+
 # Add 401 responses to authenticated endpoints in OpenAPI spec
 add_auth_responses_to_openapi(app)
 
@@ -273,16 +298,28 @@ class AgentServer(backend.util.service.AppProcess):
             allow_methods=["*"],  # Allows all methods
             allow_headers=["*"],  # Allows all headers
         )
-        uvicorn.run(
-            server_app,
-            host=backend.util.settings.Config().agent_api_host,
-            port=backend.util.settings.Config().agent_api_port,
-            log_config=None,
-        )
+        config = backend.util.settings.Config()
 
-    def cleanup(self):
-        super().cleanup()
-        logger.info(f"[{self.service_name}] ⏳ Shutting down Agent Server...")
+        # Configure uvicorn with performance optimizations from Kludex FastAPI tips
+        uvicorn_config = {
+            "app": server_app,
+            "host": config.agent_api_host,
+            "port": config.agent_api_port,
+            "log_config": None,
+            # Use httptools for HTTP parsing (if available)
+            "http": "httptools",
+            # Only use uvloop on Unix-like systems (not supported on Windows)
+            "loop": "uvloop" if platform.system() != "Windows" else "auto",
+        }
+
+        # Only add debug in local environment (not supported in all uvicorn versions)
+        if config.app_env == backend.util.settings.AppEnvironment.LOCAL:
+            import os
+
+            # Enable asyncio debug mode via environment variable
+            os.environ["PYTHONASYNCIODEBUG"] = "1"
+
+        uvicorn.run(**uvicorn_config)
 
     @staticmethod
     async def test_execute_graph(

autogpt_platform/backend/backend/server/routers/v1.py

@@ -11,7 +11,6 @@ import pydantic
 import stripe
 from autogpt_libs.auth import get_user_id, requires_user
 from autogpt_libs.auth.jwt_utils import get_jwt_payload
-from autogpt_libs.utils.cache import cached
 from fastapi import (
     APIRouter,
     Body,
@@ -24,6 +23,8 @@ from fastapi import (
     Security,
     UploadFile,
 )
+from fastapi.concurrency import run_in_threadpool
+from pydantic import BaseModel
 from starlette.status import HTTP_204_NO_CONTENT, HTTP_404_NOT_FOUND
 from typing_extensions import Optional, TypedDict
 
@@ -38,6 +39,7 @@ from backend.data.credit import (
     AutoTopUpConfig,
     RefundRequest,
     TransactionHistory,
+    UserCredit,
     get_auto_top_up,
     get_user_credit_model,
     set_auto_top_up,
@@ -82,9 +84,11 @@ from backend.server.model import (
     UpdateTimezoneRequest,
     UploadFileResponse,
 )
+from backend.util.cache import cached
 from backend.util.clients import get_scheduler_client
 from backend.util.cloud_storage import get_cloud_storage_handler
 from backend.util.exceptions import GraphValidationError, NotFoundError
+from backend.util.json import dumps
 from backend.util.settings import Settings
 from backend.util.timezone_utils import (
     convert_utc_time_to_user_timezone,
@@ -104,9 +108,6 @@ def _create_file_size_error(size_bytes: int, max_size_mb: int) -> HTTPException:
 settings = Settings()
 logger = logging.getLogger(__name__)
 
-
-_user_credit_model = get_user_credit_model()
-
 # Define the API routes
 v1_router = APIRouter()
 
@@ -263,13 +264,10 @@ async def is_onboarding_enabled():
 ########################################################
 
 
-@cached()
-def _get_cached_blocks() -> Sequence[dict[Any, Any]]:
+def _compute_blocks_sync() -> str:
     """
-    Get cached blocks with thundering herd protection.
-
-    Uses sync_cache decorator to prevent multiple concurrent requests
-    from all executing the expensive block loading operation.
+    Synchronous function to compute blocks data.
+    This does the heavy lifting: instantiate 226+ blocks, compute costs, serialize.
     """
     from backend.data.credit import get_block_cost
 
