refactor: move is_uuid and is_creator_slug helpers to shared utils

MCP tools have their own run_mcp_tool path — don't tell users they can
"only be used in agent graphs" when looking up an MCP block by UUID.
Also use block.id instead of raw query in all excluded-block messages.

.claude/skills/backend-check/SKILL.md

@@ -1,17 +0,0 @@
----
-name: backend-check
-description: Run the full backend formatting, linting, and test suite. Ensures code quality before commits and PRs. TRIGGER when backend Python code has been modified and needs validation.
-user-invocable: true
-metadata:
-  author: autogpt-team
-  version: "1.0.0"
----
-
-# Backend Check
-
-## Steps
-
-1. **Format**: `poetry run format` — runs formatting AND linting. NEVER run ruff/black/isort individually
-2. **Fix** any remaining errors manually, re-run until clean
-3. **Test**: `poetry run test` (runs DB setup + pytest). For specific files: `poetry run pytest -s -vvv <test_files>`
-4. **Snapshots** (if needed): `poetry run pytest path/to/test.py --snapshot-update` — review with `git diff`

.claude/skills/code-style/SKILL.md

@@ -1,35 +0,0 @@
----
-name: code-style
-description: Python code style preferences for the AutoGPT backend. Apply when writing or reviewing Python code. TRIGGER when writing new Python code, reviewing PRs, or refactoring backend code.
-user-invocable: false
-metadata:
-  author: autogpt-team
-  version: "1.0.0"
----
-
-# Code Style
-
-## Imports
-
-- **Top-level only** — no local/inner imports. Move all imports to the top of the file.
-
-## Typing
-
-- **No duck typing** — avoid `hasattr`, `getattr`, `isinstance` for type dispatch. Use proper typed interfaces, unions, or protocols.
-- **Pydantic models** over dataclass, namedtuple, or raw dict for structured data.
-- **No linter suppressors** — avoid `# type: ignore`, `# noqa`, `# pyright: ignore` etc. 99% of the time the right fix is fixing the type/code, not silencing the tool.
-
-## Code Structure
-
-- **List comprehensions** over manual loop-and-append.
-- **Early return** — guard clauses first, avoid deep nesting.
-- **Flatten inline** — prefer short, concise expressions. Reduce `if/else` chains with direct returns or ternaries when readable.
-- **Modular functions** — break complex logic into small, focused functions rather than long blocks with nested conditionals.
-
-## Review Checklist
-
-Before finishing, always ask:
-- Can any function be split into smaller pieces?
-- Is there unnecessary nesting that an early return would eliminate?
-- Can any loop be a comprehension?
-- Is there a simpler way to express this logic?

.claude/skills/frontend-check/SKILL.md

@@ -1,16 +0,0 @@
----
-name: frontend-check
-description: Run the full frontend formatting, linting, and type checking suite. Ensures code quality before commits and PRs. TRIGGER when frontend TypeScript/React code has been modified and needs validation.
-user-invocable: true
-metadata:
-  author: autogpt-team
-  version: "1.0.0"
----
-
-# Frontend Check
-
-## Steps (in order)
-
-1. **Format**: `pnpm format` — NEVER run individual formatters
-2. **Lint**: `pnpm lint` — fix errors, re-run until clean
-3. **Types**: `pnpm types` — if it keeps failing after multiple attempts, stop and ask the user

.claude/skills/new-block/SKILL.md

@@ -1,29 +0,0 @@
----
-name: new-block
-description: Create a new backend block following the Block SDK Guide. Guides through provider configuration, schema definition, authentication, and testing. TRIGGER when user asks to create a new block, add a new integration, or build a new node for the graph editor.
-user-invocable: true
-metadata:
-  author: autogpt-team
-  version: "1.0.0"
----
-
-# New Block Creation
-
-Read `docs/platform/block-sdk-guide.md` first for the full guide.
-
-## Steps
-
-1. **Provider config** (if external service): create `_config.py` with `ProviderBuilder`
-2. **Block file** in `backend/blocks/` (from `autogpt_platform/backend/`):
-   - Generate a UUID once with `uuid.uuid4()`, then **hard-code that string** as `id` (IDs must be stable across imports)
-   - `Input(BlockSchema)` and `Output(BlockSchema)` classes
-   - `async def run` that `yield`s output fields
-3. **Files**: use `store_media_file()` with `"for_block_output"` for outputs
-4. **Test**: `poetry run pytest 'backend/blocks/test/test_block.py::test_available_blocks[MyBlock]' -xvs`
-5. **Format**: `poetry run format`
-
-## Rules
-
-- Analyze interfaces: do inputs/outputs connect well with other blocks in a graph?
-- Use top-level imports, avoid duck typing
-- Always use `for_block_output` for block outputs

.claude/skills/openapi-regen/SKILL.md

@@ -1,28 +0,0 @@
----
-name: openapi-regen
-description: Regenerate the OpenAPI spec and frontend API client. Starts the backend REST server, fetches the spec, and regenerates the typed frontend hooks. TRIGGER when API routes change, new endpoints are added, or frontend API types are stale.
-user-invocable: true
-metadata:
-  author: autogpt-team
-  version: "1.0.0"
----
-
-# OpenAPI Spec Regeneration
-
-## Steps
-
-1. **Run end-to-end** in a single shell block (so `REST_PID` persists):
-   ```bash
-   cd autogpt_platform/backend && poetry run rest &
-   REST_PID=$!
-   WAIT=0; until curl -sf http://localhost:8006/health > /dev/null 2>&1; do sleep 1; WAIT=$((WAIT+1)); [ $WAIT -ge 60 ] && echo "Timed out" && kill $REST_PID && exit 1; done
-   cd ../frontend && pnpm generate:api:force
-   kill $REST_PID
-   pnpm types && pnpm lint && pnpm format
-   ```
-
-## Rules
-
-- Always use `pnpm generate:api:force` (not `pnpm generate:api`)
-- Don't manually edit files in `src/app/api/__generated__/`
-- Generated hooks follow: `use{Method}{Version}{OperationName}`

.claude/skills/pr-address/SKILL.md

@@ -0,0 +1,79 @@
+---
+name: pr-address
+description: Address PR review comments and loop until CI green and all comments resolved. TRIGGER when user asks to address comments, fix PR feedback, respond to reviewers, or babysit/monitor a PR.
+user-invocable: true
+args: "[PR number or URL] — if omitted, finds PR for current branch."
+metadata:
+  author: autogpt-team
+  version: "1.0.0"
+---
+
+# PR Address
+
+## Find the PR
+
+```bash
+gh pr list --head $(git branch --show-current) --repo Significant-Gravitas/AutoGPT
+gh pr view {N}
+```
+
+## Fetch comments (all sources)
+
+```bash
+gh api repos/Significant-Gravitas/AutoGPT/pulls/{N}/reviews       # top-level reviews
+gh api repos/Significant-Gravitas/AutoGPT/pulls/{N}/comments      # inline review comments
+gh api repos/Significant-Gravitas/AutoGPT/issues/{N}/comments     # PR conversation comments
+```
+
+**Bots to watch for:**
+- `autogpt-reviewer` — posts "Blockers", "Should Fix", "Nice to Have". Address ALL of them.
+- `sentry[bot]` — bug predictions. Fix real bugs, explain false positives.
+- `coderabbitai[bot]` — automated review. Address actionable items.
+
+## For each unaddressed comment
+
+Address comments **one at a time**: fix → commit → push → inline reply → next.
+
+1. Read the referenced code, make the fix (or reply explaining why it's not needed)
+2. Commit and push the fix
+3. Reply **inline** (not as a new top-level comment) referencing the fixing commit — this is what resolves the conversation for bot reviewers (coderabbitai, sentry):
+
+| Comment type | How to reply |
+|---|---|
+| Inline review (`pulls/{N}/comments`) | `gh api repos/Significant-Gravitas/AutoGPT/pulls/{N}/comments/{ID}/replies -f body="Fixed in <commit-sha>: <description>"` |
+| Conversation (`issues/{N}/comments`) | `gh api repos/Significant-Gravitas/AutoGPT/issues/{N}/comments -f body="Fixed in <commit-sha>: <description>"` |
+
+## Format and commit
+
+After fixing, format the changed code:
+
+- **Backend** (from `autogpt_platform/backend/`): `poetry run format`
+- **Frontend** (from `autogpt_platform/frontend/`): `pnpm format && pnpm lint && pnpm types`
+
+If API routes changed, regenerate the frontend client:
+```bash
+cd autogpt_platform/backend && poetry run rest &
+REST_PID=$!
+trap "kill $REST_PID 2>/dev/null" EXIT
+WAIT=0; until curl -sf http://localhost:8006/health > /dev/null 2>&1; do sleep 1; WAIT=$((WAIT+1)); [ $WAIT -ge 60 ] && echo "Timed out" && exit 1; done
+cd ../frontend && pnpm generate:api:force
+kill $REST_PID 2>/dev/null; trap - EXIT
+```
+Never manually edit files in `src/app/api/__generated__/`.
+
+Then commit and **push immediately** — never batch commits without pushing.
+
+For backend commits in worktrees: `poetry run git commit` (pre-commit hooks).
+
+## The loop
+
+```text
+address comments → format → commit → push
+→ re-check comments → fix new ones → push
+→ wait for CI → re-check comments after CI settles
+→ repeat until: all comments addressed AND CI green AND no new comments arriving
+```
+
+While CI runs, stay productive: run local tests, address remaining comments.
+
+**The loop ends when:** CI fully green + all comments addressed + no new comments since CI settled.

.claude/skills/pr-create/SKILL.md

@@ -1,31 +0,0 @@
----
-name: pr-create
-description: Create a pull request for the current branch. TRIGGER when user asks to create a PR, open a pull request, push changes for review, or submit work for merging.
-user-invocable: true
-metadata:
-  author: autogpt-team
-  version: "1.0.0"
----
-
-# Create Pull Request
-
-## Steps
-
-1. **Check for existing PR**: `gh pr view --json url -q .url 2>/dev/null` — if a PR already exists, output its URL and stop
-2. **Understand changes**: `git status`, `git diff dev...HEAD`, `git log dev..HEAD --oneline`
-3. **Read PR template**: `.github/PULL_REQUEST_TEMPLATE.md`
-4. **Draft PR title**: Use conventional commits format (see CLAUDE.md for types and scopes)
-5. **Fill out PR template** as the body — be thorough in the Changes section
-6. **Format first** (if relevant changes exist):
-   - Backend: `cd autogpt_platform/backend && poetry run format`
-   - Frontend: `cd autogpt_platform/frontend && pnpm format`
-   - Fix any lint errors, then commit formatting changes before pushing
-7. **Push**: `git push -u origin HEAD`
-8. **Create PR**: `gh pr create --base dev`
-9. **Output** the PR URL
-
-## Rules
-
-- Always target `dev` branch
-- Do NOT run tests — CI will handle that
-- Use the PR template from `.github/PULL_REQUEST_TEMPLATE.md`

.claude/skills/pr-review/SKILL.md

@@ -1,51 +1,74 @@
 ---
 name: pr-review
-description: Address all open PR review comments systematically. Fetches comments, addresses each one, reacts +1/-1, and replies when clarification is needed. Keeps iterating until all comments are addressed and CI is green. TRIGGER when user shares a PR URL, asks to address review comments, fix PR feedback, or respond to reviewer comments.
+description: Review a PR for correctness, security, code quality, and testing issues. TRIGGER when user asks to review a PR, check PR quality, or give feedback on a PR.
 user-invocable: true
+args: "[PR number or URL] — if omitted, finds PR for current branch."
 metadata:
   author: autogpt-team
   version: "1.0.0"
 ---
 
-# PR Review Comment Workflow
+# PR Review
 
-## Steps
+## Find the PR
 
-1. **Find PR**: `gh pr list --head $(git branch --show-current) --repo Significant-Gravitas/AutoGPT`
-2. **Fetch comments** (all three sources):
-   - `gh api repos/Significant-Gravitas/AutoGPT/pulls/{N}/reviews` (top-level reviews)
-   - `gh api repos/Significant-Gravitas/AutoGPT/pulls/{N}/comments` (inline review comments)
-   - `gh api repos/Significant-Gravitas/AutoGPT/issues/{N}/comments` (PR conversation comments)
-3. **Skip** comments already reacted to by PR author
-4. **For each unreacted comment**:
-   - Read referenced code, make the fix (or reply if you disagree/need info)
-   - **Inline review comments** (`pulls/{N}/comments`):
-     - React: `gh api repos/.../pulls/comments/{ID}/reactions -f content="+1"` (or `-1`)
-     - Reply: `gh api repos/.../pulls/{N}/comments/{ID}/replies -f body="..."`
-   - **PR conversation comments** (`issues/{N}/comments`):
-     - React: `gh api repos/.../issues/comments/{ID}/reactions -f content="+1"` (or `-1`)
-     - No threaded replies — post a new issue comment if needed
-   - **Top-level reviews**: no reaction API — address in code, reply via issue comment if needed
-5. **Include autogpt-reviewer bot fixes** too
-6. **Format**: `cd autogpt_platform/backend && poetry run format`, `cd autogpt_platform/frontend && pnpm format`
-7. **Commit & push**
-8. **Re-fetch comments** immediately — address any new unreacted ones before waiting on CI
-9. **Stay productive while CI runs** — don't idle. In priority order:
-   - Run any pending local tests (`poetry run pytest`, e2e, etc.) and fix failures
-   - Address any remaining comments
-   - Only poll `gh pr checks {N}` as the last resort when there's truly nothing left to do
-10. **If CI fails** — fix, go back to step 6
-11. **Re-fetch comments again** after CI is green — address anything that appeared while CI was running
-12. **Done** only when: all comments reacted AND CI is green.
+```bash
+gh pr list --head $(git branch --show-current) --repo Significant-Gravitas/AutoGPT
+gh pr view {N}
+```
 
-## CRITICAL: Do Not Stop
+## Read the diff
 
-**Loop is: address → format → commit → push → re-check comments → run local tests → wait CI → re-check comments → repeat.**
+```bash
+gh pr diff {N}
+```
 
-Never idle. If CI is running and you have nothing to address, run local tests. Waiting on CI is the last resort.
+## Fetch existing review comments
 
-## Rules
+Before posting anything, fetch existing inline comments to avoid duplicates:
 
-- One todo per comment
-- For inline review comments: reply on existing threads. For PR conversation comments: post a new issue comment (API doesn't support threaded replies)
-- React to every comment: +1 addressed, -1 disagreed (with explanation)
+```bash
+gh api repos/Significant-Gravitas/AutoGPT/pulls/{N}/comments
+gh api repos/Significant-Gravitas/AutoGPT/pulls/{N}/reviews
+```
+
+## What to check
+
+**Correctness:** logic errors, off-by-one, missing edge cases, race conditions (TOCTOU in file access, credit charging), error handling gaps, async correctness (missing `await`, unclosed resources).
+
+**Security:** input validation at boundaries, no injection (command, XSS, SQL), secrets not logged, file paths sanitized (`os.path.basename()` in error messages).
+
+**Code quality:** apply rules from backend/frontend CLAUDE.md files.
+
+**Architecture:** DRY, single responsibility, modular functions. `Security()` vs `Depends()` for FastAPI auth. `data:` for SSE events, `: comment` for heartbeats. `transaction=True` for Redis pipelines.
+
+**Testing:** edge cases covered, colocated `*_test.py` (backend) / `__tests__/` (frontend), mocks target where symbol is **used** not defined, `AsyncMock` for async.
+
+## Output format
+
+Every comment **must** be prefixed with `🤖` and a criticality badge:
+
+| Tier | Badge | Meaning |
+|---|---|---|
+| Blocker | `🔴 **Blocker**` | Must fix before merge |
+| Should Fix | `🟠 **Should Fix**` | Important improvement |
+| Nice to Have | `🟡 **Nice to Have**` | Minor suggestion |
+| Nit | `🔵 **Nit**` | Style / wording |
+
+Example: `🤖 🔴 **Blocker**: Missing error handling for X — suggest wrapping in try/except.`
+
+## Post inline comments
+
+For each finding, post an inline comment on the PR (do not just write a local report):
+
+```bash
+# Get the latest commit SHA for the PR
+COMMIT_SHA=$(gh api repos/Significant-Gravitas/AutoGPT/pulls/{N} --jq '.head.sha')
+
+# Post an inline comment on a specific file/line
+gh api repos/Significant-Gravitas/AutoGPT/pulls/{N}/comments \
+  -f body="🤖 🔴 **Blocker**: <description>" \
+  -f commit_id="$COMMIT_SHA" \
+  -f path="<file path>" \
+  -F line=<line number>
+```

.claude/skills/worktree-setup/SKILL.md

@@ -1,45 +0,0 @@
----
-name: worktree-setup
-description: Set up a new git worktree for parallel development. Creates the worktree, copies .env files, installs dependencies, generates Prisma client, and optionally starts the app (with port conflict resolution) or runs tests. TRIGGER when user asks to set up a worktree, work on a branch in isolation, or needs a separate environment for a branch or PR.
-user-invocable: true
-metadata:
-  author: autogpt-team
-  version: "1.0.0"
----
-
-# Worktree Setup
-
-## Preferred: Use Branchlet
-
-The repo has a `.branchlet.json` config — it handles env file copying, dependency installation, and Prisma generation automatically.
-
-```bash
-npm install -g branchlet                                      # install once
-branchlet create -n <name> -s <source-branch> -b <new-branch>
-branchlet list --json   # list all worktrees
-```
-
-## Manual Fallback
-
-If branchlet isn't available:
-
-1. `git worktree add ../<RepoName><N> <branch-name>`
-2. Copy `.env` files: `backend/.env`, `frontend/.env`, `autogpt_platform/.env`, `db/docker/.env`
-3. Install deps:
-   - `cd autogpt_platform/backend && poetry install && poetry run prisma generate`
-   - `cd autogpt_platform/frontend && pnpm install`
-
-## Running the App
-
-Free ports first — backend uses: 8001, 8002, 8003, 8005, 8006, 8007, 8008.
-
-```bash
-for port in 8001 8002 8003 8005 8006 8007 8008; do
-  lsof -ti :$port | xargs kill -9 2>/dev/null || true
-done
-cd <worktree>/autogpt_platform/backend && poetry run app
-```
-
-## CoPilot Testing Gotcha
-
-SDK mode spawns a Claude subprocess — **won't work inside Claude Code**. Set `CHAT_USE_CLAUDE_AGENT_SDK=false` in `backend/.env` to use baseline mode.

.claude/skills/worktree/SKILL.md

@@ -0,0 +1,85 @@
+---
+name: worktree
+description: Set up a new git worktree for parallel development. Creates the worktree, copies .env files, installs dependencies, and generates Prisma client. TRIGGER when user asks to set up a worktree, work on a branch in isolation, or needs a separate environment for a branch or PR.
+user-invocable: true
+args: "[name] — optional worktree name (e.g., 'AutoGPT7'). If omitted, uses next available AutoGPT<N>."
+metadata:
+  author: autogpt-team
+  version: "3.0.0"
+---
+
+# Worktree Setup
+
+## Create the worktree
+
+Derive paths from the git toplevel. If a name is provided as argument, use it. Otherwise, check `git worktree list` and pick the next `AutoGPT<N>`.
+
+```bash
+ROOT=$(git rev-parse --show-toplevel)
+PARENT=$(dirname "$ROOT")
+
+# From an existing branch
+git worktree add "$PARENT/<NAME>" <branch-name>
+
+# From a new branch off dev
+git worktree add -b <new-branch> "$PARENT/<NAME>" dev
+```
+
+## Copy environment files
+
+Copy `.env` from the root worktree. Falls back to `.env.default` if `.env` doesn't exist.
+
+```bash
+ROOT=$(git rev-parse --show-toplevel)
+TARGET="$(dirname "$ROOT")/<NAME>"
+
+for envpath in autogpt_platform/backend autogpt_platform/frontend autogpt_platform; do
+  if [ -f "$ROOT/$envpath/.env" ]; then
+    cp "$ROOT/$envpath/.env" "$TARGET/$envpath/.env"
+  elif [ -f "$ROOT/$envpath/.env.default" ]; then
+    cp "$ROOT/$envpath/.env.default" "$TARGET/$envpath/.env"
+  fi
+done
+```
+
+## Install dependencies
+
+```bash
+TARGET="$(dirname "$(git rev-parse --show-toplevel)")/<NAME>"
+cd "$TARGET/autogpt_platform/autogpt_libs" && poetry install
+cd "$TARGET/autogpt_platform/backend" && poetry install && poetry run prisma generate
+cd "$TARGET/autogpt_platform/frontend" && pnpm install
+```
+
+Replace `<NAME>` with the actual worktree name (e.g., `AutoGPT7`).
+
+## Running the app (optional)
+
+Backend uses ports: 8001, 8002, 8003, 8005, 8006, 8007, 8008. Free them first if needed:
+
+```bash
+TARGET="$(dirname "$(git rev-parse --show-toplevel)")/<NAME>"
+for port in 8001 8002 8003 8005 8006 8007 8008; do
+  lsof -ti :$port | xargs kill -9 2>/dev/null || true
+done
+cd "$TARGET/autogpt_platform/backend" && poetry run app
+```
+
+## CoPilot testing
+
+SDK mode spawns a Claude subprocess — won't work inside Claude Code. Set `CHAT_USE_CLAUDE_AGENT_SDK=false` in `backend/.env` to use baseline mode.
+
+## Cleanup
+
+```bash
+# Replace <NAME> with the actual worktree name (e.g., AutoGPT7)
+git worktree remove "$(dirname "$(git rev-parse --show-toplevel)")/<NAME>"
+```
+
+## Alternative: Branchlet (optional)
+
+If [branchlet](https://www.npmjs.com/package/branchlet) is installed:
+
+```bash
+branchlet create -n <name> -s <source-branch> -b <new-branch>
+```

autogpt_platform/CLAUDE.md

@@ -60,9 +60,12 @@ AutoGPT Platform is a monorepo containing:
 