@@ -279,11 +277,27 @@ def _get_cached_blocks() -> Sequence[dict[Any, Any]]:
     for block_class in block_classes.values():
         block_instance = block_class()
         if not block_instance.disabled:
-            # Get costs for this specific block class without creating another instance
             costs = get_block_cost(block_instance)
-            result.append({**block_instance.to_dict(), "costs": costs})
+            # Convert BlockCost BaseModel objects to dictionaries for JSON serialization
+            costs_dict = [
+                cost.model_dump() if isinstance(cost, BaseModel) else cost
+                for cost in costs
+            ]
+            result.append({**block_instance.to_dict(), "costs": costs_dict})
 
-    return result
+    # Use our JSON utility which properly handles complex types through to_dict conversion
+    return dumps(result)
+
+
+@cached(ttl_seconds=3600)
+async def _get_cached_blocks() -> str:
+    """
+    Async cached function with thundering herd protection.
+    On cache miss: runs heavy work in thread pool
+    On cache hit: returns cached string immediately (no thread pool needed)
+    """
+    # Only run in thread pool on cache miss - cache hits return immediately
+    return await run_in_threadpool(_compute_blocks_sync)
 
 
 @v1_router.get(
@@ -291,9 +305,28 @@ def _get_cached_blocks() -> Sequence[dict[Any, Any]]:
     summary="List available blocks",
     tags=["blocks"],
     dependencies=[Security(requires_user)],
+    responses={
+        200: {
+            "description": "Successful Response",
+            "content": {
+                "application/json": {
+                    "schema": {
+                        "items": {"additionalProperties": True, "type": "object"},
+                        "type": "array",
+                        "title": "Response Getv1List Available Blocks",
+                    }
+                }
+            },
+        }
+    },
 )
-async def get_graph_blocks() -> Sequence[dict[Any, Any]]:
-    return _get_cached_blocks()
+async def get_graph_blocks() -> Response:
+    # Cache hit: returns immediately, Cache miss: runs in thread pool
+    content = await _get_cached_blocks()
+    return Response(
+        content=content,
+        media_type="application/json",
+    )
 
 
 @v1_router.post(
@@ -443,7 +476,8 @@ async def upload_file(
 async def get_user_credits(
     user_id: Annotated[str, Security(get_user_id)],
 ) -> dict[str, int]:
-    return {"credits": await _user_credit_model.get_credits(user_id)}
+    user_credit_model = await get_user_credit_model(user_id)
+    return {"credits": await user_credit_model.get_credits(user_id)}
 
 
 @v1_router.post(
@@ -455,9 +489,8 @@ async def get_user_credits(
 async def request_top_up(
     request: RequestTopUp, user_id: Annotated[str, Security(get_user_id)]
 ):
-    checkout_url = await _user_credit_model.top_up_intent(
-        user_id, request.credit_amount
-    )
+    user_credit_model = await get_user_credit_model(user_id)
+    checkout_url = await user_credit_model.top_up_intent(user_id, request.credit_amount)
     return {"checkout_url": checkout_url}
 
 
@@ -472,7 +505,8 @@ async def refund_top_up(
     transaction_key: str,
     metadata: dict[str, str],
 ) -> int:
-    return await _user_credit_model.top_up_refund(user_id, transaction_key, metadata)
+    user_credit_model = await get_user_credit_model(user_id)
+    return await user_credit_model.top_up_refund(user_id, transaction_key, metadata)
 
 
 @v1_router.patch(
@@ -482,7 +516,8 @@ async def refund_top_up(
     dependencies=[Security(requires_user)],
 )
 async def fulfill_checkout(user_id: Annotated[str, Security(get_user_id)]):
-    await _user_credit_model.fulfill_checkout(user_id=user_id)
+    user_credit_model = await get_user_credit_model(user_id)
+    await user_credit_model.fulfill_checkout(user_id=user_id)
     return Response(status_code=200)
 
 
@@ -496,18 +531,23 @@ async def configure_user_auto_top_up(
     request: AutoTopUpConfig, user_id: Annotated[str, Security(get_user_id)]
 ) -> str:
     if request.threshold < 0:
-        raise ValueError("Threshold must be greater than 0")
+        raise HTTPException(status_code=422, detail="Threshold must be greater than 0")
     if request.amount < 500 and request.amount != 0:
-        raise ValueError("Amount must be greater than or equal to 500")
-    if request.amount < request.threshold:
-        raise ValueError("Amount must be greater than or equal to threshold")
+        raise HTTPException(
+            status_code=422, detail="Amount must be greater than or equal to 500"
+        )
+    if request.amount != 0 and request.amount < request.threshold:
+        raise HTTPException(
+            status_code=422, detail="Amount must be greater than or equal to threshold"
+        )
 
-    current_balance = await _user_credit_model.get_credits(user_id)
+    user_credit_model = await get_user_credit_model(user_id)
+    current_balance = await user_credit_model.get_credits(user_id)
 
     if current_balance < request.threshold:
-        await _user_credit_model.top_up_credits(user_id, request.amount)
+        await user_credit_model.top_up_credits(user_id, request.amount)
     else:
-        await _user_credit_model.top_up_credits(user_id, 0)
+        await user_credit_model.top_up_credits(user_id, 0)
 
     await set_auto_top_up(
         user_id, AutoTopUpConfig(threshold=request.threshold, amount=request.amount)
@@ -555,15 +595,13 @@ async def stripe_webhook(request: Request):
         event["type"] == "checkout.session.completed"
         or event["type"] == "checkout.session.async_payment_succeeded"
     ):
-        await _user_credit_model.fulfill_checkout(
-            session_id=event["data"]["object"]["id"]
-        )
+        await UserCredit().fulfill_checkout(session_id=event["data"]["object"]["id"])
 
     if event["type"] == "charge.dispute.created":
-        await _user_credit_model.handle_dispute(event["data"]["object"])
+        await UserCredit().handle_dispute(event["data"]["object"])
 
     if event["type"] == "refund.created" or event["type"] == "charge.dispute.closed":
-        await _user_credit_model.deduct_credits(event["data"]["object"])
+        await UserCredit().deduct_credits(event["data"]["object"])
 
     return Response(status_code=200)
 
@@ -577,7 +615,8 @@ async def stripe_webhook(request: Request):
 async def manage_payment_method(
     user_id: Annotated[str, Security(get_user_id)],
 ) -> dict[str, str]:
-    return {"url": await _user_credit_model.create_billing_portal_session(user_id)}
+    user_credit_model = await get_user_credit_model(user_id)
+    return {"url": await user_credit_model.create_billing_portal_session(user_id)}
 
 
 @v1_router.get(
@@ -595,7 +634,8 @@ async def get_credit_history(
     if transaction_count_limit < 1 or transaction_count_limit > 1000:
         raise ValueError("Transaction count limit must be between 1 and 1000")
 
-    return await _user_credit_model.get_transaction_history(
+    user_credit_model = await get_user_credit_model(user_id)
+    return await user_credit_model.get_transaction_history(
         user_id=user_id,
         transaction_time_ceiling=transaction_time,
         transaction_count_limit=transaction_count_limit,
@@ -612,7 +652,8 @@ async def get_credit_history(
 async def get_refund_requests(
     user_id: Annotated[str, Security(get_user_id)],
 ) -> list[RefundRequest]:
-    return await _user_credit_model.get_refund_requests(user_id)
+    user_credit_model = await get_user_credit_model(user_id)
+    return await user_credit_model.get_refund_requests(user_id)
 
 
 ########################################################
@@ -834,7 +875,8 @@ async def execute_graph(
     graph_version: Optional[int] = None,
     preset_id: Optional[str] = None,
 ) -> execution_db.GraphExecutionMeta:
-    current_balance = await _user_credit_model.get_credits(user_id)
+    user_credit_model = await get_user_credit_model(user_id)
+    current_balance = await user_credit_model.get_credits(user_id)
     if current_balance <= 0:
         raise HTTPException(
             status_code=402,

autogpt_platform/backend/backend/server/routers/v1_test.py

@@ -23,10 +23,13 @@ client = fastapi.testclient.TestClient(app)
 