 ### Reviewing/Revising Pull Requests
 
-- When the user runs /pr-comments or tries to fetch them, also run gh api /repos/Significant-Gravitas/AutoGPT/pulls/[issuenum]/reviews to get the reviews
-- Use gh api /repos/Significant-Gravitas/AutoGPT/pulls/[issuenum]/reviews/[review_id]/comments to get the review contents
-- Use gh api /repos/Significant-Gravitas/AutoGPT/issues/9924/comments to get the pr specific comments
+Use `/pr-review` to review a PR or `/pr-address` to address comments.
+
+When fetching comments manually:
+- `gh api repos/Significant-Gravitas/AutoGPT/pulls/{N}/reviews` — top-level reviews
+- `gh api repos/Significant-Gravitas/AutoGPT/pulls/{N}/comments` — inline review comments
+- `gh api repos/Significant-Gravitas/AutoGPT/issues/{N}/comments` — PR conversation comments
 
 ### Conventional Commits
 

autogpt_platform/analytics/queries/auth_activities.sql

@@ -0,0 +1,40 @@
+-- =============================================================
+-- View: analytics.auth_activities
+-- Looker source alias: ds49  |  Charts: 1
+-- =============================================================
+-- DESCRIPTION
+--   Tracks authentication events (login, logout, SSO, password
+--   reset, etc.) from Supabase's internal audit log.
+--   Useful for monitoring sign-in patterns and detecting anomalies.
+--
+-- SOURCE TABLES
+--   auth.audit_log_entries  — Supabase internal auth event log
+--
+-- OUTPUT COLUMNS
+--   created_at      TIMESTAMPTZ  When the auth event occurred
+--   actor_id        TEXT         User ID who triggered the event
+--   actor_via_sso   TEXT         Whether the action was via SSO ('true'/'false')
+--   action          TEXT         Event type (e.g. 'login', 'logout', 'token_refreshed')
+--
+-- WINDOW
+--   Rolling 90 days from current date
+--
+-- EXAMPLE QUERIES
+--   -- Daily login counts
+--   SELECT DATE_TRUNC('day', created_at) AS day, COUNT(*) AS logins
+--   FROM analytics.auth_activities
+--   WHERE action = 'login'
+--   GROUP BY 1 ORDER BY 1;
+--
+--   -- SSO vs password login breakdown
+--   SELECT actor_via_sso, COUNT(*) FROM analytics.auth_activities
+--   WHERE action = 'login' GROUP BY 1;
+-- =============================================================
+
+SELECT
+    created_at,
+    payload->>'actor_id'      AS actor_id,
+    payload->>'actor_via_sso' AS actor_via_sso,
+    payload->>'action'        AS action
+FROM auth.audit_log_entries
+WHERE created_at >= NOW() - INTERVAL '90 days'

autogpt_platform/analytics/queries/graph_execution.sql

@@ -0,0 +1,105 @@
+-- =============================================================
+-- View: analytics.graph_execution
+-- Looker source alias: ds16  |  Charts: 21
+-- =============================================================
+-- DESCRIPTION
+--   One row per agent graph execution (last 90 days).
+--   Unpacks the JSONB stats column into individual numeric columns
+--   and normalises the executionStatus — runs that failed due to
+--   insufficient credits are reclassified as 'NO_CREDITS' for
+--   easier filtering.  Error messages are scrubbed of IDs and URLs
+--   to allow safe grouping.
+--
+-- SOURCE TABLES
+--   platform.AgentGraphExecution  — Execution records
+--   platform.AgentGraph           — Agent graph metadata (for name)
+--   platform.LibraryAgent         — To flag possibly-AI (safe-mode) agents
+--
+-- OUTPUT COLUMNS
+--   id                TEXT         Execution UUID
+--   agentGraphId      TEXT         Agent graph UUID
+--   agentGraphVersion INT          Graph version number
+--   executionStatus   TEXT         COMPLETED | FAILED | NO_CREDITS | RUNNING | QUEUED | TERMINATED
+--   createdAt         TIMESTAMPTZ  When the execution was queued
+--   updatedAt         TIMESTAMPTZ  Last status update time
+--   userId            TEXT         Owner user UUID
+--   agentGraphName    TEXT         Human-readable agent name
+--   cputime           DECIMAL      Total CPU seconds consumed
+--   walltime          DECIMAL      Total wall-clock seconds
+--   node_count        DECIMAL      Number of nodes in the graph
+--   nodes_cputime     DECIMAL      CPU time across all nodes
+--   nodes_walltime    DECIMAL      Wall time across all nodes
+--   execution_cost    DECIMAL      Credit cost of this execution
+--   correctness_score FLOAT        AI correctness score (if available)
+--   possibly_ai       BOOLEAN      True if agent has sensitive_action_safe_mode enabled
+--   groupedErrorMessage TEXT       Scrubbed error string (IDs/URLs replaced with wildcards)
+--
+-- WINDOW
+--   Rolling 90 days (createdAt > CURRENT_DATE - 90 days)
+--
+-- EXAMPLE QUERIES
+--   -- Daily execution counts by status
+--   SELECT DATE_TRUNC('day', "createdAt") AS day, "executionStatus", COUNT(*)
+--   FROM analytics.graph_execution
+--   GROUP BY 1, 2 ORDER BY 1;
+--
+--   -- Average cost per execution by agent
+--   SELECT "agentGraphName", AVG("execution_cost") AS avg_cost, COUNT(*) AS runs
+--   FROM analytics.graph_execution
+--   WHERE "executionStatus" = 'COMPLETED'
+--   GROUP BY 1 ORDER BY avg_cost DESC;
+--
+--   -- Top error messages
+--   SELECT "groupedErrorMessage", COUNT(*) AS occurrences
+--   FROM analytics.graph_execution
+--   WHERE "executionStatus" = 'FAILED'
+--   GROUP BY 1 ORDER BY 2 DESC LIMIT 20;
+-- =============================================================
+
+SELECT
+    ge."id"                                                        AS id,
+    ge."agentGraphId"                                              AS agentGraphId,
+    ge."agentGraphVersion"                                         AS agentGraphVersion,
+    CASE
+        WHEN jsonb_exists(ge."stats"::jsonb, 'error')
+         AND (
+               (ge."stats"::jsonb->>'error') ILIKE '%insufficient balance%'
+            OR (ge."stats"::jsonb->>'error') ILIKE '%you have no credits left%'
+             )
+        THEN 'NO_CREDITS'
+        ELSE CAST(ge."executionStatus" AS TEXT)
+    END                                                            AS executionStatus,
+    ge."createdAt"                                                 AS createdAt,
+    ge."updatedAt"                                                 AS updatedAt,
+    ge."userId"                                                    AS userId,
+    g."name"                                                       AS agentGraphName,
+    (ge."stats"::jsonb->>'cputime')::decimal                       AS cputime,
+    (ge."stats"::jsonb->>'walltime')::decimal                      AS walltime,
+    (ge."stats"::jsonb->>'node_count')::decimal                    AS node_count,
+    (ge."stats"::jsonb->>'nodes_cputime')::decimal                 AS nodes_cputime,
+    (ge."stats"::jsonb->>'nodes_walltime')::decimal                AS nodes_walltime,
+    (ge."stats"::jsonb->>'cost')::decimal                          AS execution_cost,
+    (ge."stats"::jsonb->>'correctness_score')::float               AS correctness_score,
+    COALESCE(la.possibly_ai, FALSE)                                AS possibly_ai,
+    REGEXP_REPLACE(
+        REGEXP_REPLACE(
+            TRIM(BOTH '"' FROM ge."stats"::jsonb->>'error'),
+            '(https?://)([A-Za-z0-9.-]+)(:[0-9]+)?(/[^\s]*)?',
+            '\1\2/...', 'gi'
+        ),
+        '[a-zA-Z0-9_:-]*\d[a-zA-Z0-9_:-]*', '*', 'g'
+    )                                                              AS groupedErrorMessage
+FROM platform."AgentGraphExecution" ge
+LEFT JOIN platform."AgentGraph" g
+       ON ge."agentGraphId" = g."id"
+      AND ge."agentGraphVersion" = g."version"
+LEFT JOIN (
+    SELECT DISTINCT ON ("userId", "agentGraphId")
+           "userId", "agentGraphId",
+           ("settings"::jsonb->>'sensitive_action_safe_mode')::boolean AS possibly_ai
+    FROM platform."LibraryAgent"
+    WHERE "isDeleted"  = FALSE
+      AND "isArchived" = FALSE
+    ORDER BY "userId", "agentGraphId", "agentGraphVersion" DESC
+) la ON la."userId" = ge."userId" AND la."agentGraphId" = ge."agentGraphId"
+WHERE ge."createdAt" > CURRENT_DATE - INTERVAL '90 days'

autogpt_platform/analytics/queries/node_block_execution.sql

@@ -0,0 +1,101 @@
+-- =============================================================
+-- View: analytics.node_block_execution
+-- Looker source alias: ds14  |  Charts: 11
+-- =============================================================
+-- DESCRIPTION
+--   One row per node (block) execution (last 90 days).
+--   Unpacks stats JSONB and joins to identify which block type
+--   was run.  For failed nodes, joins the error output and
+--   scrubs it for safe grouping.
+--
+-- SOURCE TABLES
+--   platform.AgentNodeExecution              — Node execution records
+--   platform.AgentNode                       — Node → block mapping
+--   platform.AgentBlock                      — Block name/ID
+--   platform.AgentNodeExecutionInputOutput   — Error output values
+--
+-- OUTPUT COLUMNS
+--   id                    TEXT         Node execution UUID
+--   agentGraphExecutionId TEXT         Parent graph execution UUID
+--   agentNodeId           TEXT         Node UUID within the graph
+--   executionStatus       TEXT         COMPLETED | FAILED | QUEUED | RUNNING | TERMINATED
+--   addedTime             TIMESTAMPTZ  When the node was queued
+--   queuedTime            TIMESTAMPTZ  When it entered the queue
+--   startedTime           TIMESTAMPTZ  When execution started
+--   endedTime             TIMESTAMPTZ  When execution finished
+--   inputSize             BIGINT       Input payload size in bytes
+--   outputSize            BIGINT       Output payload size in bytes
+--   walltime              NUMERIC      Wall-clock seconds for this node
+--   cputime               NUMERIC      CPU seconds for this node
+--   llmRetryCount         INT          Number of LLM retries
+--   llmCallCount          INT          Number of LLM API calls made
+--   inputTokenCount       BIGINT       LLM input tokens consumed
+--   outputTokenCount      BIGINT       LLM output tokens produced
+--   blockName             TEXT         Human-readable block name (e.g. 'OpenAIBlock')
+--   blockId               TEXT         Block UUID
+--   groupedErrorMessage   TEXT         Scrubbed error (IDs/URLs wildcarded)
+--   errorMessage          TEXT         Raw error output (only set when FAILED)
+--
+-- WINDOW
+--   Rolling 90 days (addedTime > CURRENT_DATE - 90 days)
+--
+-- EXAMPLE QUERIES
+--   -- Most-used blocks by execution count
+--   SELECT "blockName", COUNT(*) AS executions,
+--          COUNT(*) FILTER (WHERE "executionStatus"='FAILED') AS failures
+--   FROM analytics.node_block_execution
+--   GROUP BY 1 ORDER BY executions DESC LIMIT 20;
+--
+--   -- Average LLM token usage per block
+--   SELECT "blockName",
+--          AVG("inputTokenCount") AS avg_input_tokens,
+--          AVG("outputTokenCount") AS avg_output_tokens
+--   FROM analytics.node_block_execution
+--   WHERE "llmCallCount" > 0
+--   GROUP BY 1 ORDER BY avg_input_tokens DESC;
+--
+--   -- Top failure reasons
+--   SELECT "blockName", "groupedErrorMessage", COUNT(*) AS count
+--   FROM analytics.node_block_execution
+--   WHERE "executionStatus" = 'FAILED'
+--   GROUP BY 1, 2 ORDER BY count DESC LIMIT 20;
+-- =============================================================
+
+SELECT
+    ne."id"                                                            AS id,
+    ne."agentGraphExecutionId"                                         AS agentGraphExecutionId,
+    ne."agentNodeId"                                                   AS agentNodeId,
+    CAST(ne."executionStatus" AS TEXT)                                 AS executionStatus,
+    ne."addedTime"                                                     AS addedTime,
+    ne."queuedTime"                                                    AS queuedTime,
+    ne."startedTime"                                                   AS startedTime,
+    ne."endedTime"                                                     AS endedTime,
+    (ne."stats"::jsonb->>'input_size')::bigint                         AS inputSize,
+    (ne."stats"::jsonb->>'output_size')::bigint                        AS outputSize,
+    (ne."stats"::jsonb->>'walltime')::numeric                          AS walltime,
+    (ne."stats"::jsonb->>'cputime')::numeric                           AS cputime,
+    (ne."stats"::jsonb->>'llm_retry_count')::int                       AS llmRetryCount,
+    (ne."stats"::jsonb->>'llm_call_count')::int                        AS llmCallCount,
+    (ne."stats"::jsonb->>'input_token_count')::bigint                  AS inputTokenCount,
+    (ne."stats"::jsonb->>'output_token_count')::bigint                 AS outputTokenCount,
+    b."name"                                                           AS blockName,
+    b."id"                                                             AS blockId,
+    REGEXP_REPLACE(
+        REGEXP_REPLACE(
+            TRIM(BOTH '"' FROM eio."data"::text),
+            '(https?://)([A-Za-z0-9.-]+)(:[0-9]+)?(/[^\s]*)?',
+            '\1\2/...', 'gi'
+        ),
+        '[a-zA-Z0-9_:-]*\d[a-zA-Z0-9_:-]*', '*', 'g'
+    )                                                                  AS groupedErrorMessage,
+    eio."data"                                                         AS errorMessage
+FROM platform."AgentNodeExecution" ne
+LEFT JOIN platform."AgentNode" nd
+       ON ne."agentNodeId" = nd."id"
+LEFT JOIN platform."AgentBlock" b
+       ON nd."agentBlockId" = b."id"
+LEFT JOIN platform."AgentNodeExecutionInputOutput" eio
+       ON eio."referencedByOutputExecId" = ne."id"
+      AND eio."name" = 'error'
+      AND ne."executionStatus" = 'FAILED'
+WHERE ne."addedTime" > CURRENT_DATE - INTERVAL '90 days'

autogpt_platform/analytics/queries/retention_agent.sql

@@ -0,0 +1,97 @@
+-- =============================================================
+-- View: analytics.retention_agent
+-- Looker source alias: ds35  |  Charts: 2
+-- =============================================================
+-- DESCRIPTION
+--   Weekly cohort retention broken down per individual agent.
+--   Cohort = week of a user's first use of THAT specific agent.
+--   Tells you which agents keep users coming back vs. one-shot
+--   use. Only includes cohorts from the last 180 days.
+--
+-- SOURCE TABLES
+--   platform.AgentGraphExecution  — Execution records (user × agent × time)
+--   platform.AgentGraph           — Agent names
+--
+-- OUTPUT COLUMNS
+--   agent_id            TEXT   Agent graph UUID
+--   agent_label         TEXT   'AgentName [first8chars]'
+--   agent_label_n       TEXT   'AgentName [first8chars] (n=total_users)'
+--   cohort_week_start   DATE   Week users first ran this agent
+--   cohort_label        TEXT   ISO week label
+--   cohort_label_n      TEXT   ISO week label with cohort size
+--   user_lifetime_week  INT    Weeks since first use of this agent
+--   cohort_users        BIGINT Users in this cohort for this agent
+--   active_users        BIGINT Users who ran the agent again in week k
+--   retention_rate      FLOAT  active_users / cohort_users
+--   cohort_users_w0     BIGINT cohort_users only at week 0 (safe to SUM)
+--   agent_total_users   BIGINT Total users across all cohorts for this agent
+--
+-- EXAMPLE QUERIES
+--   -- Best-retained agents at week 2
+--   SELECT agent_label, AVG(retention_rate) AS w2_retention
+--   FROM analytics.retention_agent
+--   WHERE user_lifetime_week = 2 AND cohort_users >= 10
+--   GROUP BY 1 ORDER BY w2_retention DESC LIMIT 10;
+--
+--   -- Agents with most unique users
+--   SELECT DISTINCT agent_label, agent_total_users
+--   FROM analytics.retention_agent
+--   ORDER BY agent_total_users DESC LIMIT 20;
+-- =============================================================
+
+WITH params AS (SELECT 12::int AS max_weeks, (CURRENT_DATE - INTERVAL '180 days') AS cohort_start),
+events AS (
+  SELECT e."userId"::text AS user_id, e."agentGraphId" AS agent_id,
+         e."createdAt"::timestamptz AS created_at,
+         DATE_TRUNC('week', e."createdAt")::date AS week_start
+  FROM platform."AgentGraphExecution" e
+),
+first_use AS (
+  SELECT user_id, agent_id, MIN(created_at) AS first_use_at,
+         DATE_TRUNC('week', MIN(created_at))::date AS cohort_week_start
+  FROM events GROUP BY 1,2
+  HAVING MIN(created_at) >= (SELECT cohort_start FROM params)
+),
+activity_weeks AS (SELECT DISTINCT user_id, agent_id, week_start FROM events),
+user_week_age AS (
+  SELECT aw.user_id, aw.agent_id, fu.cohort_week_start,
+         ((aw.week_start - DATE_TRUNC('week',fu.first_use_at)::date)/7)::int AS user_lifetime_week
+  FROM activity_weeks aw JOIN first_use fu USING (user_id, agent_id)
+  WHERE aw.week_start >= DATE_TRUNC('week',fu.first_use_at)::date
+),
+active_counts AS (
+  SELECT agent_id, cohort_week_start, user_lifetime_week, COUNT(DISTINCT user_id) AS active_users
+  FROM user_week_age WHERE user_lifetime_week >= 0 GROUP BY 1,2,3
+),
+cohort_sizes AS (
+  SELECT agent_id, cohort_week_start, COUNT(DISTINCT user_id) AS cohort_users FROM first_use GROUP BY 1,2
+),
+cohort_caps AS (
+  SELECT cs.agent_id, cs.cohort_week_start, cs.cohort_users,
+         LEAST((SELECT max_weeks FROM params),
+               GREATEST(0,((DATE_TRUNC('week',CURRENT_DATE)::date-cs.cohort_week_start)/7)::int)) AS cap_weeks
+  FROM cohort_sizes cs
+),
+grid AS (
+  SELECT cc.agent_id, cc.cohort_week_start, gs AS user_lifetime_week, cc.cohort_users
+  FROM cohort_caps cc CROSS JOIN LATERAL generate_series(0, cc.cap_weeks) gs
+),
+agent_names AS (SELECT DISTINCT ON (g."id") g."id" AS agent_id, g."name" AS agent_name FROM platform."AgentGraph" g ORDER BY g."id", g."version" DESC),
+agent_total_users AS (SELECT agent_id, SUM(cohort_users) AS agent_total_users FROM cohort_sizes GROUP BY 1)
+SELECT
+  g.agent_id,
+  COALESCE(an.agent_name,'(unnamed)')||' ['||LEFT(g.agent_id::text,8)||']'  AS agent_label,
+  COALESCE(an.agent_name,'(unnamed)')||' ['||LEFT(g.agent_id::text,8)||'] (n='||COALESCE(atu.agent_total_users,0)||')' AS agent_label_n,
+  g.cohort_week_start,
+  TO_CHAR(g.cohort_week_start,'IYYY-"W"IW')                               AS cohort_label,
+  TO_CHAR(g.cohort_week_start,'IYYY-"W"IW')||' (n='||g.cohort_users||')'  AS cohort_label_n,
+  g.user_lifetime_week, g.cohort_users,
+  COALESCE(ac.active_users,0)                                              AS active_users,
+  COALESCE(ac.active_users,0)::float / NULLIF(g.cohort_users,0)           AS retention_rate,
+  CASE WHEN g.user_lifetime_week=0 THEN g.cohort_users ELSE 0 END         AS cohort_users_w0,
+  COALESCE(atu.agent_total_users,0)                                        AS agent_total_users
+FROM grid g
+LEFT JOIN active_counts     ac  ON ac.agent_id=g.agent_id AND ac.cohort_week_start=g.cohort_week_start AND ac.user_lifetime_week=g.user_lifetime_week
+LEFT JOIN agent_names       an  ON an.agent_id=g.agent_id
+LEFT JOIN agent_total_users atu ON atu.agent_id=g.agent_id
+ORDER BY agent_label, g.cohort_week_start, g.user_lifetime_week;

autogpt_platform/analytics/queries/retention_execution_daily.sql

@@ -0,0 +1,81 @@
+-- =============================================================
+-- View: analytics.retention_execution_daily
+-- Looker source alias: ds111  |  Charts: 1
+-- =============================================================
+-- DESCRIPTION
+--   Daily cohort retention based on agent executions.
+--   Cohort anchor = day of user's FIRST ever execution.
+--   Only includes cohorts from the last 90 days, up to day 30.
+--   Great for early engagement analysis (did users run another
+--   agent the next day?).
+--
+-- SOURCE TABLES
+--   platform.AgentGraphExecution  — Execution records
+--
+-- OUTPUT COLUMNS
+--   Same pattern as retention_login_daily.
+--   cohort_day_start = day of first execution (not first login)
+--
+-- EXAMPLE QUERIES
+--   -- Day-3 execution retention
+--   SELECT cohort_label, retention_rate_bounded AS d3_retention
+--   FROM analytics.retention_execution_daily
+--   WHERE user_lifetime_day = 3 ORDER BY cohort_day_start;
+-- =============================================================
+
+WITH params AS (SELECT 30::int AS max_days, (CURRENT_DATE - INTERVAL '90 days') AS cohort_start),
+events AS (
+  SELECT e."userId"::text AS user_id, e."createdAt"::timestamptz AS created_at,
+         DATE_TRUNC('day', e."createdAt")::date AS day_start
+  FROM platform."AgentGraphExecution" e WHERE e."userId" IS NOT NULL
+),
+first_exec AS (
+  SELECT user_id, MIN(created_at) AS first_exec_at,
+         DATE_TRUNC('day', MIN(created_at))::date AS cohort_day_start
+  FROM events GROUP BY 1
+  HAVING MIN(created_at) >= (SELECT cohort_start FROM params)
+),
+activity_days AS (SELECT DISTINCT user_id, day_start FROM events),
+user_day_age AS (
+  SELECT ad.user_id, fe.cohort_day_start,
+         (ad.day_start - DATE_TRUNC('day',fe.first_exec_at)::date)::int AS user_lifetime_day
+  FROM activity_days ad JOIN first_exec fe USING (user_id)
+  WHERE ad.day_start >= DATE_TRUNC('day',fe.first_exec_at)::date
+),
+bounded_counts AS (
+  SELECT cohort_day_start, user_lifetime_day, COUNT(DISTINCT user_id) AS active_users_bounded
+  FROM user_day_age WHERE user_lifetime_day >= 0 GROUP BY 1,2
+),
+last_active AS (
+  SELECT cohort_day_start, user_id, MAX(user_lifetime_day) AS last_active_day FROM user_day_age GROUP BY 1,2
+),
+unbounded_counts AS (
+  SELECT la.cohort_day_start, gs AS user_lifetime_day, COUNT(*) AS retained_users_unbounded
+  FROM last_active la
+  CROSS JOIN LATERAL generate_series(0, LEAST(la.last_active_day,(SELECT max_days FROM params))) gs
+  GROUP BY 1,2
+),
+cohort_sizes AS (SELECT cohort_day_start, COUNT(DISTINCT user_id) AS cohort_users FROM first_exec GROUP BY 1),
+cohort_caps AS (
+  SELECT cs.cohort_day_start, cs.cohort_users,
+         LEAST((SELECT max_days FROM params), GREATEST(0,(CURRENT_DATE-cs.cohort_day_start)::int)) AS cap_days
+  FROM cohort_sizes cs
+),
+grid AS (
+  SELECT cc.cohort_day_start, gs AS user_lifetime_day, cc.cohort_users
+  FROM cohort_caps cc CROSS JOIN LATERAL generate_series(0, cc.cap_days) gs
+)
+SELECT
+  g.cohort_day_start,
+  TO_CHAR(g.cohort_day_start,'YYYY-MM-DD')                                AS cohort_label,
+  TO_CHAR(g.cohort_day_start,'YYYY-MM-DD')||' (n='||g.cohort_users||')'   AS cohort_label_n,
+  g.user_lifetime_day, g.cohort_users,
+  COALESCE(b.active_users_bounded,0)     AS active_users_bounded,
+  COALESCE(u.retained_users_unbounded,0) AS retained_users_unbounded,
+  CASE WHEN g.cohort_users>0 THEN COALESCE(b.active_users_bounded,0)::float/g.cohort_users END    AS retention_rate_bounded,
+  CASE WHEN g.cohort_users>0 THEN COALESCE(u.retained_users_unbounded,0)::float/g.cohort_users END AS retention_rate_unbounded,
+  CASE WHEN g.user_lifetime_day=0 THEN g.cohort_users ELSE 0 END          AS cohort_users_d0
+FROM grid g
+LEFT JOIN bounded_counts   b ON b.cohort_day_start=g.cohort_day_start AND b.user_lifetime_day=g.user_lifetime_day
+LEFT JOIN unbounded_counts u ON u.cohort_day_start=g.cohort_day_start AND u.user_lifetime_day=g.user_lifetime_day
+ORDER BY g.cohort_day_start, g.user_lifetime_day;

autogpt_platform/analytics/queries/retention_execution_weekly.sql

@@ -0,0 +1,81 @@
+-- =============================================================
+-- View: analytics.retention_execution_weekly
+-- Looker source alias: ds92  |  Charts: 2
+-- =============================================================
+-- DESCRIPTION
+--   Weekly cohort retention based on agent executions.
+--   Cohort anchor = week of user's FIRST ever agent execution
+--   (not first login). Only includes cohorts from the last 180 days.
+--   Useful when you care about product engagement, not just visits.
+--
+-- SOURCE TABLES
+--   platform.AgentGraphExecution  — Execution records
+--
+-- OUTPUT COLUMNS
+--   Same pattern as retention_login_weekly.
+--   cohort_week_start = week of first execution (not first login)
+--
+-- EXAMPLE QUERIES
+--   -- Week-2 execution retention
+--   SELECT cohort_label, retention_rate_bounded
+--   FROM analytics.retention_execution_weekly
+--   WHERE user_lifetime_week = 2 ORDER BY cohort_week_start;
+-- =============================================================
+
+WITH params AS (SELECT 12::int AS max_weeks, (CURRENT_DATE - INTERVAL '180 days') AS cohort_start),
+events AS (
+  SELECT e."userId"::text AS user_id, e."createdAt"::timestamptz AS created_at,
+         DATE_TRUNC('week', e."createdAt")::date AS week_start
+  FROM platform."AgentGraphExecution" e WHERE e."userId" IS NOT NULL
+),
+first_exec AS (
+  SELECT user_id, MIN(created_at) AS first_exec_at,
+         DATE_TRUNC('week', MIN(created_at))::date AS cohort_week_start
+  FROM events GROUP BY 1
+  HAVING MIN(created_at) >= (SELECT cohort_start FROM params)
+),
+activity_weeks AS (SELECT DISTINCT user_id, week_start FROM events),
+user_week_age AS (
+  SELECT aw.user_id, fe.cohort_week_start,
+         ((aw.week_start - DATE_TRUNC('week',fe.first_exec_at)::date)/7)::int AS user_lifetime_week
+  FROM activity_weeks aw JOIN first_exec fe USING (user_id)
+  WHERE aw.week_start >= DATE_TRUNC('week',fe.first_exec_at)::date
+),
+bounded_counts AS (
+  SELECT cohort_week_start, user_lifetime_week, COUNT(DISTINCT user_id) AS active_users_bounded
+  FROM user_week_age WHERE user_lifetime_week >= 0 GROUP BY 1,2
+),
+last_active AS (
+  SELECT cohort_week_start, user_id, MAX(user_lifetime_week) AS last_active_week FROM user_week_age GROUP BY 1,2
+),
+unbounded_counts AS (
+  SELECT la.cohort_week_start, gs AS user_lifetime_week, COUNT(*) AS retained_users_unbounded
+  FROM last_active la
+  CROSS JOIN LATERAL generate_series(0, LEAST(la.last_active_week,(SELECT max_weeks FROM params))) gs
+  GROUP BY 1,2
+),
+cohort_sizes AS (SELECT cohort_week_start, COUNT(DISTINCT user_id) AS cohort_users FROM first_exec GROUP BY 1),
+cohort_caps AS (
+  SELECT cs.cohort_week_start, cs.cohort_users,
+         LEAST((SELECT max_weeks FROM params),
+               GREATEST(0,((DATE_TRUNC('week',CURRENT_DATE)::date-cs.cohort_week_start)/7)::int)) AS cap_weeks
+  FROM cohort_sizes cs
+),
+grid AS (
+  SELECT cc.cohort_week_start, gs AS user_lifetime_week, cc.cohort_users
+  FROM cohort_caps cc CROSS JOIN LATERAL generate_series(0, cc.cap_weeks) gs
+)
+SELECT
+  g.cohort_week_start,
+  TO_CHAR(g.cohort_week_start,'IYYY-"W"IW')                               AS cohort_label,
+  TO_CHAR(g.cohort_week_start,'IYYY-"W"IW')||' (n='||g.cohort_users||')'  AS cohort_label_n,
+  g.user_lifetime_week, g.cohort_users,
+  COALESCE(b.active_users_bounded,0)     AS active_users_bounded,
+  COALESCE(u.retained_users_unbounded,0) AS retained_users_unbounded,
+  CASE WHEN g.cohort_users>0 THEN COALESCE(b.active_users_bounded,0)::float/g.cohort_users END    AS retention_rate_bounded,
+  CASE WHEN g.cohort_users>0 THEN COALESCE(u.retained_users_unbounded,0)::float/g.cohort_users END AS retention_rate_unbounded,
+  CASE WHEN g.user_lifetime_week=0 THEN g.cohort_users ELSE 0 END         AS cohort_users_w0
+FROM grid g
+LEFT JOIN bounded_counts   b ON b.cohort_week_start=g.cohort_week_start AND b.user_lifetime_week=g.user_lifetime_week
+LEFT JOIN unbounded_counts u ON u.cohort_week_start=g.cohort_week_start AND u.user_lifetime_week=g.user_lifetime_week
+ORDER BY g.cohort_week_start, g.user_lifetime_week;

autogpt_platform/analytics/queries/retention_login_daily.sql

@@ -0,0 +1,94 @@
+-- =============================================================
+-- View: analytics.retention_login_daily
+-- Looker source alias: ds112  |  Charts: 1
+-- =============================================================
+-- DESCRIPTION
+--   Daily cohort retention based on login sessions.
+--   Same logic as retention_login_weekly but at day granularity,
+--   showing up to day 30 for cohorts from the last 90 days.
+--   Useful for analysing early activation (days 1-7) in detail.
+--
+-- SOURCE TABLES
+--   auth.sessions  — Login session records
+--
+-- OUTPUT COLUMNS (same pattern as retention_login_weekly)
+--   cohort_day_start          DATE     First day the cohort logged in
+--   cohort_label              TEXT     Date string (e.g. '2025-03-01')
+--   cohort_label_n            TEXT     Date + cohort size (e.g. '2025-03-01 (n=12)')
+--   user_lifetime_day         INT      Days since first login (0 = signup day)
+--   cohort_users              BIGINT   Total users in cohort
+--   active_users_bounded      BIGINT   Users active on exactly day k
+--   retained_users_unbounded  BIGINT   Users active any time on/after day k
+--   retention_rate_bounded    FLOAT    bounded / cohort_users
+--   retention_rate_unbounded  FLOAT    unbounded / cohort_users
+--   cohort_users_d0           BIGINT   cohort_users only at day 0, else 0 (safe to SUM)
+--
+-- EXAMPLE QUERIES
+--   -- Day-1 retention rate (came back next day)
+--   SELECT cohort_label, retention_rate_bounded AS d1_retention
+--   FROM analytics.retention_login_daily
+--   WHERE user_lifetime_day = 1 ORDER BY cohort_day_start;
+--
+--   -- Average retention curve across all cohorts
+--   SELECT user_lifetime_day,
+--          SUM(active_users_bounded)::float / NULLIF(SUM(cohort_users_d0), 0) AS avg_retention
+--   FROM analytics.retention_login_daily
+--   GROUP BY 1 ORDER BY 1;
+-- =============================================================
+
+WITH params AS (SELECT 30::int AS max_days, (CURRENT_DATE - INTERVAL '90 days')::date AS cohort_start),
+events AS (
+  SELECT s.user_id::text AS user_id, s.created_at::timestamptz AS created_at,
+         DATE_TRUNC('day', s.created_at)::date AS day_start
+  FROM auth.sessions s WHERE s.user_id IS NOT NULL
+),
+first_login AS (
+  SELECT user_id, MIN(created_at) AS first_login_time,
+         DATE_TRUNC('day', MIN(created_at))::date AS cohort_day_start
+  FROM events GROUP BY 1
+  HAVING MIN(created_at) >= (SELECT cohort_start FROM params)
+),
+activity_days AS (SELECT DISTINCT user_id, day_start FROM events),
+user_day_age AS (
+  SELECT ad.user_id, fl.cohort_day_start,
+         (ad.day_start - DATE_TRUNC('day', fl.first_login_time)::date)::int AS user_lifetime_day
+  FROM activity_days ad JOIN first_login fl USING (user_id)
+  WHERE ad.day_start >= DATE_TRUNC('day', fl.first_login_time)::date
+),
+bounded_counts AS (
+  SELECT cohort_day_start, user_lifetime_day, COUNT(DISTINCT user_id) AS active_users_bounded
+  FROM user_day_age WHERE user_lifetime_day >= 0 GROUP BY 1,2
+),
+last_active AS (
+  SELECT cohort_day_start, user_id, MAX(user_lifetime_day) AS last_active_day FROM user_day_age GROUP BY 1,2
+),
+unbounded_counts AS (
+  SELECT la.cohort_day_start, gs AS user_lifetime_day, COUNT(*) AS retained_users_unbounded
+  FROM last_active la
+  CROSS JOIN LATERAL generate_series(0, LEAST(la.last_active_day,(SELECT max_days FROM params))) gs
+  GROUP BY 1,2
+),
+cohort_sizes AS (SELECT cohort_day_start, COUNT(DISTINCT user_id) AS cohort_users FROM first_login GROUP BY 1),
+cohort_caps AS (
+  SELECT cs.cohort_day_start, cs.cohort_users,
+         LEAST((SELECT max_days FROM params), GREATEST(0,(CURRENT_DATE-cs.cohort_day_start)::int)) AS cap_days
+  FROM cohort_sizes cs
+),
+grid AS (
+  SELECT cc.cohort_day_start, gs AS user_lifetime_day, cc.cohort_users
+  FROM cohort_caps cc CROSS JOIN LATERAL generate_series(0, cc.cap_days) gs
+)
+SELECT
+  g.cohort_day_start,
+  TO_CHAR(g.cohort_day_start,'YYYY-MM-DD')                                  AS cohort_label,
+  TO_CHAR(g.cohort_day_start,'YYYY-MM-DD')||' (n='||g.cohort_users||')'     AS cohort_label_n,
+  g.user_lifetime_day, g.cohort_users,
+  COALESCE(b.active_users_bounded,0)     AS active_users_bounded,
+  COALESCE(u.retained_users_unbounded,0) AS retained_users_unbounded,
+  CASE WHEN g.cohort_users>0 THEN COALESCE(b.active_users_bounded,0)::float/g.cohort_users END    AS retention_rate_bounded,
+  CASE WHEN g.cohort_users>0 THEN COALESCE(u.retained_users_unbounded,0)::float/g.cohort_users END AS retention_rate_unbounded,
+  CASE WHEN g.user_lifetime_day=0 THEN g.cohort_users ELSE 0 END            AS cohort_users_d0
+FROM grid g
+LEFT JOIN bounded_counts   b ON b.cohort_day_start=g.cohort_day_start AND b.user_lifetime_day=g.user_lifetime_day
+LEFT JOIN unbounded_counts u ON u.cohort_day_start=g.cohort_day_start AND u.user_lifetime_day=g.user_lifetime_day
+ORDER BY g.cohort_day_start, g.user_lifetime_day;

autogpt_platform/analytics/queries/retention_login_onboarded_weekly.sql

@@ -0,0 +1,96 @@
+-- =============================================================
+-- View: analytics.retention_login_onboarded_weekly
+-- Looker source alias: ds101  |  Charts: 2
+-- =============================================================
+-- DESCRIPTION
+--   Weekly cohort retention from login sessions, restricted to
+--   users who "onboarded" — defined as running at least one
+--   agent within 365 days of their first login.
+--   Filters out users who signed up but never activated,
+--   giving a cleaner view of engaged-user retention.
+--
+-- SOURCE TABLES
+--   auth.sessions                  — Login session records
+--   platform.AgentGraphExecution   — Used to identify onboarders
+--
+-- OUTPUT COLUMNS
+--   Same as retention_login_weekly (cohort_week_start, user_lifetime_week,
+--   retention_rate_bounded, retention_rate_unbounded, etc.)
+--   Only difference: cohort is filtered to onboarded users only.
+--
+-- EXAMPLE QUERIES
+--   -- Compare week-4 retention: all users vs onboarded only
+--   SELECT 'all_users' AS segment, AVG(retention_rate_bounded) AS w4_retention
+--   FROM analytics.retention_login_weekly WHERE user_lifetime_week = 4
+--   UNION ALL
+--   SELECT 'onboarded', AVG(retention_rate_bounded)
+--   FROM analytics.retention_login_onboarded_weekly WHERE user_lifetime_week = 4;
+-- =============================================================
+
+WITH params AS (SELECT 12::int AS max_weeks, 365::int AS onboarding_window_days),
+events AS (
+  SELECT s.user_id::text AS user_id, s.created_at::timestamptz AS created_at,
+         DATE_TRUNC('week', s.created_at)::date AS week_start
+  FROM auth.sessions s WHERE s.user_id IS NOT NULL
+),
+first_login_all AS (
+  SELECT user_id, MIN(created_at) AS first_login_time,
+         DATE_TRUNC('week', MIN(created_at))::date AS cohort_week_start
+  FROM events GROUP BY 1
+),
+onboarders AS (
+  SELECT fl.user_id FROM first_login_all fl
+  WHERE EXISTS (
+    SELECT 1 FROM platform."AgentGraphExecution" e
+    WHERE e."userId"::text = fl.user_id
+      AND e."createdAt" >= fl.first_login_time
+      AND e."createdAt" < fl.first_login_time
+          + make_interval(days => (SELECT onboarding_window_days FROM params))
+  )
+),
+first_login AS (SELECT * FROM first_login_all WHERE user_id IN (SELECT user_id FROM onboarders)),
+activity_weeks AS (SELECT DISTINCT user_id, week_start FROM events),
+user_week_age AS (
+  SELECT aw.user_id, fl.cohort_week_start,
+         ((aw.week_start - DATE_TRUNC('week',fl.first_login_time)::date)/7)::int AS user_lifetime_week
+  FROM activity_weeks aw JOIN first_login fl USING (user_id)
+  WHERE aw.week_start >= DATE_TRUNC('week',fl.first_login_time)::date
+),
+bounded_counts AS (
+  SELECT cohort_week_start, user_lifetime_week, COUNT(DISTINCT user_id) AS active_users_bounded
+  FROM user_week_age WHERE user_lifetime_week >= 0 GROUP BY 1,2
+),
+last_active AS (
+  SELECT cohort_week_start, user_id, MAX(user_lifetime_week) AS last_active_week FROM user_week_age GROUP BY 1,2
+),
+unbounded_counts AS (
+  SELECT la.cohort_week_start, gs AS user_lifetime_week, COUNT(*) AS retained_users_unbounded
+  FROM last_active la
+  CROSS JOIN LATERAL generate_series(0, LEAST(la.last_active_week,(SELECT max_weeks FROM params))) gs
+  GROUP BY 1,2
+),
+cohort_sizes AS (SELECT cohort_week_start, COUNT(DISTINCT user_id) AS cohort_users FROM first_login GROUP BY 1),
+cohort_caps AS (
+  SELECT cs.cohort_week_start, cs.cohort_users,
+         LEAST((SELECT max_weeks FROM params),
+               GREATEST(0,((DATE_TRUNC('week',CURRENT_DATE)::date-cs.cohort_week_start)/7)::int)) AS cap_weeks
+  FROM cohort_sizes cs
+),
+grid AS (
+  SELECT cc.cohort_week_start, gs AS user_lifetime_week, cc.cohort_users
+  FROM cohort_caps cc CROSS JOIN LATERAL generate_series(0, cc.cap_weeks) gs
+)
+SELECT
+  g.cohort_week_start,
+  TO_CHAR(g.cohort_week_start,'IYYY-"W"IW')                               AS cohort_label,
+  TO_CHAR(g.cohort_week_start,'IYYY-"W"IW')||' (n='||g.cohort_users||')'  AS cohort_label_n,
+  g.user_lifetime_week, g.cohort_users,
+  COALESCE(b.active_users_bounded,0)     AS active_users_bounded,
+  COALESCE(u.retained_users_unbounded,0) AS retained_users_unbounded,
+  CASE WHEN g.cohort_users>0 THEN COALESCE(b.active_users_bounded,0)::float/g.cohort_users END    AS retention_rate_bounded,
+  CASE WHEN g.cohort_users>0 THEN COALESCE(u.retained_users_unbounded,0)::float/g.cohort_users END AS retention_rate_unbounded,
+  CASE WHEN g.user_lifetime_week=0 THEN g.cohort_users ELSE 0 END         AS cohort_users_w0
+FROM grid g
+LEFT JOIN bounded_counts   b ON b.cohort_week_start=g.cohort_week_start AND b.user_lifetime_week=g.user_lifetime_week
+LEFT JOIN unbounded_counts u ON u.cohort_week_start=g.cohort_week_start AND u.user_lifetime_week=g.user_lifetime_week
+ORDER BY g.cohort_week_start, g.user_lifetime_week;