 
 @pytest.fixture(autouse=True)
-def setup_app_auth(mock_jwt_user):
+def setup_app_auth(mock_jwt_user, setup_test_user):
     """Setup auth overrides for all tests in this module"""
     from autogpt_libs.auth.jwt_utils import get_jwt_payload
 
+    # setup_test_user fixture already executed and user is created in database
+    # It returns the user_id which we don't need to await
+
     app.dependency_overrides[get_jwt_payload] = mock_jwt_user["get_jwt_payload"]
     yield
     app.dependency_overrides.clear()
@@ -194,8 +197,12 @@ def test_get_user_credits(
     snapshot: Snapshot,
 ) -> None:
     """Test get user credits endpoint"""
-    mock_credit_model = mocker.patch("backend.server.routers.v1._user_credit_model")
+    mock_credit_model = Mock()
     mock_credit_model.get_credits = AsyncMock(return_value=1000)
+    mocker.patch(
+        "backend.server.routers.v1.get_user_credit_model",
+        return_value=mock_credit_model,
+    )
 
     response = client.get("/credits")
 
@@ -215,10 +222,14 @@ def test_request_top_up(
     snapshot: Snapshot,
 ) -> None:
     """Test request top up endpoint"""
-    mock_credit_model = mocker.patch("backend.server.routers.v1._user_credit_model")
+    mock_credit_model = Mock()
     mock_credit_model.top_up_intent = AsyncMock(
         return_value="https://checkout.example.com/session123"
     )
+    mocker.patch(
+        "backend.server.routers.v1.get_user_credit_model",
+        return_value=mock_credit_model,
+    )
 
     request_data = {"credit_amount": 500}
 
@@ -261,6 +272,74 @@ def test_get_auto_top_up(
     )
 
 
+def test_configure_auto_top_up(
+    mocker: pytest_mock.MockFixture,
+    snapshot: Snapshot,
+) -> None:
+    """Test configure auto top-up endpoint - this test would have caught the enum casting bug"""
+    # Mock the set_auto_top_up function to avoid database operations
+    mocker.patch(
+        "backend.server.routers.v1.set_auto_top_up",
+        return_value=None,
+    )
+
+    # Mock credit model to avoid Stripe API calls
+    mock_credit_model = mocker.AsyncMock()
+    mock_credit_model.get_credits.return_value = 50  # Current balance below threshold
+    mock_credit_model.top_up_credits.return_value = None
+
+    mocker.patch(
+        "backend.server.routers.v1.get_user_credit_model",
+        return_value=mock_credit_model,
+    )
+
+    # Test data
+    request_data = {
+        "threshold": 100,
+        "amount": 500,
+    }
+
+    response = client.post("/credits/auto-top-up", json=request_data)
+
+    # This should succeed with our fix, but would have failed before with the enum casting error
+    assert response.status_code == 200
+    assert response.json() == "Auto top-up settings updated"
+
+
+def test_configure_auto_top_up_validation_errors(
+    mocker: pytest_mock.MockFixture,
+) -> None:
+    """Test configure auto top-up endpoint validation"""
+    # Mock set_auto_top_up to avoid database operations for successful case
+    mocker.patch("backend.server.routers.v1.set_auto_top_up")
+
+    # Mock credit model to avoid Stripe API calls for the successful case
+    mock_credit_model = mocker.AsyncMock()
+    mock_credit_model.get_credits.return_value = 50
+    mock_credit_model.top_up_credits.return_value = None
+
+    mocker.patch(
+        "backend.server.routers.v1.get_user_credit_model",
+        return_value=mock_credit_model,
+    )
+
+    # Test negative threshold
+    response = client.post(
+        "/credits/auto-top-up", json={"threshold": -1, "amount": 500}
+    )
+    assert response.status_code == 422  # Validation error
+
+    # Test amount too small (but not 0)
+    response = client.post(
+        "/credits/auto-top-up", json={"threshold": 100, "amount": 100}
+    )
+    assert response.status_code == 422  # Validation error
+
+    # Test amount = 0 (should be allowed)
+    response = client.post("/credits/auto-top-up", json={"threshold": 100, "amount": 0})
+    assert response.status_code == 200  # Should succeed
+
+
 # Graphs endpoints tests
 def test_get_graphs(
     mocker: pytest_mock.MockFixture,