autogpt_platform/analytics/queries/retention_login_weekly.sql

@@ -0,0 +1,103 @@
+-- =============================================================
+-- View: analytics.retention_login_weekly
+-- Looker source alias: ds83  |  Charts: 2
+-- =============================================================
+-- DESCRIPTION
+--   Weekly cohort retention based on login sessions.
+--   Users are grouped by the ISO week of their first ever login.
+--   For each cohort × lifetime-week combination, outputs both:
+--     - bounded rate: % active in exactly that week
+--     - unbounded rate: % who were ever active on or after that week
+--   Weeks are capped to the cohort's actual age (no future data points).
+--
+-- SOURCE TABLES
+--   auth.sessions  — Login session records
+--
+-- HOW TO READ THE OUTPUT
+--   cohort_week_start   The Monday of the week users first logged in
+--   user_lifetime_week  0 = signup week, 1 = one week later, etc.
+--   retention_rate_bounded   = active_users_bounded / cohort_users
+--   retention_rate_unbounded = retained_users_unbounded / cohort_users
+--
+-- OUTPUT COLUMNS
+--   cohort_week_start         DATE     First day of the cohort's signup week
+--   cohort_label              TEXT     ISO week label (e.g. '2025-W01')
+--   cohort_label_n            TEXT     ISO week label with cohort size (e.g. '2025-W01 (n=42)')
+--   user_lifetime_week        INT      Weeks since first login (0 = signup week)
+--   cohort_users              BIGINT   Total users in this cohort (denominator)
+--   active_users_bounded      BIGINT   Users active in exactly week k
+--   retained_users_unbounded  BIGINT   Users active any time on/after week k
+--   retention_rate_bounded    FLOAT    bounded active / cohort_users
+--   retention_rate_unbounded  FLOAT    unbounded retained / cohort_users
+--   cohort_users_w0           BIGINT   cohort_users only at week 0, else 0 (safe to SUM in pivot tables)
+--
+-- EXAMPLE QUERIES
+--   -- Week-1 retention rate per cohort
+--   SELECT cohort_label, retention_rate_bounded AS w1_retention
+--   FROM analytics.retention_login_weekly
+--   WHERE user_lifetime_week = 1
+--   ORDER BY cohort_week_start;
+--
+--   -- Overall average retention curve (all cohorts combined)
+--   SELECT user_lifetime_week,
+--          SUM(active_users_bounded)::float / NULLIF(SUM(cohort_users_w0), 0) AS avg_retention
+--   FROM analytics.retention_login_weekly
+--   GROUP BY 1 ORDER BY 1;
+-- =============================================================
+
+WITH params AS (SELECT 12::int AS max_weeks),
+events AS (
+  SELECT s.user_id::text AS user_id, s.created_at::timestamptz AS created_at,
+         DATE_TRUNC('week', s.created_at)::date AS week_start
+  FROM auth.sessions s WHERE s.user_id IS NOT NULL
+),
+first_login AS (
+  SELECT user_id, MIN(created_at) AS first_login_time,
+         DATE_TRUNC('week', MIN(created_at))::date AS cohort_week_start
+  FROM events GROUP BY 1
+),
+activity_weeks AS (SELECT DISTINCT user_id, week_start FROM events),
+user_week_age AS (
+  SELECT aw.user_id, fl.cohort_week_start,
+         ((aw.week_start - DATE_TRUNC('week', fl.first_login_time)::date) / 7)::int AS user_lifetime_week
+  FROM activity_weeks aw JOIN first_login fl USING (user_id)
+  WHERE aw.week_start >= DATE_TRUNC('week', fl.first_login_time)::date
+),
+bounded_counts AS (
+  SELECT cohort_week_start, user_lifetime_week, COUNT(DISTINCT user_id) AS active_users_bounded
+  FROM user_week_age WHERE user_lifetime_week >= 0 GROUP BY 1,2
+),
+last_active AS (
+  SELECT cohort_week_start, user_id, MAX(user_lifetime_week) AS last_active_week FROM user_week_age GROUP BY 1,2
+),
+unbounded_counts AS (
+  SELECT la.cohort_week_start, gs AS user_lifetime_week, COUNT(*) AS retained_users_unbounded
+  FROM last_active la
+  CROSS JOIN LATERAL generate_series(0, LEAST(la.last_active_week,(SELECT max_weeks FROM params))) gs
+  GROUP BY 1,2
+),
+cohort_sizes AS (SELECT cohort_week_start, COUNT(DISTINCT user_id) AS cohort_users FROM first_login GROUP BY 1),
+cohort_caps AS (
+  SELECT cs.cohort_week_start, cs.cohort_users,
+         LEAST((SELECT max_weeks FROM params),
+               GREATEST(0,((DATE_TRUNC('week',CURRENT_DATE)::date - cs.cohort_week_start)/7)::int)) AS cap_weeks
+  FROM cohort_sizes cs
+),
+grid AS (
+  SELECT cc.cohort_week_start, gs AS user_lifetime_week, cc.cohort_users
+  FROM cohort_caps cc CROSS JOIN LATERAL generate_series(0, cc.cap_weeks) gs
+)
+SELECT
+  g.cohort_week_start,
+  TO_CHAR(g.cohort_week_start,'IYYY-"W"IW')                                    AS cohort_label,
+  TO_CHAR(g.cohort_week_start,'IYYY-"W"IW')||' (n='||g.cohort_users||')'       AS cohort_label_n,
+  g.user_lifetime_week, g.cohort_users,
+  COALESCE(b.active_users_bounded,0)     AS active_users_bounded,
+  COALESCE(u.retained_users_unbounded,0) AS retained_users_unbounded,
+  CASE WHEN g.cohort_users>0 THEN COALESCE(b.active_users_bounded,0)::float/g.cohort_users END    AS retention_rate_bounded,
+  CASE WHEN g.cohort_users>0 THEN COALESCE(u.retained_users_unbounded,0)::float/g.cohort_users END AS retention_rate_unbounded,
+  CASE WHEN g.user_lifetime_week=0 THEN g.cohort_users ELSE 0 END               AS cohort_users_w0
+FROM grid g
+LEFT JOIN bounded_counts   b ON b.cohort_week_start=g.cohort_week_start AND b.user_lifetime_week=g.user_lifetime_week
+LEFT JOIN unbounded_counts u ON u.cohort_week_start=g.cohort_week_start AND u.user_lifetime_week=g.user_lifetime_week
+ORDER BY g.cohort_week_start, g.user_lifetime_week

autogpt_platform/analytics/queries/user_block_spending.sql

@@ -0,0 +1,71 @@
+-- =============================================================
+-- View: analytics.user_block_spending
+-- Looker source alias: ds6  |  Charts: 5
+-- =============================================================
+-- DESCRIPTION
+--   One row per credit transaction (last 90 days).
+--   Shows how users spend credits broken down by block type,
+--   LLM provider and model.  Joins node execution stats for
+--   token-level detail.
+--
+-- SOURCE TABLES
+--   platform.CreditTransaction   — Credit debit/credit records
+--   platform.AgentNodeExecution  — Node execution stats (for token counts)
+--
+-- OUTPUT COLUMNS
+--   transactionKey        TEXT         Unique transaction identifier
+--   userId                TEXT         User who was charged
+--   amount                DECIMAL      Credit amount (positive = credit, negative = debit)
+--   negativeAmount        DECIMAL      amount * -1 (convenience for spend charts)
+--   transactionType       TEXT         Transaction type (e.g. 'USAGE', 'REFUND', 'TOP_UP')
+--   transactionTime       TIMESTAMPTZ  When the transaction was recorded
+--   blockId               TEXT         Block UUID that triggered the spend
+--   blockName             TEXT         Human-readable block name
+--   llm_provider          TEXT         LLM provider (e.g. 'openai', 'anthropic')
+--   llm_model             TEXT         Model name (e.g. 'gpt-4o', 'claude-3-5-sonnet')
+--   node_exec_id          TEXT         Linked node execution UUID
+--   llm_call_count        INT          LLM API calls made in that execution
+--   llm_retry_count       INT          LLM retries in that execution
+--   llm_input_token_count INT          Input tokens consumed
+--   llm_output_token_count INT         Output tokens produced
+--
+-- WINDOW
+--   Rolling 90 days (createdAt > CURRENT_DATE - 90 days)
+--
+-- EXAMPLE QUERIES
+--   -- Total spend per user (last 90 days)
+--   SELECT "userId", SUM("negativeAmount") AS total_spent
+--   FROM analytics.user_block_spending
+--   WHERE "transactionType" = 'USAGE'
+--   GROUP BY 1 ORDER BY total_spent DESC;
+--
+--   -- Spend by LLM provider + model
+--   SELECT "llm_provider", "llm_model",
+--          SUM("negativeAmount") AS total_cost,
+--          SUM("llm_input_token_count") AS input_tokens,
+--          SUM("llm_output_token_count") AS output_tokens
+--   FROM analytics.user_block_spending
+--   WHERE "llm_provider" IS NOT NULL
+--   GROUP BY 1, 2 ORDER BY total_cost DESC;
+-- =============================================================
+
+SELECT
+    c."transactionKey"                                        AS transactionKey,
+    c."userId"                                                AS userId,
+    c."amount"                                                AS amount,
+    c."amount" * -1                                           AS negativeAmount,
+    c."type"                                                  AS transactionType,
+    c."createdAt"                                             AS transactionTime,
+    c.metadata->>'block_id'                                   AS blockId,
+    c.metadata->>'block'                                      AS blockName,
+    c.metadata->'input'->'credentials'->>'provider'           AS llm_provider,
+    c.metadata->'input'->>'model'                             AS llm_model,
+    c.metadata->>'node_exec_id'                               AS node_exec_id,
+    (ne."stats"->>'llm_call_count')::int                       AS llm_call_count,
+    (ne."stats"->>'llm_retry_count')::int                      AS llm_retry_count,
+    (ne."stats"->>'input_token_count')::int                    AS llm_input_token_count,
+    (ne."stats"->>'output_token_count')::int                   AS llm_output_token_count
+FROM platform."CreditTransaction" c
+LEFT JOIN platform."AgentNodeExecution" ne
+       ON (c.metadata->>'node_exec_id') = ne."id"::text
+WHERE c."createdAt" > CURRENT_DATE - INTERVAL '90 days'

autogpt_platform/analytics/queries/user_onboarding.sql

@@ -0,0 +1,45 @@
+-- =============================================================
+-- View: analytics.user_onboarding
+-- Looker source alias: ds68  |  Charts: 3
+-- =============================================================
+-- DESCRIPTION
+--   One row per user onboarding record.  Contains the user's
+--   stated usage reason, selected integrations, completed
+--   onboarding steps and optional first agent selection.
+--   Full history (no date filter) since onboarding happens
+--   once per user.
+--
+-- SOURCE TABLES
+--   platform.UserOnboarding  — Onboarding state per user
+--
+-- OUTPUT COLUMNS
+--   id                            TEXT         Onboarding record UUID
+--   createdAt                     TIMESTAMPTZ  When onboarding started
+--   updatedAt                     TIMESTAMPTZ  Last update to onboarding state
+--   usageReason                   TEXT         Why user signed up (e.g. 'work', 'personal')
+--   integrations                  TEXT[]       Array of integration names the user selected
+--   userId                        TEXT         User UUID
+--   completedSteps                TEXT[]       Array of onboarding step enums completed
+--   selectedStoreListingVersionId TEXT         First marketplace agent the user chose (if any)
+--
+-- EXAMPLE QUERIES
+--   -- Usage reason breakdown
+--   SELECT "usageReason", COUNT(*) FROM analytics.user_onboarding GROUP BY 1;
+--
+--   -- Completion rate per step
+--   SELECT step, COUNT(*) AS users_completed
+--   FROM analytics.user_onboarding
+--   CROSS JOIN LATERAL UNNEST("completedSteps") AS step
+--   GROUP BY 1 ORDER BY users_completed DESC;
+-- =============================================================
+
+SELECT
+    id,
+    "createdAt",
+    "updatedAt",
+    "usageReason",
+    integrations,
+    "userId",
+    "completedSteps",
+    "selectedStoreListingVersionId"
+FROM platform."UserOnboarding"

autogpt_platform/analytics/queries/user_onboarding_funnel.sql

@@ -0,0 +1,100 @@
+-- =============================================================
+-- View: analytics.user_onboarding_funnel
+-- Looker source alias: ds74  |  Charts: 1
+-- =============================================================
+-- DESCRIPTION
+--   Pre-aggregated onboarding funnel showing how many users
+--   completed each step and the drop-off percentage from the
+--   previous step.  One row per onboarding step (all 22 steps
+--   always present, even with 0 completions — prevents sparse
+--   gaps from making LAG compare the wrong predecessors).
+--
+-- SOURCE TABLES
+--   platform.UserOnboarding  — Onboarding records with completedSteps array
+--
+-- OUTPUT COLUMNS
+--   step             TEXT     Onboarding step enum name (e.g. 'WELCOME', 'CONGRATS')
+--   step_order       INT      Numeric position in the funnel (1=first, 22=last)
+--   users_completed  BIGINT   Distinct users who completed this step
+--   pct_from_prev    NUMERIC  % of users from the previous step who reached this one
+--
+-- STEP ORDER
+--   1  WELCOME               9  MARKETPLACE_VISIT     17  SCHEDULE_AGENT
+--   2  USAGE_REASON         10  MARKETPLACE_ADD_AGENT  18  RUN_AGENTS
+--   3  INTEGRATIONS         11  MARKETPLACE_RUN_AGENT  19  RUN_3_DAYS
+--   4  AGENT_CHOICE         12  BUILDER_OPEN           20  TRIGGER_WEBHOOK
+--   5  AGENT_NEW_RUN        13  BUILDER_SAVE_AGENT     21  RUN_14_DAYS
+--   6  AGENT_INPUT          14  BUILDER_RUN_AGENT      22  RUN_AGENTS_100
+--   7  CONGRATS             15  VISIT_COPILOT
+--   8  GET_RESULTS          16  RE_RUN_AGENT
+--
+-- WINDOW
+--   Users who started onboarding in the last 90 days
+--
+-- EXAMPLE QUERIES
+--   -- Full funnel
+--   SELECT * FROM analytics.user_onboarding_funnel ORDER BY step_order;
+--
+--   -- Biggest drop-off point
+--   SELECT step, pct_from_prev FROM analytics.user_onboarding_funnel
+--   ORDER BY pct_from_prev ASC LIMIT 3;
+-- =============================================================
+
+WITH all_steps AS (
+  -- Complete ordered grid of all 22 steps so zero-completion steps
+  -- are always present, keeping LAG comparisons correct.
+  SELECT step_name, step_order
+  FROM (VALUES
+    ('WELCOME',               1),
+    ('USAGE_REASON',          2),
+    ('INTEGRATIONS',          3),
+    ('AGENT_CHOICE',          4),
+    ('AGENT_NEW_RUN',         5),
+    ('AGENT_INPUT',           6),
+    ('CONGRATS',              7),
+    ('GET_RESULTS',           8),
+    ('MARKETPLACE_VISIT',     9),
+    ('MARKETPLACE_ADD_AGENT', 10),
+    ('MARKETPLACE_RUN_AGENT', 11),
+    ('BUILDER_OPEN',          12),
+    ('BUILDER_SAVE_AGENT',    13),
+    ('BUILDER_RUN_AGENT',     14),
+    ('VISIT_COPILOT',         15),
+    ('RE_RUN_AGENT',          16),
+    ('SCHEDULE_AGENT',        17),
+    ('RUN_AGENTS',            18),
+    ('RUN_3_DAYS',            19),
+    ('TRIGGER_WEBHOOK',       20),
+    ('RUN_14_DAYS',           21),
+    ('RUN_AGENTS_100',        22)
+  ) AS t(step_name, step_order)
+),
+raw AS (
+  SELECT
+      u."userId",
+      step_txt::text AS step
+  FROM platform."UserOnboarding" u
+  CROSS JOIN LATERAL UNNEST(u."completedSteps") AS step_txt
+  WHERE u."createdAt" >= CURRENT_DATE - INTERVAL '90 days'
+),
+step_counts AS (
+  SELECT step, COUNT(DISTINCT "userId") AS users_completed
+  FROM raw GROUP BY step
+),
+funnel AS (
+  SELECT
+      a.step_name                          AS step,
+      a.step_order,
+      COALESCE(sc.users_completed, 0)      AS users_completed,
+      ROUND(
+        100.0 * COALESCE(sc.users_completed, 0)
+        / NULLIF(
+            LAG(COALESCE(sc.users_completed, 0)) OVER (ORDER BY a.step_order),
+            0
+          ),
+        2
+      )                                    AS pct_from_prev
+  FROM all_steps a
+  LEFT JOIN step_counts sc ON sc.step = a.step_name
+)
+SELECT * FROM funnel ORDER BY step_order

autogpt_platform/analytics/queries/user_onboarding_integration.sql

@@ -0,0 +1,41 @@
+-- =============================================================
+-- View: analytics.user_onboarding_integration
+-- Looker source alias: ds75  |  Charts: 1
+-- =============================================================
+-- DESCRIPTION
+--   Pre-aggregated count of users who selected each integration
+--   during onboarding.  One row per integration type, sorted
+--   by popularity.
+--
+-- SOURCE TABLES
+--   platform.UserOnboarding  — integrations array column
+--
+-- OUTPUT COLUMNS
+--   integration            TEXT    Integration name (e.g. 'github', 'slack', 'notion')
+--   users_with_integration BIGINT  Distinct users who selected this integration
+--
+-- WINDOW
+--   Users who started onboarding in the last 90 days
+--
+-- EXAMPLE QUERIES
+--   -- Full integration popularity ranking
+--   SELECT * FROM analytics.user_onboarding_integration;
+--
+--   -- Top 5 integrations
+--   SELECT * FROM analytics.user_onboarding_integration LIMIT 5;
+-- =============================================================
+
+WITH exploded AS (
+  SELECT
+      u."userId" AS user_id,
+      UNNEST(u."integrations") AS integration
+  FROM platform."UserOnboarding" u
+  WHERE u."createdAt" >= CURRENT_DATE - INTERVAL '90 days'
+)
+SELECT
+    integration,
+    COUNT(DISTINCT user_id) AS users_with_integration
+FROM exploded
+WHERE integration IS NOT NULL AND integration <> ''
+GROUP BY integration
+ORDER BY users_with_integration DESC

autogpt_platform/analytics/queries/users_activities.sql

@@ -0,0 +1,145 @@
+-- =============================================================
+-- View: analytics.users_activities
+-- Looker source alias: ds56  |  Charts: 5
+-- =============================================================
+-- DESCRIPTION
+--   One row per user with lifetime activity summary.
+--   Joins login sessions with agent graphs, executions and
+--   node-level runs to give a full picture of how engaged
+--   each user is.  Includes a convenience flag for 7-day
+--   activation (did the user return at least 7 days after
+--   their first login?).
+--
+-- SOURCE TABLES
+--   auth.sessions                    — Login/session records
+--   platform.AgentGraph              — Graphs (agents) built by the user
+--   platform.AgentGraphExecution     — Agent run history
+--   platform.AgentNodeExecution      — Individual block execution history
+--
+-- PERFORMANCE NOTE
+--   Each CTE aggregates its own table independently by userId.
+--   This avoids the fan-out that occurs when driving every join
+--   from user_logins across the two largest tables
+--   (AgentGraphExecution and AgentNodeExecution).
+--
+-- OUTPUT COLUMNS
+--   user_id                   TEXT         Supabase user UUID
+--   first_login_time          TIMESTAMPTZ  First ever session created_at
+--   last_login_time           TIMESTAMPTZ  Most recent session created_at
+--   last_visit_time           TIMESTAMPTZ  Max of last refresh or login
+--   last_agent_save_time      TIMESTAMPTZ  Last time user saved an agent graph
+--   agent_count               BIGINT       Number of distinct active graphs built (0 if none)
+--   first_agent_run_time      TIMESTAMPTZ  First ever graph execution
+--   last_agent_run_time       TIMESTAMPTZ  Most recent graph execution
+--   unique_agent_runs         BIGINT       Distinct agent graphs ever run (0 if none)
+--   agent_runs                BIGINT       Total graph execution count (0 if none)
+--   node_execution_count      BIGINT       Total node executions across all runs
+--   node_execution_failed     BIGINT       Node executions with FAILED status
+--   node_execution_completed  BIGINT       Node executions with COMPLETED status
+--   node_execution_terminated BIGINT       Node executions with TERMINATED status
+--   node_execution_queued     BIGINT       Node executions with QUEUED status
+--   node_execution_running    BIGINT       Node executions with RUNNING status
+--   is_active_after_7d        INT          1=returned after day 7, 0=did not, NULL=too early to tell
+--   node_execution_incomplete BIGINT       Node executions with INCOMPLETE status
+--   node_execution_review     BIGINT       Node executions with REVIEW status
+--
+-- EXAMPLE QUERIES
+--   -- Users who ran at least one agent and returned after 7 days
+--   SELECT COUNT(*) FROM analytics.users_activities
+--   WHERE agent_runs > 0 AND is_active_after_7d = 1;
+--
+--   -- Top 10 most active users by agent runs
+--   SELECT user_id, agent_runs, node_execution_count
+--   FROM analytics.users_activities
+--   ORDER BY agent_runs DESC LIMIT 10;
+--
+--   -- 7-day activation rate
+--   SELECT
+--     SUM(CASE WHEN is_active_after_7d = 1 THEN 1 ELSE 0 END)::float
+--     / NULLIF(COUNT(CASE WHEN is_active_after_7d IS NOT NULL THEN 1 END), 0)
+--     AS activation_rate
+--   FROM analytics.users_activities;
+-- =============================================================
+
+WITH user_logins AS (
+  SELECT
+    user_id::text                                    AS user_id,
+    MIN(created_at)                                  AS first_login_time,
+    MAX(created_at)                                  AS last_login_time,
+    GREATEST(
+      MAX(refreshed_at)::timestamptz,
+      MAX(created_at)::timestamptz
+    )                                                AS last_visit_time
+  FROM auth.sessions
+  GROUP BY user_id
+),
+user_agents AS (
+  -- Aggregate AgentGraph directly by userId (no fan-out from user_logins)
+  SELECT
+    "userId"::text                AS user_id,
+    MAX("updatedAt")              AS last_agent_save_time,
+    COUNT(DISTINCT "id")          AS agent_count
+  FROM platform."AgentGraph"
+  WHERE "isActive"
+  GROUP BY "userId"
+),
+user_graph_runs AS (
+  -- Aggregate AgentGraphExecution directly by userId
+  SELECT
+    "userId"::text                        AS user_id,
+    MIN("createdAt")                      AS first_agent_run_time,
+    MAX("createdAt")                      AS last_agent_run_time,
+    COUNT(DISTINCT "agentGraphId")        AS unique_agent_runs,
+    COUNT("id")                           AS agent_runs
+  FROM platform."AgentGraphExecution"
+  GROUP BY "userId"
+),
+user_node_runs AS (
+  -- Aggregate AgentNodeExecution directly; resolve userId via a
+  -- single join to AgentGraphExecution instead of fanning out from
+  -- user_logins through both large tables.
+  SELECT
+    g."userId"::text                                                   AS user_id,
+    COUNT(*)                                                           AS node_execution_count,
+    COUNT(*) FILTER (WHERE n."executionStatus" = 'FAILED')             AS node_execution_failed,
+    COUNT(*) FILTER (WHERE n."executionStatus" = 'COMPLETED')          AS node_execution_completed,
+    COUNT(*) FILTER (WHERE n."executionStatus" = 'TERMINATED')         AS node_execution_terminated,
+    COUNT(*) FILTER (WHERE n."executionStatus" = 'QUEUED')             AS node_execution_queued,
+    COUNT(*) FILTER (WHERE n."executionStatus" = 'RUNNING')            AS node_execution_running,
+    COUNT(*) FILTER (WHERE n."executionStatus" = 'INCOMPLETE')         AS node_execution_incomplete,
+    COUNT(*) FILTER (WHERE n."executionStatus" = 'REVIEW')             AS node_execution_review
+  FROM platform."AgentNodeExecution" n
+  JOIN platform."AgentGraphExecution" g
+    ON g."id" = n."agentGraphExecutionId"
+  GROUP BY g."userId"
+)
+SELECT
+  ul.user_id,
+  ul.first_login_time,
+  ul.last_login_time,
+  ul.last_visit_time,
+  ua.last_agent_save_time,
+  COALESCE(ua.agent_count, 0)             AS agent_count,
+  gr.first_agent_run_time,
+  gr.last_agent_run_time,
+  COALESCE(gr.unique_agent_runs, 0)       AS unique_agent_runs,
+  COALESCE(gr.agent_runs, 0)              AS agent_runs,
+  COALESCE(nr.node_execution_count, 0)      AS node_execution_count,
+  COALESCE(nr.node_execution_failed, 0)     AS node_execution_failed,
+  COALESCE(nr.node_execution_completed, 0)  AS node_execution_completed,
+  COALESCE(nr.node_execution_terminated, 0) AS node_execution_terminated,
+  COALESCE(nr.node_execution_queued, 0)     AS node_execution_queued,
+  COALESCE(nr.node_execution_running, 0)    AS node_execution_running,
+  CASE
+    WHEN ul.first_login_time < NOW() - INTERVAL '7 days'
+     AND ul.last_visit_time  >= ul.first_login_time + INTERVAL '7 days' THEN 1
+    WHEN ul.first_login_time < NOW() - INTERVAL '7 days'
+     AND ul.last_visit_time  <  ul.first_login_time + INTERVAL '7 days' THEN 0
+    ELSE NULL
+  END AS is_active_after_7d,
+  COALESCE(nr.node_execution_incomplete, 0) AS node_execution_incomplete,
+  COALESCE(nr.node_execution_review, 0)     AS node_execution_review
+FROM user_logins ul
+LEFT JOIN user_agents     ua ON ul.user_id = ua.user_id
+LEFT JOIN user_graph_runs gr ON ul.user_id = gr.user_id
+LEFT JOIN user_node_runs  nr ON ul.user_id = nr.user_id

autogpt_platform/backend/.env.default

@@ -37,6 +37,10 @@ JWT_VERIFY_KEY=your-super-secret-jwt-token-with-at-least-32-characters-long
 ENCRYPTION_KEY=dvziYgz0KSK8FENhju0ZYi8-fRTfAdlz6YLhdB_jhNw=
 UNSUBSCRIBE_SECRET_KEY=HlP8ivStJjmbf6NKi78m_3FnOogut0t5ckzjsIqeaio=
 