autogpt_platform/backend/backend/server/v2/admin/credit_admin_routes.py

@@ -11,8 +11,6 @@ from backend.util.json import SafeJson
 
 logger = logging.getLogger(__name__)
 
-_user_credit_model = get_user_credit_model()
-
 
 router = APIRouter(
     prefix="/admin",
@@ -33,7 +31,8 @@ async def add_user_credits(
     logger.info(
         f"Admin user {admin_user_id} is adding {amount} credits to user {user_id}"
     )
-    new_balance, transaction_key = await _user_credit_model._add_transaction(
+    user_credit_model = await get_user_credit_model(user_id)
+    new_balance, transaction_key = await user_credit_model._add_transaction(
         user_id,
         amount,
         transaction_type=CreditTransactionType.GRANT,

autogpt_platform/backend/backend/server/v2/admin/credit_admin_routes_test.py

@@ -1,5 +1,5 @@
 import json
-from unittest.mock import AsyncMock
+from unittest.mock import AsyncMock, Mock
 
 import fastapi
 import fastapi.testclient
@@ -7,12 +7,12 @@ import prisma.enums
 import pytest
 import pytest_mock
 from autogpt_libs.auth.jwt_utils import get_jwt_payload
-from prisma import Json
 from pytest_snapshot.plugin import Snapshot
 
 import backend.server.v2.admin.credit_admin_routes as credit_admin_routes
 import backend.server.v2.admin.model as admin_model
 from backend.data.model import UserTransaction
+from backend.util.json import SafeJson
 from backend.util.models import Pagination
 
 app = fastapi.FastAPI()
@@ -37,12 +37,14 @@ def test_add_user_credits_success(
 ) -> None:
     """Test successful credit addition by admin"""
     # Mock the credit model
-    mock_credit_model = mocker.patch(
-        "backend.server.v2.admin.credit_admin_routes._user_credit_model"
-    )
+    mock_credit_model = Mock()
     mock_credit_model._add_transaction = AsyncMock(
         return_value=(1500, "transaction-123-uuid")
     )
+    mocker.patch(
+        "backend.server.v2.admin.credit_admin_routes.get_user_credit_model",
+        return_value=mock_credit_model,
+    )
 
     request_data = {
         "user_id": target_user_id,
@@ -62,11 +64,17 @@ def test_add_user_credits_success(
     call_args = mock_credit_model._add_transaction.call_args
     assert call_args[0] == (target_user_id, 500)
     assert call_args[1]["transaction_type"] == prisma.enums.CreditTransactionType.GRANT
-    # Check that metadata is a Json object with the expected content
-    assert isinstance(call_args[1]["metadata"], Json)
-    assert call_args[1]["metadata"] == Json(
-        {"admin_id": admin_user_id, "reason": "Test credit grant for debugging"}
-    )
+    # Check that metadata is a SafeJson object with the expected content
+    assert isinstance(call_args[1]["metadata"], SafeJson)
+    actual_metadata = call_args[1]["metadata"]
+    expected_data = {
+        "admin_id": admin_user_id,
+        "reason": "Test credit grant for debugging",
+    }
+
+    # SafeJson inherits from Json which stores parsed data in the .data attribute
+    assert actual_metadata.data["admin_id"] == expected_data["admin_id"]
+    assert actual_metadata.data["reason"] == expected_data["reason"]
 
     # Snapshot test the response
     configured_snapshot.assert_match(
@@ -81,12 +89,14 @@ def test_add_user_credits_negative_amount(
 ) -> None:
     """Test credit deduction by admin (negative amount)"""
     # Mock the credit model
-    mock_credit_model = mocker.patch(
-        "backend.server.v2.admin.credit_admin_routes._user_credit_model"
-    )
+    mock_credit_model = Mock()
     mock_credit_model._add_transaction = AsyncMock(
         return_value=(200, "transaction-456-uuid")
     )
+    mocker.patch(
+        "backend.server.v2.admin.credit_admin_routes.get_user_credit_model",
+        return_value=mock_credit_model,
+    )
 
     request_data = {
         "user_id": "target-user-id",

autogpt_platform/backend/backend/server/v2/admin/store_admin_routes.py

@@ -7,6 +7,7 @@ import fastapi
 import fastapi.responses
 import prisma.enums
 
+import backend.server.v2.store.cache as store_cache
 import backend.server.v2.store.db
 import backend.server.v2.store.model
 import backend.util.json
@@ -86,6 +87,11 @@ async def review_submission(
         StoreSubmission with updated review information
     """
     try:
+        already_approved = (
+            await backend.server.v2.store.db.check_submission_already_approved(
+                store_listing_version_id=store_listing_version_id,
+            )
+        )
         submission = await backend.server.v2.store.db.review_store_submission(
             store_listing_version_id=store_listing_version_id,
             is_approved=request.is_approved,
@@ -93,6 +99,11 @@ async def review_submission(
             internal_comments=request.internal_comments or "",
             reviewer_id=user_id,
         )
+
+        state_changed = already_approved != request.is_approved
+        # Clear caches when the request is approved as it updates what is shown on the store
+        if state_changed:
+            store_cache.clear_all_caches()
         return submission
     except Exception as e:
         logger.exception("Error reviewing submission: %s", e)

autogpt_platform/backend/backend/server/v2/builder/db.py

@@ -2,7 +2,6 @@ import logging
 from datetime import datetime, timedelta, timezone
 
 import prisma
-from autogpt_libs.utils.cache import cached
 
 import backend.data.block
 from backend.blocks import load_all_blocks
@@ -18,6 +17,7 @@ from backend.server.v2.builder.model import (
     ProviderResponse,
     SearchBlocksResponse,
 )
+from backend.util.cache import cached
 from backend.util.models import Pagination
 
 logger = logging.getLogger(__name__)
@@ -118,6 +118,17 @@ def get_blocks(
     )
 
 
+def get_block_by_id(block_id: str) -> BlockInfo | None:
+    """
+    Get a specific block by its ID.
+    """
+    for block_type in load_all_blocks().values():
+        block: Block[BlockSchema, BlockSchema] = block_type()
+        if block.id == block_id:
+            return block.get_info()
+    return None
+
+
 def search_blocks(
     include_blocks: bool = True,
     include_integrations: bool = True,
@@ -296,7 +307,7 @@ def _matches_llm_model(schema_cls: type[BlockSchema], query: str) -> bool:
     return False
 
 
-@cached()
+@cached(ttl_seconds=3600)
 def _get_all_providers() -> dict[ProviderName, Provider]:
     providers: dict[ProviderName, Provider] = {}
 

autogpt_platform/backend/backend/server/v2/builder/model.py

@@ -53,16 +53,6 @@ class ProviderResponse(BaseModel):
     pagination: Pagination
 
 
-# Search
-class SearchRequest(BaseModel):
-    search_query: str | None = None
-    filter: list[FilterType] | None = None
-    by_creator: list[str] | None = None
-    search_id: str | None = None
-    page: int | None = None
-    page_size: int | None = None
-
-
 class SearchBlocksResponse(BaseModel):
     blocks: BlockResponse
     total_block_count: int

autogpt_platform/backend/backend/server/v2/builder/routes.py

@@ -110,6 +110,25 @@ async def get_blocks(
     )
 