+## ===== SIGNUP / INVITE GATE ===== ##
+# Set to true to require an invite before users can sign up
+ENABLE_INVITE_GATE=false
+
 ## ===== IMPORTANT OPTIONAL CONFIGURATION ===== ##
 # Platform URLs (set these for webhooks and OAuth to work)
 PLATFORM_BASE_URL=http://localhost:8000

autogpt_platform/backend/CLAUDE.md

@@ -58,10 +58,31 @@ poetry run pytest path/to/test.py --snapshot-update
 - **Authentication**: JWT-based with Supabase integration
 - **Security**: Cache protection middleware prevents sensitive data caching in browsers/proxies
 
+## Code Style
+
+- **Top-level imports only** — no local/inner imports (lazy imports only for heavy optional deps like `openpyxl`)
+- **No duck typing** — no `hasattr`/`getattr`/`isinstance` for type dispatch; use typed interfaces/unions/protocols
+- **Pydantic models** over dataclass/namedtuple/dict for structured data
+- **No linter suppressors** — no `# type: ignore`, `# noqa`, `# pyright: ignore`; fix the type/code
+- **List comprehensions** over manual loop-and-append
+- **Early return** — guard clauses first, avoid deep nesting
+- **Lazy `%s` logging** — `logger.info("Processing %s items", count)` not `logger.info(f"Processing {count} items")`
+- **Sanitize error paths** — `os.path.basename()` in error messages to avoid leaking directory structure
+- **TOCTOU awareness** — avoid check-then-act patterns for file access and credit charging
+- **`Security()` vs `Depends()`** — use `Security()` for auth deps to get proper OpenAPI security spec
+- **Redis pipelines** — `transaction=True` for atomicity on multi-step operations
+- **`max(0, value)` guards** — for computed values that should never be negative
+- **SSE protocol** — `data:` lines for frontend-parsed events (must match Zod schema), `: comment` lines for heartbeats/status
+- **File length** — keep files under ~300 lines; if a file grows beyond this, split by responsibility (e.g. extract helpers, models, or a sub-module into a new file). Never keep appending to a long file.
+- **Function length** — keep functions under ~40 lines; extract named helpers when a function grows longer. Long functions are a sign of mixed concerns, not complexity.
+
 ## Testing Approach
 
 - Uses pytest with snapshot testing for API responses
 - Test files are colocated with source files (`*_test.py`)
+- Mock at boundaries — mock where the symbol is **used**, not where it's **defined**
+- After refactoring, update mock targets to match new module paths
+- Use `AsyncMock` for async functions (`from unittest.mock import AsyncMock`)
 
 ## Database Schema
 

autogpt_platform/backend/backend/api/features/admin/model.py

@@ -1,8 +1,17 @@
-from pydantic import BaseModel
+from __future__ import annotations
+
+from datetime import datetime
+from typing import TYPE_CHECKING, Any, Literal, Optional
+
+import prisma.enums
+from pydantic import BaseModel, EmailStr
 
 from backend.data.model import UserTransaction
 from backend.util.models import Pagination
 
+if TYPE_CHECKING:
+    from backend.data.invited_user import BulkInvitedUsersResult, InvitedUserRecord
+
 
 class UserHistoryResponse(BaseModel):
     """Response model for listings with version history"""
@@ -14,3 +23,70 @@ class UserHistoryResponse(BaseModel):
 class AddUserCreditsResponse(BaseModel):
     new_balance: int
     transaction_key: str
+
+
+class CreateInvitedUserRequest(BaseModel):
+    email: EmailStr
+    name: Optional[str] = None
+
+
+class InvitedUserResponse(BaseModel):
+    id: str
+    email: str
+    status: prisma.enums.InvitedUserStatus
+    auth_user_id: Optional[str] = None
+    name: Optional[str] = None
+    tally_understanding: Optional[dict[str, Any]] = None
+    tally_status: prisma.enums.TallyComputationStatus
+    tally_computed_at: Optional[datetime] = None
+    tally_error: Optional[str] = None
+    created_at: datetime
+    updated_at: datetime
+
+    @classmethod
+    def from_record(cls, record: InvitedUserRecord) -> InvitedUserResponse:
+        return cls.model_validate(record.model_dump())
+
+
+class InvitedUsersResponse(BaseModel):
+    invited_users: list[InvitedUserResponse]
+    pagination: Pagination
+
+
+class BulkInvitedUserRowResponse(BaseModel):
+    row_number: int
+    email: Optional[str] = None
+    name: Optional[str] = None
+    status: Literal["CREATED", "SKIPPED", "ERROR"]
+    message: str
+    invited_user: Optional[InvitedUserResponse] = None
+
+
+class BulkInvitedUsersResponse(BaseModel):
+    created_count: int
+    skipped_count: int
+    error_count: int
+    results: list[BulkInvitedUserRowResponse]
+
+    @classmethod
+    def from_result(cls, result: BulkInvitedUsersResult) -> BulkInvitedUsersResponse:
+        return cls(
+            created_count=result.created_count,
+            skipped_count=result.skipped_count,
+            error_count=result.error_count,
+            results=[
+                BulkInvitedUserRowResponse(
+                    row_number=row.row_number,
+                    email=row.email,
+                    name=row.name,
+                    status=row.status,
+                    message=row.message,
+                    invited_user=(
+                        InvitedUserResponse.from_record(row.invited_user)
+                        if row.invited_user is not None
+                        else None
+                    ),
+                )
+                for row in result.results
+            ],
+        )

autogpt_platform/backend/backend/api/features/admin/user_admin_routes.py

@@ -0,0 +1,137 @@
+import logging
+import math
+
+from autogpt_libs.auth import get_user_id, requires_admin_user
+from fastapi import APIRouter, File, Query, Security, UploadFile
+
+from backend.data.invited_user import (
+    bulk_create_invited_users_from_file,
+    create_invited_user,
+    list_invited_users,
+    retry_invited_user_tally,
+    revoke_invited_user,
+)
+from backend.data.tally import mask_email
+from backend.util.models import Pagination
+
+from .model import (
+    BulkInvitedUsersResponse,
+    CreateInvitedUserRequest,
+    InvitedUserResponse,
+    InvitedUsersResponse,
+)
+
+logger = logging.getLogger(__name__)
+
+
+router = APIRouter(
+    prefix="/admin",
+    tags=["users", "admin"],
+    dependencies=[Security(requires_admin_user)],
+)
+
+
+@router.get(
+    "/invited-users",
+    response_model=InvitedUsersResponse,
+    summary="List Invited Users",
+)
+async def get_invited_users(
+    admin_user_id: str = Security(get_user_id),
+    page: int = Query(1, ge=1),
+    page_size: int = Query(50, ge=1, le=200),
+) -> InvitedUsersResponse:
+    logger.info("Admin user %s requested invited users", admin_user_id)
+    invited_users, total = await list_invited_users(page=page, page_size=page_size)
+    return InvitedUsersResponse(
+        invited_users=[InvitedUserResponse.from_record(iu) for iu in invited_users],
+        pagination=Pagination(
+            total_items=total,
+            total_pages=max(1, math.ceil(total / page_size)),
+            current_page=page,
+            page_size=page_size,
+        ),
+    )
+
+
+@router.post(
+    "/invited-users",
+    response_model=InvitedUserResponse,
+    summary="Create Invited User",
+)
+async def create_invited_user_route(
+    request: CreateInvitedUserRequest,
+    admin_user_id: str = Security(get_user_id),
+) -> InvitedUserResponse:
+    logger.info(
+        "Admin user %s creating invited user for %s",
+        admin_user_id,
+        mask_email(request.email),
+    )
+    invited_user = await create_invited_user(request.email, request.name)
+    logger.info(
+        "Admin user %s created invited user %s",
+        admin_user_id,
+        invited_user.id,
+    )
+    return InvitedUserResponse.from_record(invited_user)
+
+
+@router.post(
+    "/invited-users/bulk",
+    response_model=BulkInvitedUsersResponse,
+    summary="Bulk Create Invited Users",
+    operation_id="postV2BulkCreateInvitedUsers",
+)
+async def bulk_create_invited_users_route(
+    file: UploadFile = File(...),
+    admin_user_id: str = Security(get_user_id),
+) -> BulkInvitedUsersResponse:
+    logger.info(
+        "Admin user %s bulk invited users from %s",
+        admin_user_id,
+        file.filename or "<unnamed>",
+    )
+    content = await file.read()
+    result = await bulk_create_invited_users_from_file(file.filename, content)
+    return BulkInvitedUsersResponse.from_result(result)
+
+
+@router.post(
+    "/invited-users/{invited_user_id}/revoke",
+    response_model=InvitedUserResponse,
+    summary="Revoke Invited User",
+)
+async def revoke_invited_user_route(
+    invited_user_id: str,
+    admin_user_id: str = Security(get_user_id),
+) -> InvitedUserResponse:
+    logger.info(
+        "Admin user %s revoking invited user %s", admin_user_id, invited_user_id
+    )
+    invited_user = await revoke_invited_user(invited_user_id)
+    logger.info("Admin user %s revoked invited user %s", admin_user_id, invited_user_id)
+    return InvitedUserResponse.from_record(invited_user)
+
+
+@router.post(
+    "/invited-users/{invited_user_id}/retry-tally",
+    response_model=InvitedUserResponse,
+    summary="Retry Invited User Tally",
+)
+async def retry_invited_user_tally_route(
+    invited_user_id: str,
+    admin_user_id: str = Security(get_user_id),
+) -> InvitedUserResponse:
+    logger.info(
+        "Admin user %s retrying Tally seed for invited user %s",
+        admin_user_id,
+        invited_user_id,
+    )
+    invited_user = await retry_invited_user_tally(invited_user_id)
+    logger.info(
+        "Admin user %s retried Tally seed for invited user %s",
+        admin_user_id,
+        invited_user_id,
+    )
+    return InvitedUserResponse.from_record(invited_user)

autogpt_platform/backend/backend/api/features/admin/user_admin_routes_test.py

@@ -0,0 +1,168 @@
+from datetime import datetime, timezone
+from unittest.mock import AsyncMock
+
+import fastapi
+import fastapi.testclient
+import prisma.enums
+import pytest
+import pytest_mock
+from autogpt_libs.auth.jwt_utils import get_jwt_payload
+
+from backend.data.invited_user import (
+    BulkInvitedUserRowResult,
+    BulkInvitedUsersResult,
+    InvitedUserRecord,
+)
+
+from .user_admin_routes import router as user_admin_router
+
+app = fastapi.FastAPI()
+app.include_router(user_admin_router)
+
+client = fastapi.testclient.TestClient(app)
+
+
+@pytest.fixture(autouse=True)
+def setup_app_admin_auth(mock_jwt_admin):
+    app.dependency_overrides[get_jwt_payload] = mock_jwt_admin["get_jwt_payload"]
+    yield
+    app.dependency_overrides.clear()
+
+
+def _sample_invited_user() -> InvitedUserRecord:
+    now = datetime.now(timezone.utc)
+    return InvitedUserRecord(
+        id="invite-1",
+        email="invited@example.com",
+        status=prisma.enums.InvitedUserStatus.INVITED,
+        auth_user_id=None,
+        name="Invited User",
+        tally_understanding=None,
+        tally_status=prisma.enums.TallyComputationStatus.PENDING,
+        tally_computed_at=None,
+        tally_error=None,
+        created_at=now,
+        updated_at=now,
+    )
+
+
+def _sample_bulk_invited_users_result() -> BulkInvitedUsersResult:
+    return BulkInvitedUsersResult(
+        created_count=1,
+        skipped_count=1,
+        error_count=0,
+        results=[
+            BulkInvitedUserRowResult(
+                row_number=1,
+                email="invited@example.com",
+                name=None,
+                status="CREATED",
+                message="Invite created",
+                invited_user=_sample_invited_user(),
+            ),
+            BulkInvitedUserRowResult(
+                row_number=2,
+                email="duplicate@example.com",
+                name=None,
+                status="SKIPPED",
+                message="An invited user with this email already exists",
+                invited_user=None,
+            ),
+        ],
+    )
+
+
+def test_get_invited_users(
+    mocker: pytest_mock.MockerFixture,
+) -> None:
+    mocker.patch(
+        "backend.api.features.admin.user_admin_routes.list_invited_users",
+        AsyncMock(return_value=([_sample_invited_user()], 1)),
+    )
+
+    response = client.get("/admin/invited-users")
+
+    assert response.status_code == 200
+    data = response.json()
+    assert len(data["invited_users"]) == 1
+    assert data["invited_users"][0]["email"] == "invited@example.com"
+    assert data["invited_users"][0]["status"] == "INVITED"
+    assert data["pagination"]["total_items"] == 1
+    assert data["pagination"]["current_page"] == 1
+    assert data["pagination"]["page_size"] == 50
+
+
+def test_create_invited_user(
+    mocker: pytest_mock.MockerFixture,
+) -> None:
+    mocker.patch(
+        "backend.api.features.admin.user_admin_routes.create_invited_user",
+        AsyncMock(return_value=_sample_invited_user()),
+    )
+
+    response = client.post(
+        "/admin/invited-users",
+        json={"email": "invited@example.com", "name": "Invited User"},
+    )
+
+    assert response.status_code == 200
+    data = response.json()
+    assert data["email"] == "invited@example.com"
+    assert data["name"] == "Invited User"
+
+
+def test_bulk_create_invited_users(
+    mocker: pytest_mock.MockerFixture,
+) -> None:
+    mocker.patch(
+        "backend.api.features.admin.user_admin_routes.bulk_create_invited_users_from_file",
+        AsyncMock(return_value=_sample_bulk_invited_users_result()),
+    )
+
+    response = client.post(
+        "/admin/invited-users/bulk",
+        files={
+            "file": ("invites.txt", b"invited@example.com\nduplicate@example.com\n")
+        },
+    )
+
+    assert response.status_code == 200
+    data = response.json()
+    assert data["created_count"] == 1
+    assert data["skipped_count"] == 1
+    assert data["results"][0]["status"] == "CREATED"
+    assert data["results"][1]["status"] == "SKIPPED"
+
+
+def test_revoke_invited_user(
+    mocker: pytest_mock.MockerFixture,
+) -> None:
+    revoked = _sample_invited_user().model_copy(
+        update={"status": prisma.enums.InvitedUserStatus.REVOKED}
+    )
+    mocker.patch(
+        "backend.api.features.admin.user_admin_routes.revoke_invited_user",
+        AsyncMock(return_value=revoked),
+    )
+
+    response = client.post("/admin/invited-users/invite-1/revoke")
+
+    assert response.status_code == 200
+    assert response.json()["status"] == "REVOKED"
+
+
+def test_retry_invited_user_tally(
+    mocker: pytest_mock.MockerFixture,
+) -> None:
+    retried = _sample_invited_user().model_copy(
+        update={"tally_status": prisma.enums.TallyComputationStatus.RUNNING}
+    )
+    mocker.patch(
+        "backend.api.features.admin.user_admin_routes.retry_invited_user_tally",
+        AsyncMock(return_value=retried),
+    )
+
+    response = client.post("/admin/invited-users/invite-1/retry-tally")
+
+    assert response.status_code == 200
+    assert response.json()["tally_status"] == "RUNNING"

autogpt_platform/backend/backend/api/features/chat/routes.py

@@ -54,6 +54,7 @@ from backend.copilot.tools.models import (
 )
 from backend.copilot.tracking import track_user_message
 from backend.data.redis_client import get_redis_async
+from backend.data.understanding import get_business_understanding
 from backend.data.workspace import get_or_create_workspace
 from backend.util.exceptions import NotFoundError
 
@@ -853,6 +854,36 @@ async def session_assign_user(
     return {"status": "ok"}
 
 
+# ========== Suggested Prompts ==========
+
+
+class SuggestedPromptsResponse(BaseModel):
+    """Response model for user-specific suggested prompts."""
+
+    prompts: list[str]
+
+
+@router.get(
+    "/suggested-prompts",
+    dependencies=[Security(auth.requires_user)],
+)
+async def get_suggested_prompts(
+    user_id: Annotated[str, Security(auth.get_user_id)],
+) -> SuggestedPromptsResponse:
+    """
+    Get LLM-generated suggested prompts for the authenticated user.
+
+    Returns personalized quick-action prompts based on the user's
+    business understanding. Returns an empty list if no custom prompts
+    are available.
+    """
+    understanding = await get_business_understanding(user_id)
+    if understanding is None:
+        return SuggestedPromptsResponse(prompts=[])
+
+    return SuggestedPromptsResponse(prompts=understanding.suggested_prompts)
+
+
 # ========== Configuration ==========
 
 

autogpt_platform/backend/backend/api/features/chat/routes_test.py

@@ -1,6 +1,6 @@
-"""Tests for chat API routes: session title update and file attachment validation."""
+"""Tests for chat API routes: session title update, file attachment validation, and suggested prompts."""
 
-from unittest.mock import AsyncMock
+from unittest.mock import AsyncMock, MagicMock
 
 import fastapi
 import fastapi.testclient
@@ -249,3 +249,62 @@ def test_file_ids_scoped_to_workspace(mocker: pytest_mock.MockFixture):
     call_kwargs = mock_prisma.find_many.call_args[1]
     assert call_kwargs["where"]["workspaceId"] == "my-workspace-id"
     assert call_kwargs["where"]["isDeleted"] is False
+
+
+# ─── Suggested prompts endpoint ──────────────────────────────────────
+
+
+def _mock_get_business_understanding(
+    mocker: pytest_mock.MockerFixture,
+    *,
+    return_value=None,
+):
+    """Mock get_business_understanding."""
+    return mocker.patch(
+        "backend.api.features.chat.routes.get_business_understanding",
+        new_callable=AsyncMock,
+        return_value=return_value,
+    )
+
+
+def test_suggested_prompts_returns_prompts(
+    mocker: pytest_mock.MockerFixture,
+    test_user_id: str,
+) -> None:
+    """User with understanding and prompts gets them back."""
+    mock_understanding = MagicMock()
+    mock_understanding.suggested_prompts = ["Do X", "Do Y", "Do Z"]
+    _mock_get_business_understanding(mocker, return_value=mock_understanding)
+
+    response = client.get("/suggested-prompts")
+
+    assert response.status_code == 200
+    assert response.json() == {"prompts": ["Do X", "Do Y", "Do Z"]}
+
+
+def test_suggested_prompts_no_understanding(
+    mocker: pytest_mock.MockerFixture,
+    test_user_id: str,
+) -> None:
+    """User with no understanding gets empty list."""
+    _mock_get_business_understanding(mocker, return_value=None)
+
+    response = client.get("/suggested-prompts")
+
+    assert response.status_code == 200
+    assert response.json() == {"prompts": []}
+
+
+def test_suggested_prompts_empty_prompts(
+    mocker: pytest_mock.MockerFixture,
+    test_user_id: str,
+) -> None:
+    """User with understanding but no prompts gets empty list."""
+    mock_understanding = MagicMock()
+    mock_understanding.suggested_prompts = []
+    _mock_get_business_understanding(mocker, return_value=mock_understanding)
+
+    response = client.get("/suggested-prompts")
+
+    assert response.status_code == 200
+    assert response.json() == {"prompts": []}

autogpt_platform/backend/backend/api/features/v1.py

@@ -55,6 +55,7 @@ from backend.data.credit import (
     set_auto_top_up,
 )
 from backend.data.graph import GraphSettings
+from backend.data.invited_user import get_or_activate_user
 from backend.data.model import CredentialsMetaInput, UserOnboarding
 from backend.data.notifications import NotificationPreference, NotificationPreferenceDTO
 from backend.data.onboarding import (
@@ -70,7 +71,6 @@ from backend.data.onboarding import (
     update_user_onboarding,
 )
 from backend.data.user import (
-    get_or_create_user,
     get_user_by_id,
     get_user_notification_preference,
     update_user_email,
@@ -136,12 +136,10 @@ _tally_background_tasks: set[asyncio.Task] = set()
     dependencies=[Security(requires_user)],
 )
 async def get_or_create_user_route(user_data: dict = Security(get_jwt_payload)):
-    user = await get_or_create_user(user_data)
+    user = await get_or_activate_user(user_data)
 
-    # Fire-and-forget: populate business understanding from Tally form.
-    # We use created_at proximity instead of an is_new flag because
-    # get_or_create_user is cached — a separate is_new return value would be
-    # unreliable on repeated calls within the cache TTL.
+    # Fire-and-forget: backfill Tally understanding when invite pre-seeding did
+    # not produce a stored result before first activation.
     age_seconds = (datetime.now(timezone.utc) - user.created_at).total_seconds()
     if age_seconds < 30:
         try:
@@ -165,7 +163,8 @@ async def get_or_create_user_route(user_data: dict = Security(get_jwt_payload)):
     dependencies=[Security(requires_user)],
 )
 async def update_user_email_route(
-    user_id: Annotated[str, Security(get_user_id)], email: str = Body(...)
+    user_id: Annotated[str, Security(get_user_id)],
+    email: str = Body(...),
 ) -> dict[str, str]:
     await update_user_email(user_id, email)
 
@@ -179,10 +178,16 @@ async def update_user_email_route(
     dependencies=[Security(requires_user)],
 )
 async def get_user_timezone_route(
-    user_data: dict = Security(get_jwt_payload),
+    user_id: Annotated[str, Security(get_user_id)],
 ) -> TimezoneResponse:
     """Get user timezone setting."""
-    user = await get_or_create_user(user_data)
+    try:
+        user = await get_user_by_id(user_id)
+    except ValueError:
+        raise HTTPException(
+            status_code=HTTP_404_NOT_FOUND,
+            detail="User not found. Please complete activation via /auth/user first.",
+        )
     return TimezoneResponse(timezone=user.timezone)
 
 
@@ -193,7 +198,8 @@ async def get_user_timezone_route(
     dependencies=[Security(requires_user)],
 )
 async def update_user_timezone_route(
-    user_id: Annotated[str, Security(get_user_id)], request: UpdateTimezoneRequest
+    user_id: Annotated[str, Security(get_user_id)],
+    request: UpdateTimezoneRequest,
 ) -> TimezoneResponse:
     """Update user timezone. The timezone should be a valid IANA timezone identifier."""
     user = await update_user_timezone(user_id, str(request.timezone))

autogpt_platform/backend/backend/api/features/v1_test.py

@@ -51,7 +51,7 @@ def test_get_or_create_user_route(
     }
 
     mocker.patch(
-        "backend.api.features.v1.get_or_create_user",
+        "backend.api.features.v1.get_or_activate_user",
         return_value=mock_user,
     )
 

autogpt_platform/backend/backend/api/rest_api.py

@@ -19,6 +19,7 @@ from prisma.errors import PrismaError
 import backend.api.features.admin.credit_admin_routes
 import backend.api.features.admin.execution_analytics_routes
 import backend.api.features.admin.store_admin_routes
+import backend.api.features.admin.user_admin_routes
 import backend.api.features.builder
 import backend.api.features.builder.routes
 import backend.api.features.chat.routes as chat_routes
@@ -311,6 +312,11 @@ app.include_router(
     tags=["v2", "admin"],
     prefix="/api/executions",
 )
+app.include_router(
+    backend.api.features.admin.user_admin_routes.router,
+    tags=["v2", "admin"],
+    prefix="/api/users",
+)
 app.include_router(
     backend.api.features.executions.review.routes.router,
     tags=["v2", "executions", "review"],

autogpt_platform/backend/backend/copilot/config.py

@@ -115,7 +115,7 @@ class ChatConfig(BaseSettings):
         description="E2B sandbox template to use for copilot sessions.",
     )
     e2b_sandbox_timeout: int = Field(
-        default=10800,  # 3 hours — wall-clock timeout, not idle; explicit pause is primary
+        default=300,  # 5 min safety net — explicit per-turn pause is the primary mechanism
         description="E2B sandbox running-time timeout (seconds). "
         "E2B timeout is wall-clock (not idle). Explicit per-turn pause is the primary "
         "mechanism; this is the safety net.",

autogpt_platform/backend/backend/copilot/prompting.py

@@ -52,6 +52,11 @@ Examples:
 You can embed a reference inside any string argument, or use it as the entire
 value.  Multiple references in one argument are all expanded.
 
+**Type coercion**: The platform automatically coerces expanded string values
+to match the block's expected input types.  For example, if a block expects
+`list[list[str]]` and you pass a string containing a JSON array (e.g. from
+an @@agptfile: expansion), the string will be parsed into the correct type.
+
 
 ### Sub-agent tasks
 - When using the Task tool, NEVER set `run_in_background` to true.

autogpt_platform/backend/backend/copilot/sdk/compaction.py

@@ -11,7 +11,7 @@ persistence, and the ``CompactionTracker`` state machine.
 import asyncio
 import logging
 import uuid
-from collections.abc import Callable
+from dataclasses import dataclass, field
 
 from ..constants import COMPACTION_DONE_MSG, COMPACTION_TOOL_NAME
 from ..model import ChatMessage, ChatSession
@@ -27,6 +27,19 @@ from ..response_model import (
 logger = logging.getLogger(__name__)
 
 
+@dataclass
+class CompactionResult:
+    """Result of emit_end_if_ready — bundles events with compaction metadata.
+
+    Eliminates the need for separate ``compaction_just_ended`` checks,
+    preventing TOCTOU races between the emit call and the flag read.
+    """
+
+    events: list[StreamBaseResponse] = field(default_factory=list)
+    just_ended: bool = False
+    transcript_path: str = ""
+
+
 # ---------------------------------------------------------------------------
 # Event builders (private — use CompactionTracker or compaction_events)
 # ---------------------------------------------------------------------------
@@ -177,11 +190,22 @@ class CompactionTracker:
         self._start_emitted = False
         self._done = False
         self._tool_call_id = ""
+        self._transcript_path: str = ""
 
-    @property
-    def on_compact(self) -> Callable[[], None]:
-        """Callback for the PreCompact hook."""
-        return self._compact_start.set
+    def on_compact(self, transcript_path: str = "") -> None:
+        """Callback for the PreCompact hook. Stores transcript_path."""
+        if (
+            self._transcript_path
+            and transcript_path
+            and self._transcript_path != transcript_path
+        ):
+            logger.warning(
+                "[Compaction] Overwriting transcript_path %s -> %s",
+                self._transcript_path,
+                transcript_path,
+            )
+        self._transcript_path = transcript_path
+        self._compact_start.set()
 
     # ------------------------------------------------------------------
     # Pre-query compaction
@@ -201,6 +225,7 @@ class CompactionTracker:
         self._done = False
         self._start_emitted = False
         self._tool_call_id = ""
+        self._transcript_path = ""
 
     def emit_start_if_ready(self) -> list[StreamBaseResponse]:
         """If the PreCompact hook fired, emit start events (spinning tool)."""
@@ -211,15 +236,20 @@ class CompactionTracker:
             return _start_events(self._tool_call_id)
         return []
 
-    async def emit_end_if_ready(self, session: ChatSession) -> list[StreamBaseResponse]:
-        """If compaction is in progress, emit end events and persist."""
+    async def emit_end_if_ready(self, session: ChatSession) -> CompactionResult:
+        """If compaction is in progress, emit end events and persist.
+
+        Returns a ``CompactionResult`` with ``just_ended=True`` and the
+        captured ``transcript_path`` when a compaction cycle completes.
+        This avoids a separate flag check (TOCTOU-safe).
+        """
         # Yield so pending hook tasks can set compact_start
         await asyncio.sleep(0)
 
         if self._done:
-            return []
+            return CompactionResult()
         if not self._start_emitted and not self._compact_start.is_set():
-            return []
+            return CompactionResult()
 
         if self._start_emitted:
             # Close the open spinner
@@ -232,8 +262,12 @@ class CompactionTracker:
                 COMPACTION_DONE_MSG, tool_call_id=persist_id
             )
 
+        transcript_path = self._transcript_path
         self._compact_start.clear()
         self._start_emitted = False
         self._done = True
+        self._transcript_path = ""
         _persist(session, persist_id, COMPACTION_DONE_MSG)
-        return done_events
+        return CompactionResult(
+            events=done_events, just_ended=True, transcript_path=transcript_path
+        )

autogpt_platform/backend/backend/copilot/sdk/compaction_test.py

@@ -195,10 +195,11 @@ class TestCompactionTracker:
         session = _make_session()
         tracker.on_compact()
         tracker.emit_start_if_ready()
-        evts = await tracker.emit_end_if_ready(session)
-        assert len(evts) == 2
-        assert isinstance(evts[0], StreamToolOutputAvailable)
-        assert isinstance(evts[1], StreamFinishStep)
+        result = await tracker.emit_end_if_ready(session)
+        assert result.just_ended is True
+        assert len(result.events) == 2
+        assert isinstance(result.events[0], StreamToolOutputAvailable)
+        assert isinstance(result.events[1], StreamFinishStep)
         # Should persist
         assert len(session.messages) == 2
 
@@ -210,28 +211,32 @@ class TestCompactionTracker:
         session = _make_session()
         tracker.on_compact()
         # Don't call emit_start_if_ready
-        evts = await tracker.emit_end_if_ready(session)
-        assert len(evts) == 5  # Full self-contained event
-        assert isinstance(evts[0], StreamStartStep)
+        result = await tracker.emit_end_if_ready(session)
+        assert result.just_ended is True
+        assert len(result.events) == 5  # Full self-contained event
+        assert isinstance(result.events[0], StreamStartStep)
         assert len(session.messages) == 2
 
     @pytest.mark.asyncio
-    async def test_emit_end_no_op_when_done(self):
+    async def test_emit_end_no_op_when_no_new_compaction(self):
         tracker = CompactionTracker()
         session = _make_session()
         tracker.on_compact()
         tracker.emit_start_if_ready()
-        await tracker.emit_end_if_ready(session)
-        # Second call should be no-op
-        evts = await tracker.emit_end_if_ready(session)
-        assert evts == []
+        result1 = await tracker.emit_end_if_ready(session)
+        assert result1.just_ended is True
+        # Second call should be no-op (no new on_compact)
+        result2 = await tracker.emit_end_if_ready(session)
+        assert result2.just_ended is False
+        assert result2.events == []
 
     @pytest.mark.asyncio
     async def test_emit_end_no_op_when_nothing_happened(self):
         tracker = CompactionTracker()
         session = _make_session()
-        evts = await tracker.emit_end_if_ready(session)
-        assert evts == []
+        result = await tracker.emit_end_if_ready(session)
+        assert result.just_ended is False
+        assert result.events == []
 
     def test_emit_pre_query(self):
         tracker = CompactionTracker()
@@ -246,20 +251,29 @@ class TestCompactionTracker:
         tracker._done = True
         tracker._start_emitted = True
         tracker._tool_call_id = "old"
+        tracker._transcript_path = "/some/path"
         tracker.reset_for_query()
         assert tracker._done is False
         assert tracker._start_emitted is False
         assert tracker._tool_call_id == ""
+        assert tracker._transcript_path == ""
 
     @pytest.mark.asyncio
-    async def test_pre_query_blocks_sdk_compaction(self):
-        """After pre-query compaction, SDK compaction events are suppressed."""
+    async def test_pre_query_blocks_sdk_compaction_until_reset(self):
+        """After pre-query compaction, SDK compaction is blocked until
+        reset_for_query is called."""
         tracker = CompactionTracker()
         session = _make_session()
         tracker.emit_pre_query(session)
         tracker.on_compact()
+        # _done is True so emit_start_if_ready is blocked
         evts = tracker.emit_start_if_ready()
-        assert evts == []  # _done blocks it
+        assert evts == []
+        # Reset clears _done, allowing subsequent compaction
+        tracker.reset_for_query()
+        tracker.on_compact()
+        evts = tracker.emit_start_if_ready()
+        assert len(evts) == 3
 
     @pytest.mark.asyncio
     async def test_reset_allows_new_compaction(self):
@@ -279,9 +293,9 @@ class TestCompactionTracker:
         session = _make_session()
         tracker.on_compact()
         start_evts = tracker.emit_start_if_ready()
-        end_evts = await tracker.emit_end_if_ready(session)
+        result = await tracker.emit_end_if_ready(session)
         start_evt = start_evts[1]
-        end_evt = end_evts[0]
+        end_evt = result.events[0]
         assert isinstance(start_evt, StreamToolInputStart)
         assert isinstance(end_evt, StreamToolOutputAvailable)
         assert start_evt.toolCallId == end_evt.toolCallId
@@ -289,3 +303,105 @@ class TestCompactionTracker:
         tool_calls = session.messages[0].tool_calls
         assert tool_calls is not None
         assert tool_calls[0]["id"] == start_evt.toolCallId
+
+    @pytest.mark.asyncio
+    async def test_multiple_compactions_within_query(self):
+        """Two mid-stream compactions within a single query both trigger."""
+        tracker = CompactionTracker()
+        session = _make_session()
+
+        # First compaction cycle
+        tracker.on_compact("/path/1")
+        tracker.emit_start_if_ready()
+        result1 = await tracker.emit_end_if_ready(session)
+        assert result1.just_ended is True
+        assert len(result1.events) == 2
+        assert result1.transcript_path == "/path/1"
+
+        # Second compaction cycle (should NOT be blocked — _done resets
+        # because emit_end_if_ready sets it True, but the next on_compact
+        # + emit_start_if_ready checks !_done which IS True now.
+        # So we need reset_for_query between queries, but within a single
+        # query multiple compactions work because _done blocks emit_start
+        # until the next message arrives, at which point emit_end detects it)
+        #
+        # Actually: _done=True blocks emit_start_if_ready, so we need
+        # the stream loop to reset. In practice service.py doesn't call
+        # reset between compactions within the same query — let's verify
+        # the actual behavior.
+        tracker.on_compact("/path/2")
+        # _done is True from first compaction, so start is blocked
+        start_evts = tracker.emit_start_if_ready()
+        assert start_evts == []
+        # But emit_end returns no-op because _done is True
+        result2 = await tracker.emit_end_if_ready(session)
+        assert result2.just_ended is False
+
+    @pytest.mark.asyncio
+    async def test_multiple_compactions_with_intervening_message(self):
+        """Multiple compactions work when the stream loop processes messages between them.
+
+        In the real service.py flow:
+        1. PreCompact fires → on_compact()
+        2. emit_start shows spinner
+        3. Next message arrives → emit_end completes compaction (_done=True)
+        4. Stream continues processing messages...
+        5. If a second PreCompact fires, _done=True blocks emit_start
+        6. But the next message triggers emit_end, which sees _done=True → no-op
+        7. The stream loop needs to detect this and handle accordingly
+
+        The actual flow for multiple compactions within a query requires
+        _done to be cleared between them. The service.py code uses
+        CompactionResult.just_ended to trigger replace_entries, and _done
+        stays True until reset_for_query.
+        """
+        tracker = CompactionTracker()
+        session = _make_session()
+
+        # First compaction
+        tracker.on_compact("/path/1")
+        tracker.emit_start_if_ready()
+        result1 = await tracker.emit_end_if_ready(session)
+        assert result1.just_ended is True
+        assert result1.transcript_path == "/path/1"
+
+        # Simulate reset between queries
+        tracker.reset_for_query()
+
+        # Second compaction in new query
+        tracker.on_compact("/path/2")
+        start_evts = tracker.emit_start_if_ready()
+        assert len(start_evts) == 3
+        result2 = await tracker.emit_end_if_ready(session)
+        assert result2.just_ended is True
+        assert result2.transcript_path == "/path/2"
+
+    def test_on_compact_stores_transcript_path(self):
+        tracker = CompactionTracker()
+        tracker.on_compact("/some/path.jsonl")
+        assert tracker._transcript_path == "/some/path.jsonl"
+
+    @pytest.mark.asyncio
+    async def test_emit_end_returns_transcript_path(self):
+        """CompactionResult includes the transcript_path from on_compact."""
+        tracker = CompactionTracker()
+        session = _make_session()
+        tracker.on_compact("/my/session.jsonl")
+        tracker.emit_start_if_ready()
+        result = await tracker.emit_end_if_ready(session)
+        assert result.just_ended is True
+        assert result.transcript_path == "/my/session.jsonl"
+        # transcript_path is cleared after emit_end
+        assert tracker._transcript_path == ""
+
+    @pytest.mark.asyncio
+    async def test_emit_end_clears_transcript_path(self):
+        """After emit_end, _transcript_path is reset so it doesn't leak to
+        subsequent non-compaction emit_end calls."""
+        tracker = CompactionTracker()
+        session = _make_session()
+        tracker.on_compact("/first/path.jsonl")
+        tracker.emit_start_if_ready()
+        await tracker.emit_end_if_ready(session)
+        # After compaction, _transcript_path is cleared
+        assert tracker._transcript_path == ""