 
+@router.get(
+    "/blocks/batch",
+    summary="Get specific blocks",
+    response_model=list[builder_model.BlockInfo],
+)
+async def get_specific_blocks(
+    block_ids: Annotated[list[str], fastapi.Query()],
+) -> list[builder_model.BlockInfo]:
+    """
+    Get specific blocks by their IDs.
+    """
+    blocks = []
+    for block_id in block_ids:
+        block = builder_db.get_block_by_id(block_id)
+        if block:
+            blocks.append(block)
+    return blocks
+
+
 @router.get(
     "/providers",
     summary="Get Builder integration providers",
@@ -128,30 +147,34 @@ async def get_providers(
     )
 
 
-@router.post(
+# Not using post method because on frontend, orval doesn't support Infinite Query with POST method.
+@router.get(
     "/search",
     summary="Builder search",
     tags=["store", "private"],
     response_model=builder_model.SearchResponse,
 )
 async def search(
-    options: builder_model.SearchRequest,
     user_id: Annotated[str, fastapi.Security(get_user_id)],
+    search_query: Annotated[str | None, fastapi.Query()] = None,
+    filter: Annotated[list[str] | None, fastapi.Query()] = None,
+    search_id: Annotated[str | None, fastapi.Query()] = None,
+    by_creator: Annotated[list[str] | None, fastapi.Query()] = None,
+    page: Annotated[int, fastapi.Query()] = 1,
+    page_size: Annotated[int, fastapi.Query()] = 50,
 ) -> builder_model.SearchResponse:
     """
     Search for blocks (including integrations), marketplace agents, and user library agents.
     """
     # If no filters are provided, then we will return all types
-    if not options.filter:
-        options.filter = [
+    if not filter:
+        filter = [
             "blocks",
             "integrations",
             "marketplace_agents",
             "my_agents",
         ]
-    options.search_query = sanitize_query(options.search_query)
-    options.page = options.page or 1
-    options.page_size = options.page_size or 50
+    search_query = sanitize_query(search_query)
 
     # Blocks&Integrations
     blocks = builder_model.SearchBlocksResponse(
@@ -162,13 +185,13 @@ async def search(
         total_block_count=0,
         total_integration_count=0,
     )
-    if "blocks" in options.filter or "integrations" in options.filter:
+    if "blocks" in filter or "integrations" in filter:
         blocks = builder_db.search_blocks(
-            include_blocks="blocks" in options.filter,
-            include_integrations="integrations" in options.filter,
-            query=options.search_query or "",
-            page=options.page,
-            page_size=options.page_size,
+            include_blocks="blocks" in filter,
+            include_integrations="integrations" in filter,
+            query=search_query or "",
+            page=page,
+            page_size=page_size,
         )
 
     # Library Agents
@@ -176,12 +199,12 @@ async def search(
         agents=[],
         pagination=Pagination.empty(),
     )
-    if "my_agents" in options.filter:
+    if "my_agents" in filter:
         my_agents = await library_db.list_library_agents(
             user_id=user_id,
-            search_term=options.search_query,
-            page=options.page,
-            page_size=options.page_size,
+            search_term=search_query,
+            page=page,
+            page_size=page_size,
         )
 
     # Marketplace Agents
@@ -189,12 +212,12 @@ async def search(
         agents=[],
         pagination=Pagination.empty(),
     )
-    if "marketplace_agents" in options.filter:
+    if "marketplace_agents" in filter:
         marketplace_agents = await store_db.get_store_agents(
-            creators=options.by_creator,
-            search_query=options.search_query,
-            page=options.page,
-            page_size=options.page_size,
+            creators=by_creator,
+            search_query=search_query,
+            page=page,
+            page_size=page_size,
         )
 
     more_pages = False
@@ -214,7 +237,7 @@ async def search(
             "marketplace_agents": marketplace_agents.pagination.total_items,
             "my_agents": my_agents.pagination.total_items,
         },
-        page=options.page,
+        page=page,
         more_pages=more_pages,
     )