autogpt_platform/backend/backend/copilot/sdk/e2e_compaction_test.py

@@ -0,0 +1,531 @@
+"""End-to-end compaction flow test.
+
+Simulates the full service.py compaction lifecycle using real-format
+JSONL session files — no SDK subprocess needed. Exercises:
+
+  1. TranscriptBuilder loads a "downloaded" transcript
+  2. User query appended, assistant response streamed
+  3. PreCompact hook fires → CompactionTracker.on_compact()
+  4. Next message → emit_start_if_ready() yields spinner events
+  5. Message after that → emit_end_if_ready() returns CompactionResult
+  6. read_compacted_entries() reads the CLI session file
+  7. TranscriptBuilder.replace_entries() syncs state
+  8. More messages appended post-compaction
+  9. to_jsonl() exports full state for upload
+  10. Fresh builder loads the export — roundtrip verified
+"""
+
+import asyncio
+
+from backend.copilot.model import ChatSession
+from backend.copilot.response_model import (
+    StreamFinishStep,
+    StreamStartStep,
+    StreamToolInputAvailable,
+    StreamToolInputStart,
+    StreamToolOutputAvailable,
+)
+from backend.copilot.sdk.compaction import CompactionTracker
+from backend.copilot.sdk.transcript import (
+    read_compacted_entries,
+    strip_progress_entries,
+)
+from backend.copilot.sdk.transcript_builder import TranscriptBuilder
+from backend.util import json
+
+
+def _make_jsonl(*entries: dict) -> str:
+    return "\n".join(json.dumps(e) for e in entries) + "\n"
+
+
+def _run(coro):
+    """Run an async coroutine synchronously."""
+    return asyncio.run(coro)
+
+
+# ---------------------------------------------------------------------------
+# Fixtures: realistic CLI session file content
+# ---------------------------------------------------------------------------
+
+# Pre-compaction conversation
+USER_1 = {
+    "type": "user",
+    "uuid": "u1",
+    "message": {"role": "user", "content": "What files are in this project?"},
+}
+ASST_1_THINKING = {
+    "type": "assistant",
+    "uuid": "a1-think",
+    "parentUuid": "u1",
+    "message": {
+        "role": "assistant",
+        "id": "msg_sdk_aaa",
+        "type": "message",
+        "content": [{"type": "thinking", "thinking": "Let me look at the files..."}],
+        "stop_reason": None,
+        "stop_sequence": None,
+    },
+}
+ASST_1_TOOL = {
+    "type": "assistant",
+    "uuid": "a1-tool",
+    "parentUuid": "u1",
+    "message": {
+        "role": "assistant",
+        "id": "msg_sdk_aaa",
+        "type": "message",
+        "content": [
+            {
+                "type": "tool_use",
+                "id": "tu1",
+                "name": "Bash",
+                "input": {"command": "ls"},
+            }
+        ],
+        "stop_reason": "tool_use",
+        "stop_sequence": None,
+    },
+}
+TOOL_RESULT_1 = {
+    "type": "user",
+    "uuid": "tr1",
+    "parentUuid": "a1-tool",
+    "message": {
+        "role": "user",
+        "content": [
+            {
+                "type": "tool_result",
+                "tool_use_id": "tu1",
+                "content": "file1.py\nfile2.py",
+            }
+        ],
+    },
+}
+ASST_1_TEXT = {
+    "type": "assistant",
+    "uuid": "a1-text",
+    "parentUuid": "tr1",
+    "message": {
+        "role": "assistant",
+        "id": "msg_sdk_bbb",
+        "type": "message",
+        "content": [{"type": "text", "text": "I found file1.py and file2.py."}],
+        "stop_reason": "end_turn",
+        "stop_sequence": None,
+    },
+}
+# Progress entries (should be stripped during upload)
+PROGRESS_1 = {
+    "type": "progress",
+    "uuid": "prog1",
+    "parentUuid": "a1-tool",
+    "data": {"type": "bash_progress", "stdout": "running ls..."},
+}
+# Second user message
+USER_2 = {
+    "type": "user",
+    "uuid": "u2",
+    "parentUuid": "a1-text",
+    "message": {"role": "user", "content": "Show me file1.py"},
+}
+ASST_2 = {
+    "type": "assistant",
+    "uuid": "a2",
+    "parentUuid": "u2",
+    "message": {
+        "role": "assistant",
+        "id": "msg_sdk_ccc",
+        "type": "message",
+        "content": [{"type": "text", "text": "Here is file1.py content..."}],
+        "stop_reason": "end_turn",
+        "stop_sequence": None,
+    },
+}
+
+# --- Compaction summary (written by CLI after context compaction) ---
+COMPACT_SUMMARY = {
+    "type": "summary",
+    "uuid": "cs1",
+    "isCompactSummary": True,
+    "message": {
+        "role": "user",
+        "content": (
+            "Summary: User asked about project files. Found file1.py and file2.py. "
+            "User then asked to see file1.py."
+        ),
+    },
+}
+
+# Post-compaction assistant response
+POST_COMPACT_ASST = {
+    "type": "assistant",
+    "uuid": "a3",
+    "parentUuid": "cs1",
+    "message": {
+        "role": "assistant",
+        "id": "msg_sdk_ddd",
+        "type": "message",
+        "content": [{"type": "text", "text": "Here is the content of file1.py..."}],
+        "stop_reason": "end_turn",
+        "stop_sequence": None,
+    },
+}
+
+# Post-compaction user follow-up
+USER_3 = {
+    "type": "user",
+    "uuid": "u3",
+    "parentUuid": "a3",
+    "message": {"role": "user", "content": "Now show file2.py"},
+}
+ASST_3 = {
+    "type": "assistant",
+    "uuid": "a4",
+    "parentUuid": "u3",
+    "message": {
+        "role": "assistant",
+        "id": "msg_sdk_eee",
+        "type": "message",
+        "content": [{"type": "text", "text": "Here is file2.py..."}],
+        "stop_reason": "end_turn",
+        "stop_sequence": None,
+    },
+}
+
+
+# ---------------------------------------------------------------------------
+# E2E test
+# ---------------------------------------------------------------------------
+
+
+class TestCompactionE2E:
+    def _write_session_file(self, session_dir, entries):
+        """Write a CLI session JSONL file."""
+        path = session_dir / "session.jsonl"
+        path.write_text(_make_jsonl(*entries))
+        return path
+
+    def test_full_compaction_lifecycle(self, tmp_path, monkeypatch):
+        """Simulate the complete service.py compaction flow.
+
+        Timeline:
+        1. Previous turn uploaded transcript with [USER_1, ASST_1, USER_2, ASST_2]
+        2. Current turn: download → load_previous
+        3. User sends "Now show file2.py" → append_user
+        4. SDK starts streaming response
+        5. Mid-stream: PreCompact hook fires (context too large)
+        6. CLI writes compaction summary to session file
+        7. Next SDK message → emit_start (spinner)
+        8. Following message → emit_end (CompactionResult)
+        9. read_compacted_entries reads the session file
+        10. replace_entries syncs TranscriptBuilder
+        11. More assistant messages appended
+        12. Export → upload → next turn downloads it
+        """
+        # --- Setup CLI projects directory ---
+        config_dir = tmp_path / "config"
+        projects_dir = config_dir / "projects"
+        session_dir = projects_dir / "proj"
+        session_dir.mkdir(parents=True)
+        monkeypatch.setenv("CLAUDE_CONFIG_DIR", str(config_dir))
+
+        # --- Step 1-2: Load "downloaded" transcript from previous turn ---
+        previous_transcript = _make_jsonl(
+            USER_1,
+            ASST_1_THINKING,
+            ASST_1_TOOL,
+            TOOL_RESULT_1,
+            ASST_1_TEXT,
+            USER_2,
+            ASST_2,
+        )
+        builder = TranscriptBuilder()
+        builder.load_previous(previous_transcript)
+        assert builder.entry_count == 7
+
+        # --- Step 3: User sends new query ---
+        builder.append_user("Now show file2.py")
+        assert builder.entry_count == 8
+
+        # --- Step 4: SDK starts streaming ---
+        builder.append_assistant(
+            [{"type": "thinking", "thinking": "Let me read file2.py..."}],
+            model="claude-sonnet-4-20250514",
+        )
+        assert builder.entry_count == 9
+
+        # --- Step 5-6: PreCompact fires, CLI writes session file ---
+        session_file = self._write_session_file(
+            session_dir,
+            [
+                USER_1,
+                ASST_1_THINKING,
+                ASST_1_TOOL,
+                PROGRESS_1,
+                TOOL_RESULT_1,
+                ASST_1_TEXT,
+                USER_2,
+                ASST_2,
+                COMPACT_SUMMARY,
+                POST_COMPACT_ASST,
+                USER_3,
+                ASST_3,
+            ],
+        )
+
+        # --- Step 7: CompactionTracker receives PreCompact hook ---
+        tracker = CompactionTracker()
+        session = ChatSession.new(user_id="test-user")
+        tracker.on_compact(str(session_file))
+
+        # --- Step 8: Next SDK message arrives → emit_start ---
+        start_events = tracker.emit_start_if_ready()
+        assert len(start_events) == 3
+        assert isinstance(start_events[0], StreamStartStep)
+        assert isinstance(start_events[1], StreamToolInputStart)
+        assert isinstance(start_events[2], StreamToolInputAvailable)
+
+        # Verify tool_call_id is set
+        tool_call_id = start_events[1].toolCallId
+        assert tool_call_id.startswith("compaction-")
+
+        # --- Step 9: Following message → emit_end ---
+        result = _run(tracker.emit_end_if_ready(session))
+        assert result.just_ended is True
+        assert result.transcript_path == str(session_file)
+        assert len(result.events) == 2
+        assert isinstance(result.events[0], StreamToolOutputAvailable)
+        assert isinstance(result.events[1], StreamFinishStep)
+        # Verify same tool_call_id
+        assert result.events[0].toolCallId == tool_call_id
+
+        # Session should have compaction messages persisted
+        assert len(session.messages) == 2
+        assert session.messages[0].role == "assistant"
+        assert session.messages[1].role == "tool"
+
+        # --- Step 10: read_compacted_entries + replace_entries ---
+        compacted = read_compacted_entries(str(session_file))
+        assert compacted is not None
+        # Should have: COMPACT_SUMMARY + POST_COMPACT_ASST + USER_3 + ASST_3
+        assert len(compacted) == 4
+        assert compacted[0]["uuid"] == "cs1"
+        assert compacted[0]["isCompactSummary"] is True
+
+        # Replace builder state with compacted entries
+        old_count = builder.entry_count
+        builder.replace_entries(compacted)
+        assert builder.entry_count == 4  # Only compacted entries
+        assert builder.entry_count < old_count  # Compaction reduced entries
+
+        # --- Step 11: More assistant messages after compaction ---
+        builder.append_assistant(
+            [{"type": "text", "text": "Here is file2.py:\n\ndef hello():\n    pass"}],
+            model="claude-sonnet-4-20250514",
+            stop_reason="end_turn",
+        )
+        assert builder.entry_count == 5
+
+        # --- Step 12: Export for upload ---
+        output = builder.to_jsonl()
+        assert output  # Not empty
+        output_entries = [json.loads(line) for line in output.strip().split("\n")]
+        assert len(output_entries) == 5
+
+        # Verify structure:
+        # [COMPACT_SUMMARY, POST_COMPACT_ASST, USER_3, ASST_3, new_assistant]
+        assert output_entries[0]["type"] == "summary"
+        assert output_entries[0].get("isCompactSummary") is True
+        assert output_entries[0]["uuid"] == "cs1"
+        assert output_entries[1]["uuid"] == "a3"
+        assert output_entries[2]["uuid"] == "u3"
+        assert output_entries[3]["uuid"] == "a4"
+        assert output_entries[4]["type"] == "assistant"
+
+        # Verify parent chain is intact
+        assert output_entries[1]["parentUuid"] == "cs1"  # a3 → cs1
+        assert output_entries[2]["parentUuid"] == "a3"  # u3 → a3
+        assert output_entries[3]["parentUuid"] == "u3"  # a4 → u3
+        assert output_entries[4]["parentUuid"] == "a4"  # new → a4
+
+        # --- Step 13: Roundtrip — next turn loads this export ---
+        builder2 = TranscriptBuilder()
+        builder2.load_previous(output)
+        assert builder2.entry_count == 5
+
+        # isCompactSummary survives roundtrip
+        output2 = builder2.to_jsonl()
+        first_entry = json.loads(output2.strip().split("\n")[0])
+        assert first_entry.get("isCompactSummary") is True
+
+        # Can append more messages
+        builder2.append_user("What about file3.py?")
+        assert builder2.entry_count == 6
+        final_output = builder2.to_jsonl()
+        last_entry = json.loads(final_output.strip().split("\n")[-1])
+        assert last_entry["type"] == "user"
+        # Parented to the last entry from previous turn
+        assert last_entry["parentUuid"] == output_entries[-1]["uuid"]
+
+    def test_double_compaction_within_session(self, tmp_path, monkeypatch):
+        """Two compactions in the same session (across reset_for_query)."""
+        config_dir = tmp_path / "config"
+        projects_dir = config_dir / "projects"
+        session_dir = projects_dir / "proj"
+        session_dir.mkdir(parents=True)
+        monkeypatch.setenv("CLAUDE_CONFIG_DIR", str(config_dir))
+
+        tracker = CompactionTracker()
+        session = ChatSession.new(user_id="test")
+        builder = TranscriptBuilder()
+
+        # --- First query with compaction ---
+        builder.append_user("first question")
+        builder.append_assistant([{"type": "text", "text": "first answer"}])
+
+        # Write session file for first compaction
+        first_summary = {
+            "type": "summary",
+            "uuid": "cs-first",
+            "isCompactSummary": True,
+            "message": {"role": "user", "content": "First compaction summary"},
+        }
+        first_post = {
+            "type": "assistant",
+            "uuid": "a-first",
+            "parentUuid": "cs-first",
+            "message": {"role": "assistant", "content": "first post-compact"},
+        }
+        file1 = session_dir / "session1.jsonl"
+        file1.write_text(_make_jsonl(first_summary, first_post))
+
+        tracker.on_compact(str(file1))
+        tracker.emit_start_if_ready()
+        result1 = _run(tracker.emit_end_if_ready(session))
+        assert result1.just_ended is True
+
+        compacted1 = read_compacted_entries(str(file1))
+        assert compacted1 is not None
+        builder.replace_entries(compacted1)
+        assert builder.entry_count == 2
+
+        # --- Reset for second query ---
+        tracker.reset_for_query()
+
+        # --- Second query with compaction ---
+        builder.append_user("second question")
+        builder.append_assistant([{"type": "text", "text": "second answer"}])
+
+        second_summary = {
+            "type": "summary",
+            "uuid": "cs-second",
+            "isCompactSummary": True,
+            "message": {"role": "user", "content": "Second compaction summary"},
+        }
+        second_post = {
+            "type": "assistant",
+            "uuid": "a-second",
+            "parentUuid": "cs-second",
+            "message": {"role": "assistant", "content": "second post-compact"},
+        }
+        file2 = session_dir / "session2.jsonl"
+        file2.write_text(_make_jsonl(second_summary, second_post))
+
+        tracker.on_compact(str(file2))
+        tracker.emit_start_if_ready()
+        result2 = _run(tracker.emit_end_if_ready(session))
+        assert result2.just_ended is True
+
+        compacted2 = read_compacted_entries(str(file2))
+        assert compacted2 is not None
+        builder.replace_entries(compacted2)
+        assert builder.entry_count == 2  # Only second compaction entries
+
+        # Export and verify
+        output = builder.to_jsonl()
+        entries = [json.loads(line) for line in output.strip().split("\n")]
+        assert entries[0]["uuid"] == "cs-second"
+        assert entries[0].get("isCompactSummary") is True
+
+    def test_strip_progress_then_load_then_compact_roundtrip(
+        self, tmp_path, monkeypatch
+    ):
+        """Full pipeline: strip → load → compact → replace → export → reload.
+
+        This tests the exact sequence that happens across two turns:
+        Turn 1: SDK produces transcript with progress entries
+        Upload: strip_progress_entries removes progress, upload to cloud
+        Turn 2: Download → load_previous → compaction fires → replace → export
+        Turn 3: Download the Turn 2 export → load_previous (roundtrip)
+        """
+        config_dir = tmp_path / "config"
+        projects_dir = config_dir / "projects"
+        session_dir = projects_dir / "proj"
+        session_dir.mkdir(parents=True)
+        monkeypatch.setenv("CLAUDE_CONFIG_DIR", str(config_dir))
+
+        # --- Turn 1: SDK produces raw transcript ---
+        raw_content = _make_jsonl(
+            USER_1,
+            ASST_1_THINKING,
+            ASST_1_TOOL,
+            PROGRESS_1,
+            TOOL_RESULT_1,
+            ASST_1_TEXT,
+            USER_2,
+            ASST_2,
+        )
+
+        # Strip progress for upload
+        stripped = strip_progress_entries(raw_content)
+        stripped_entries = [
+            json.loads(line) for line in stripped.strip().split("\n") if line.strip()
+        ]
+        # Progress should be gone
+        assert not any(e.get("type") == "progress" for e in stripped_entries)
+        assert len(stripped_entries) == 7  # 8 - 1 progress
+
+        # --- Turn 2: Download stripped, load, compaction happens ---
+        builder = TranscriptBuilder()
+        builder.load_previous(stripped)
+        assert builder.entry_count == 7
+
+        builder.append_user("Now show file2.py")
+        builder.append_assistant(
+            [{"type": "text", "text": "Reading file2.py..."}],
+            model="claude-sonnet-4-20250514",
+        )
+
+        # CLI writes session file with compaction
+        session_file = self._write_session_file(
+            session_dir,
+            [
+                USER_1,
+                ASST_1_TOOL,
+                TOOL_RESULT_1,
+                ASST_1_TEXT,
+                USER_2,
+                ASST_2,
+                COMPACT_SUMMARY,
+                POST_COMPACT_ASST,
+            ],
+        )
+
+        compacted = read_compacted_entries(str(session_file))
+        assert compacted is not None
+        builder.replace_entries(compacted)
+
+        # Append post-compaction message
+        builder.append_user("Thanks!")
+        output = builder.to_jsonl()
+
+        # --- Turn 3: Fresh load of Turn 2 export ---
+        builder3 = TranscriptBuilder()
+        builder3.load_previous(output)
+        # Should have: compact_summary + post_compact_asst + "Thanks!"
+        assert builder3.entry_count == 3
+
+        # Compact summary survived the full pipeline
+        first = json.loads(builder3.to_jsonl().strip().split("\n")[0])
+        assert first.get("isCompactSummary") is True
+        assert first["type"] == "summary"

autogpt_platform/backend/backend/copilot/sdk/mcp_tool_guide.md

@@ -20,7 +20,24 @@ Use these URLs directly without asking the user:
 | Cloudflare | `https://mcp.cloudflare.com/mcp` |
 | Atlassian / Jira | `https://mcp.atlassian.com/mcp` |
 
-For other services, search the MCP registry at https://registry.modelcontextprotocol.io/.
+For other services, search the MCP registry API:
+```http
+GET https://registry.modelcontextprotocol.io/v0/servers?q=<search_term>
+```
+Each result includes a `remotes` array with the exact server URL to use.
+
+### Important: Check blocks first
+
+Before using `run_mcp_tool`, always check if the platform already has blocks for the service
+using `find_block`. The platform has hundreds of built-in blocks (Google Sheets, Google Docs,
+Google Calendar, Gmail, etc.) that work without MCP setup.
+
+Only use `run_mcp_tool` when:
+- The service is in the known hosted MCP servers list above, OR
+- You searched `find_block` first and found no matching blocks
+
+**Never guess or construct MCP server URLs.** Only use URLs from the known servers list above
+or from the `remotes[].url` field in MCP registry search results.
 
 ### Authentication
 

autogpt_platform/backend/backend/copilot/sdk/security_hooks.py

@@ -127,7 +127,7 @@ def create_security_hooks(
     user_id: str | None,
     sdk_cwd: str | None = None,
     max_subtasks: int = 3,
-    on_compact: Callable[[], None] | None = None,
+    on_compact: Callable[[str], None] | None = None,
 ) -> dict[str, Any]:
     """Create the security hooks configuration for Claude Agent SDK.
 
@@ -142,6 +142,7 @@ def create_security_hooks(
         sdk_cwd: SDK working directory for workspace-scoped tool validation
         max_subtasks: Maximum concurrent Task (sub-agent) spawns allowed per session
         on_compact: Callback invoked when SDK starts compacting context.
+            Receives the transcript_path from the hook input.
 
     Returns:
         Hooks configuration dict for ClaudeAgentOptions
@@ -301,11 +302,21 @@ def create_security_hooks(
             """
             _ = context, tool_use_id
             trigger = input_data.get("trigger", "auto")
+            # Sanitize untrusted input before logging to prevent log injection
+            transcript_path = (
+                str(input_data.get("transcript_path", ""))
+                .replace("\n", "")
+                .replace("\r", "")
+            )
             logger.info(
-                f"[SDK] Context compaction triggered: {trigger}, user={user_id}"
+                "[SDK] Context compaction triggered: %s, user=%s, "
+                "transcript_path=%s",
+                trigger,
+                user_id,
+                transcript_path,
             )
             if on_compact is not None:
-                on_compact()
+                on_compact(transcript_path)
             return cast(SyncHookJSONOutput, {})
 
         hooks: dict[str, Any] = {

autogpt_platform/backend/backend/copilot/sdk/service.py

@@ -77,6 +77,7 @@ from .tool_adapter import (
 from .transcript import (
     cleanup_cli_project_dir,
     download_transcript,
+    read_compacted_entries,
     upload_transcript,
     validate_transcript,
     write_transcript_to_tempfile,
@@ -1045,6 +1046,7 @@ async def stream_chat_completion_sdk(
                             exc_info=True,
                         )
                         ended_with_stream_error = True
+
                         yield StreamError(
                             errorText=f"SDK stream error: {stream_err}",
                             code="sdk_stream_error",
@@ -1129,9 +1131,26 @@ async def stream_chat_completion_sdk(
                                 sdk_msg.result or "(no error message provided)",
                             )
 
-                    # Emit compaction end if SDK finished compacting
-                    for ev in await compaction.emit_end_if_ready(session):
+                    # Emit compaction end if SDK finished compacting.
+                    # When compaction ends, sync TranscriptBuilder with the
+                    # CLI's active context so they stay identical.
+                    compact_result = await compaction.emit_end_if_ready(session)
+                    for ev in compact_result.events:
                         yield ev
+                    # After replace_entries, skip append_assistant for this
+                    # sdk_msg — the CLI session file already contains it,
+                    # so appending again would create a duplicate.
+                    entries_replaced = False
+                    if compact_result.just_ended:
+                        compacted = await asyncio.to_thread(
+                            read_compacted_entries,
+                            compact_result.transcript_path,
+                        )
+                        if compacted is not None:
+                            transcript_builder.replace_entries(
+                                compacted, log_prefix=log_prefix
+                            )
+                            entries_replaced = True
 
                     for response in adapter.convert_message(sdk_msg):
                         if isinstance(response, StreamStart):
@@ -1218,10 +1237,11 @@ async def stream_chat_completion_sdk(
                                     tool_call_id=response.toolCallId,
                                 )
                             )
-                            transcript_builder.append_tool_result(
-                                tool_use_id=response.toolCallId,
-                                content=content,
-                            )
+                            if not entries_replaced:
+                                transcript_builder.append_tool_result(
+                                    tool_use_id=response.toolCallId,
+                                    content=content,
+                                )
                             has_tool_results = True
 
                         elif isinstance(response, StreamFinish):
@@ -1231,7 +1251,9 @@ async def stream_chat_completion_sdk(
                     # any stashed tool results from the previous turn are
                     # recorded first, preserving the required API order:
                     # assistant(tool_use) → tool_result → assistant(text).
-                    if isinstance(sdk_msg, AssistantMessage):
+                    # Skip if replace_entries just ran — the CLI session
+                    # file already contains this message.
+                    if isinstance(sdk_msg, AssistantMessage) and not entries_replaced:
                         transcript_builder.append_assistant(
                             content_blocks=_format_sdk_content_blocks(sdk_msg.content),
                             model=sdk_msg.model,
@@ -1422,13 +1444,13 @@ async def stream_chat_completion_sdk(
             task.add_done_callback(_background_tasks.discard)
 
         # --- Upload transcript for next-turn --resume ---
-        # This MUST run in finally so the transcript is uploaded even when
-        # the streaming loop raises an exception.
-        # The transcript represents the COMPLETE active context (atomic).
+        # TranscriptBuilder is the single source of truth.  It mirrors the
+        # CLI's active context: on compaction, replace_entries() syncs it
+        # with the compacted session file.  No CLI file read needed here.
         if config.claude_agent_use_resume and user_id and session is not None:
             try:
-                # Build complete transcript from captured SDK messages
                 transcript_content = transcript_builder.to_jsonl()
+                entry_count = transcript_builder.entry_count
 
                 if not transcript_content:
                     logger.warning(
@@ -1438,18 +1460,15 @@ async def stream_chat_completion_sdk(
                     logger.warning(
                         "%s Transcript invalid, skipping upload (entries=%d)",
                         log_prefix,
-                        transcript_builder.entry_count,
+                        entry_count,
                     )
                 else:
                     logger.info(
-                        "%s Uploading complete transcript (entries=%d, bytes=%d)",
+                        "%s Uploading transcript (entries=%d, bytes=%d)",
                         log_prefix,
-                        transcript_builder.entry_count,
+                        entry_count,
                         len(transcript_content),
                     )
-                    # Shield upload from cancellation - let it complete even if
-                    # the finally block is interrupted. No timeout to avoid race
-                    # conditions where backgrounded uploads overwrite newer transcripts.
                     await asyncio.shield(
                         upload_transcript(
                             user_id=user_id,

autogpt_platform/backend/backend/copilot/sdk/transcript.py

@@ -13,8 +13,10 @@ filesystem for self-hosted) — no DB column needed.
 import logging
 import os
 import re
+import shutil
 import time
 from dataclasses import dataclass
+from pathlib import Path
 
 from backend.util import json
 
@@ -82,7 +84,11 @@ def strip_progress_entries(content: str) -> str:
         parent = entry.get("parentUuid", "")
         if uid:
             uuid_to_parent[uid] = parent
-        if entry.get("type", "") in STRIPPABLE_TYPES and uid:
+        if (
+            entry.get("type", "") in STRIPPABLE_TYPES
+            and uid
+            and not entry.get("isCompactSummary")
+        ):
             stripped_uuids.add(uid)
 
     # Second pass: keep non-stripped entries, reparenting where needed.
@@ -106,7 +112,9 @@ def strip_progress_entries(content: str) -> str:
         if not isinstance(entry, dict):
             result_lines.append(line)
             continue
-        if entry.get("type", "") in STRIPPABLE_TYPES:
+        if entry.get("type", "") in STRIPPABLE_TYPES and not entry.get(
+            "isCompactSummary"
+        ):
             continue
         uid = entry.get("uuid", "")
         if uid in reparented:
@@ -137,6 +145,155 @@ def _sanitize_id(raw_id: str, max_len: int = 36) -> str:
 _SAFE_CWD_PREFIX = os.path.realpath("/tmp/copilot-")
 
 
+def _projects_base() -> str:
+    """Return the resolved path to the CLI's projects directory."""
+    config_dir = os.environ.get("CLAUDE_CONFIG_DIR") or os.path.expanduser("~/.claude")
+    return os.path.realpath(os.path.join(config_dir, "projects"))
+
+
+def _cli_project_dir(sdk_cwd: str) -> str | None:
+    """Return the CLI's project directory for a given working directory.
+
+    Returns ``None`` if the path would escape the projects base.
+    """
+    cwd_encoded = re.sub(r"[^a-zA-Z0-9]", "-", os.path.realpath(sdk_cwd))
+    projects_base = _projects_base()
+    project_dir = os.path.realpath(os.path.join(projects_base, cwd_encoded))
+
+    if not project_dir.startswith(projects_base + os.sep):
+        logger.warning(
+            "[Transcript] Project dir escaped projects base: %s", project_dir
+        )
+        return None
+    return project_dir
+
+
+def _safe_glob_jsonl(project_dir: str) -> list[Path]:
+    """Glob ``*.jsonl`` files, filtering out symlinks that escape the directory."""
+    try:
+        resolved_base = Path(project_dir).resolve()
+    except OSError as e:
+        logger.warning("[Transcript] Failed to resolve project dir: %s", e)
+        return []
+
+    result: list[Path] = []
+    for candidate in Path(project_dir).glob("*.jsonl"):
+        try:
+            resolved = candidate.resolve()
+            if resolved.is_relative_to(resolved_base):
+                result.append(resolved)
+        except (OSError, RuntimeError) as e:
+            logger.debug(
+                "[Transcript] Skipping invalid CLI session candidate %s: %s",
+                candidate,
+                e,
+            )
+    return result
+
+
+def read_compacted_entries(transcript_path: str) -> list[dict] | None:
+    """Read compacted entries from the CLI session file after compaction.
+
+    Parses the JSONL file line-by-line, finds the ``isCompactSummary: true``
+    entry, and returns it plus all entries after it.
+
+    The CLI writes the compaction summary BEFORE sending the next message,
+    so the file is guaranteed to be flushed by the time we read it.
+
+    Returns a list of parsed dicts, or ``None`` if the file cannot be read
+    or no compaction summary is found.
+    """
+    if not transcript_path:
+        return None
+
+    projects_base = _projects_base()
+    real_path = os.path.realpath(transcript_path)
+    if not real_path.startswith(projects_base + os.sep):
+        logger.warning(
+            "[Transcript] transcript_path outside projects base: %s", transcript_path
+        )
+        return None
+
+    try:
+        content = Path(real_path).read_text()
+    except OSError as e:
+        logger.warning(
+            "[Transcript] Failed to read session file %s: %s", transcript_path, e
+        )
+        return None
+
+    lines = content.strip().split("\n")
+    compact_idx: int | None = None
+
+    for idx, line in enumerate(lines):
+        if not line.strip():
+            continue
+        entry = json.loads(line, fallback=None)
+        if not isinstance(entry, dict):
+            continue
+        if entry.get("isCompactSummary"):
+            compact_idx = idx  # don't break — find the LAST summary
+
+    if compact_idx is None:
+        logger.debug("[Transcript] No compaction summary found in %s", transcript_path)
+        return None
+
+    entries: list[dict] = []
+    for line in lines[compact_idx:]:
+        if not line.strip():
+            continue
+        entry = json.loads(line, fallback=None)
+        if isinstance(entry, dict):
+            entries.append(entry)
+
+    logger.info(
+        "[Transcript] Read %d compacted entries from %s (summary at line %d)",
+        len(entries),
+        transcript_path,
+        compact_idx + 1,
+    )
+    return entries
+
+
+def read_cli_session_file(sdk_cwd: str) -> str | None:
+    """Read the CLI's own session file, which reflects any compaction.
+
+    The CLI writes its session transcript to
+    ``~/.claude/projects/<encoded_cwd>/<session_id>.jsonl``.
+    Since each SDK turn uses a unique ``sdk_cwd``, there should be
+    exactly one ``.jsonl`` file in that directory.
+
+    Returns the file content, or ``None`` if not found.
+    """
+    project_dir = _cli_project_dir(sdk_cwd)
+    if not project_dir or not os.path.isdir(project_dir):
+        return None
+
+    jsonl_files = _safe_glob_jsonl(project_dir)
+    if not jsonl_files:
+        logger.debug("[Transcript] No CLI session file found in %s", project_dir)
+        return None
+
+    # Pick the most recently modified file (should be only one per turn).
+    try:
+        session_file = max(jsonl_files, key=lambda p: p.stat().st_mtime)
+    except OSError as e:
+        logger.warning("[Transcript] Failed to inspect CLI session files: %s", e)
+        return None
+
+    try:
+        content = session_file.read_text()
+        logger.info(
+            "[Transcript] Read CLI session file: %s (%d bytes)",
+            session_file,
+            len(content),
+        )
+        return content
+    except OSError as e:
+        logger.warning("[Transcript] Failed to read CLI session file: %s", e)
+        return None
+
+
 def cleanup_cli_project_dir(sdk_cwd: str) -> None:
     """Remove the CLI's project directory for a specific working directory.
 
@@ -144,25 +301,15 @@ def cleanup_cli_project_dir(sdk_cwd: str) -> None:
     Each SDK turn uses a unique ``sdk_cwd``, so the project directory is
     safe to remove entirely after the transcript has been uploaded.
     """
-    import shutil
-
-    # Encode cwd the same way CLI does (replaces non-alphanumeric with -)
-    cwd_encoded = re.sub(r"[^a-zA-Z0-9]", "-", os.path.realpath(sdk_cwd))
-    config_dir = os.environ.get("CLAUDE_CONFIG_DIR") or os.path.expanduser("~/.claude")
-    projects_base = os.path.realpath(os.path.join(config_dir, "projects"))
-    project_dir = os.path.realpath(os.path.join(projects_base, cwd_encoded))
-
-    if not project_dir.startswith(projects_base + os.sep):
-        logger.warning(
-            f"[Transcript] Cleanup path escaped projects base: {project_dir}"
-        )
+    project_dir = _cli_project_dir(sdk_cwd)
+    if not project_dir:
         return
 
     if os.path.isdir(project_dir):
         shutil.rmtree(project_dir, ignore_errors=True)
-        logger.debug(f"[Transcript] Cleaned up CLI project dir: {project_dir}")
+        logger.debug("[Transcript] Cleaned up CLI project dir: %s", project_dir)
     else:
-        logger.debug(f"[Transcript] Project dir not found: {project_dir}")
+        logger.debug("[Transcript] Project dir not found: %s", project_dir)
 
 
 def write_transcript_to_tempfile(
@@ -259,24 +406,27 @@ def _meta_storage_path_parts(user_id: str, session_id: str) -> tuple[str, str, s
     )
 
 
-def _build_storage_path(user_id: str, session_id: str, backend: object) -> str:
-    """Build the full storage path string that ``retrieve()`` expects.
-
-    ``store()`` returns a path like ``gcs://bucket/workspaces/...`` or
-    ``local://workspace_id/file_id/filename``.  Since we use deterministic
-    arguments we can reconstruct the same path for download/delete without
-    having stored the return value.
-    """
+def _build_path_from_parts(parts: tuple[str, str, str], backend: object) -> str:
+    """Build a full storage path from (workspace_id, file_id, filename) parts."""
     from backend.util.workspace_storage import GCSWorkspaceStorage
 
-    wid, fid, fname = _storage_path_parts(user_id, session_id)
-
+    wid, fid, fname = parts
     if isinstance(backend, GCSWorkspaceStorage):
         blob = f"workspaces/{wid}/{fid}/{fname}"
         return f"gcs://{backend.bucket_name}/{blob}"
-    else:
-        # LocalWorkspaceStorage returns local://{relative_path}
-        return f"local://{wid}/{fid}/{fname}"
+    return f"local://{wid}/{fid}/{fname}"
+
+
+def _build_storage_path(user_id: str, session_id: str, backend: object) -> str:
+    """Build the full storage path string that ``retrieve()`` expects."""
+    return _build_path_from_parts(_storage_path_parts(user_id, session_id), backend)
+
+
+def _build_meta_storage_path(user_id: str, session_id: str, backend: object) -> str:
+    """Build the full storage path for the companion .meta.json file."""
+    return _build_path_from_parts(
+        _meta_storage_path_parts(user_id, session_id), backend
+    )
 
 
 async def upload_transcript(
@@ -381,15 +531,7 @@ async def download_transcript(
     message_count = 0
     uploaded_at = 0.0
     try:
-        from backend.util.workspace_storage import GCSWorkspaceStorage
-
-        mwid, mfid, mfname = _meta_storage_path_parts(user_id, session_id)
-        if isinstance(storage, GCSWorkspaceStorage):
-            blob = f"workspaces/{mwid}/{mfid}/{mfname}"
-            meta_path = f"gcs://{storage.bucket_name}/{blob}"
-        else:
-            meta_path = f"local://{mwid}/{mfid}/{mfname}"
-
+        meta_path = _build_meta_storage_path(user_id, session_id, storage)
         meta_data = await storage.retrieve(meta_path)
         meta = json.loads(meta_data.decode("utf-8"), fallback={})
         message_count = meta.get("message_count", 0)
@@ -406,7 +548,11 @@ async def download_transcript(
 
 
 async def delete_transcript(user_id: str, session_id: str) -> None:
-    """Delete transcript from bucket storage (e.g. after resume failure)."""
+    """Delete transcript and its metadata from bucket storage.
+
+    Removes both the ``.jsonl`` transcript and the companion ``.meta.json``
+    so stale ``message_count`` watermarks cannot corrupt gap-fill logic.
+    """
     from backend.util.workspace_storage import get_workspace_storage
 
     storage = await get_workspace_storage()
@@ -414,6 +560,14 @@ async def delete_transcript(user_id: str, session_id: str) -> None:
 
     try:
         await storage.delete(path)
-        logger.info(f"[Transcript] Deleted transcript for session {session_id}")
+        logger.info("[Transcript] Deleted transcript for session %s", session_id)
     except Exception as e:
-        logger.warning(f"[Transcript] Failed to delete transcript: {e}")
+        logger.warning("[Transcript] Failed to delete transcript: %s", e)
+
+    # Also delete the companion .meta.json to avoid orphaned metadata.
+    try:
+        meta_path = _build_meta_storage_path(user_id, session_id, storage)
+        await storage.delete(meta_path)
+        logger.info("[Transcript] Deleted metadata for session %s", session_id)
+    except Exception as e:
+        logger.warning("[Transcript] Failed to delete metadata: %s", e)

autogpt_platform/backend/backend/copilot/sdk/transcript_builder.py

@@ -30,6 +30,7 @@ class TranscriptEntry(BaseModel):
     type: str
     uuid: str
     parentUuid: str | None
+    isCompactSummary: bool | None = None
     message: dict[str, Any]
 
 
@@ -53,6 +54,24 @@ class TranscriptBuilder:
             return self._entries[-1].message.get("id", "")
         return ""
 
+    @staticmethod
+    def _parse_entry(data: dict) -> TranscriptEntry | None:
+        """Parse a single transcript entry, filtering strippable types.
+
+        Returns ``None`` for entries that should be skipped (strippable types
+        that are not compaction summaries).
+        """
+        entry_type = data.get("type", "")
+        if entry_type in STRIPPABLE_TYPES and not data.get("isCompactSummary"):
+            return None
+        return TranscriptEntry(
+            type=entry_type,
+            uuid=data.get("uuid") or str(uuid4()),
+            parentUuid=data.get("parentUuid"),
+            isCompactSummary=data.get("isCompactSummary") or None,
+            message=data.get("message", {}),
+        )
+
     def load_previous(self, content: str, log_prefix: str = "[Transcript]") -> None:
         """Load complete previous transcript.
 
@@ -78,18 +97,9 @@ class TranscriptBuilder:
                 )
                 continue
 
-            # Load all non-strippable entries (user/assistant/system/etc.)
-            # Skip only STRIPPABLE_TYPES to match strip_progress_entries() behavior
-            entry_type = data.get("type", "")
-            if entry_type in STRIPPABLE_TYPES:
+            entry = self._parse_entry(data)
+            if entry is None:
                 continue
-
-            entry = TranscriptEntry(
-                type=data["type"],
-                uuid=data.get("uuid") or str(uuid4()),
-                parentUuid=data.get("parentUuid"),
-                message=data.get("message", {}),
-            )
             self._entries.append(entry)
             self._last_uuid = entry.uuid
 
@@ -162,6 +172,43 @@ class TranscriptBuilder:
         )
         self._last_uuid = msg_uuid
 
+    def replace_entries(
+        self, compacted_entries: list[dict], log_prefix: str = "[Transcript]"
+    ) -> None:
+        """Replace all entries with compacted entries from the CLI session file.
+
+        Called after mid-stream compaction so TranscriptBuilder mirrors the
+        CLI's active context (compaction summary + post-compaction entries).
+
+        Builds the new list first and validates it's non-empty before swapping,
+        so corrupt input cannot wipe the conversation history.
+        """
+        new_entries: list[TranscriptEntry] = []
+        for data in compacted_entries:
+            entry = self._parse_entry(data)
+            if entry is not None:
+                new_entries.append(entry)
+
+        if not new_entries:
+            logger.warning(
+                "%s replace_entries produced 0 entries from %d inputs, keeping old (%d entries)",
+                log_prefix,
+                len(compacted_entries),
+                len(self._entries),
+            )
+            return
+
+        old_count = len(self._entries)
+        self._entries = new_entries
+        self._last_uuid = new_entries[-1].uuid
+
+        logger.info(
+            "%s TranscriptBuilder compacted: %d entries -> %d entries",
+            log_prefix,
+            old_count,
+            len(self._entries),
+        )
+
     def to_jsonl(self) -> str:
         """Export complete context as JSONL.
 

autogpt_platform/backend/backend/copilot/sdk/transcript_test.py

@@ -1,15 +1,23 @@
 """Unit tests for JSONL transcript management utilities."""
 
 import os
+from unittest.mock import AsyncMock, patch
+
+import pytest
 
 from backend.util import json
 
 from .transcript import (
     STRIPPABLE_TYPES,
+    _cli_project_dir,
+    delete_transcript,
+    read_cli_session_file,
+    read_compacted_entries,
     strip_progress_entries,
     validate_transcript,
     write_transcript_to_tempfile,
 )
+from .transcript_builder import TranscriptBuilder
 
 
 def _make_jsonl(*entries: dict) -> str:
@@ -282,3 +290,610 @@ class TestStripProgressEntries:
         lines = result.strip().split("\n")
         asst_entry = json.loads(lines[-1])
         assert asst_entry["parentUuid"] == "u1"  # reparented
+
+
+# --- read_cli_session_file ---
+
+
+class TestReadCliSessionFile:
+    def test_no_matching_files_returns_none(self, tmp_path, monkeypatch):
+        """read_cli_session_file returns None when no .jsonl files exist."""
+        # Create a project dir with no jsonl files
+        project_dir = tmp_path / "projects" / "encoded-cwd"
+        project_dir.mkdir(parents=True)
+        monkeypatch.setattr(
+            "backend.copilot.sdk.transcript._cli_project_dir",
+            lambda sdk_cwd: str(project_dir),
+        )
+        assert read_cli_session_file("/fake/cwd") is None
+
+    def test_one_jsonl_file_returns_content(self, tmp_path, monkeypatch):
+        """read_cli_session_file returns the content of a single .jsonl file."""
+        project_dir = tmp_path / "projects" / "encoded-cwd"
+        project_dir.mkdir(parents=True)
+        jsonl_file = project_dir / "session.jsonl"
+        jsonl_file.write_text("line1\nline2\n")
+        monkeypatch.setattr(
+            "backend.copilot.sdk.transcript._cli_project_dir",
+            lambda sdk_cwd: str(project_dir),
+        )
+        result = read_cli_session_file("/fake/cwd")
+        assert result == "line1\nline2\n"
+
+    def test_symlink_escaping_project_dir_is_skipped(self, tmp_path, monkeypatch):
+        """read_cli_session_file skips symlinks that escape the project dir."""
+        project_dir = tmp_path / "projects" / "encoded-cwd"
+        project_dir.mkdir(parents=True)
+
+        # Create a file outside the project dir
+        outside = tmp_path / "outside"
+        outside.mkdir()
+        outside_file = outside / "evil.jsonl"
+        outside_file.write_text("should not be read\n")
+
+        # Symlink from inside project_dir to outside file
+        symlink = project_dir / "evil.jsonl"
+        symlink.symlink_to(outside_file)
+
+        monkeypatch.setattr(
+            "backend.copilot.sdk.transcript._cli_project_dir",
+            lambda sdk_cwd: str(project_dir),
+        )
+        # The symlink target resolves outside project_dir, so it should be skipped
+        result = read_cli_session_file("/fake/cwd")
+        assert result is None
+
+
+# --- _cli_project_dir ---
+
+
+class TestCliProjectDir:
+    def test_returns_none_for_path_traversal(self, tmp_path, monkeypatch):
+        """_cli_project_dir returns None when the project dir symlink escapes projects base."""
+        config_dir = tmp_path / "config"
+        config_dir.mkdir()
+        projects_dir = config_dir / "projects"
+        projects_dir.mkdir()
+
+        monkeypatch.setenv("CLAUDE_CONFIG_DIR", str(config_dir))
+
+        # Create a symlink inside projects/ that points outside of it.
+        # _cli_project_dir encodes the cwd as all-alnum-hyphens, so use a
+        # cwd whose encoded form matches the symlink name we create.
+        evil_target = tmp_path / "escaped"
+        evil_target.mkdir()
+
+        # The encoded form of "/evil/cwd" is "-evil-cwd"
+        symlink_path = projects_dir / "-evil-cwd"
+        symlink_path.symlink_to(evil_target)
+
+        result = _cli_project_dir("/evil/cwd")
+        assert result is None
+
+
+# --- delete_transcript ---
+
+
+class TestDeleteTranscript:
+    @pytest.mark.asyncio
+    async def test_deletes_both_jsonl_and_meta(self):
+        """delete_transcript removes both the .jsonl and .meta.json files."""
+        mock_storage = AsyncMock()
+        mock_storage.delete = AsyncMock()
+
+        with patch(
+            "backend.util.workspace_storage.get_workspace_storage",
+            new_callable=AsyncMock,
+            return_value=mock_storage,
+        ):
+            await delete_transcript("user-123", "session-456")
+
+        assert mock_storage.delete.call_count == 2
+        paths = [call.args[0] for call in mock_storage.delete.call_args_list]
+        assert any(p.endswith(".jsonl") for p in paths)
+        assert any(p.endswith(".meta.json") for p in paths)
+
+    @pytest.mark.asyncio
+    async def test_continues_on_jsonl_delete_failure(self):
+        """If .jsonl delete fails, .meta.json delete is still attempted."""
+        mock_storage = AsyncMock()
+        mock_storage.delete = AsyncMock(
+            side_effect=[Exception("jsonl delete failed"), None]
+        )
+
+        with patch(
+            "backend.util.workspace_storage.get_workspace_storage",
+            new_callable=AsyncMock,
+            return_value=mock_storage,
+        ):
+            # Should not raise
+            await delete_transcript("user-123", "session-456")
+
+        assert mock_storage.delete.call_count == 2
+
+    @pytest.mark.asyncio
+    async def test_handles_meta_delete_failure(self):
+        """If .meta.json delete fails, no exception propagates."""
+        mock_storage = AsyncMock()
+        mock_storage.delete = AsyncMock(
+            side_effect=[None, Exception("meta delete failed")]
+        )
+
+        with patch(
+            "backend.util.workspace_storage.get_workspace_storage",
+            new_callable=AsyncMock,
+            return_value=mock_storage,
+        ):
+            # Should not raise
+            await delete_transcript("user-123", "session-456")
+
+
+# --- read_compacted_entries ---
+
+
+COMPACT_SUMMARY = {
+    "type": "summary",
+    "uuid": "cs1",
+    "isCompactSummary": True,
+    "message": {"role": "assistant", "content": "compacted context"},
+}
+POST_COMPACT_ASST = {
+    "type": "assistant",
+    "uuid": "a2",
+    "parentUuid": "cs1",
+    "message": {"role": "assistant", "content": "response after compaction"},
+}
+
+
+class TestReadCompactedEntries:
+    def test_returns_summary_and_entries_after(self, tmp_path, monkeypatch):
+        """File with isCompactSummary entry returns summary + entries after."""
+        config_dir = tmp_path / "config"
+        projects_dir = config_dir / "projects"
+        session_dir = projects_dir / "proj"
+        session_dir.mkdir(parents=True)
+        monkeypatch.setenv("CLAUDE_CONFIG_DIR", str(config_dir))
+
+        pre_compact = {"type": "user", "uuid": "u1", "message": {"role": "user"}}
+        path = session_dir / "session.jsonl"
+        path.write_text(_make_jsonl(pre_compact, COMPACT_SUMMARY, POST_COMPACT_ASST))
+
+        result = read_compacted_entries(str(path))
+        assert result is not None
+        assert len(result) == 2
+        assert result[0]["isCompactSummary"] is True
+        assert result[1]["uuid"] == "a2"
+
+    def test_no_compact_summary_returns_none(self, tmp_path, monkeypatch):
+        """File without isCompactSummary returns None."""
+        config_dir = tmp_path / "config"
+        projects_dir = config_dir / "projects"
+        session_dir = projects_dir / "proj"
+        session_dir.mkdir(parents=True)
+        monkeypatch.setenv("CLAUDE_CONFIG_DIR", str(config_dir))
+
+        path = session_dir / "session.jsonl"
+        path.write_text(_make_jsonl(USER_MSG, ASST_MSG))
+
+        result = read_compacted_entries(str(path))
+        assert result is None
+
+    def test_file_not_found_returns_none(self, tmp_path, monkeypatch):
+        """Non-existent file returns None."""
+        config_dir = tmp_path / "config"
+        projects_dir = config_dir / "projects"
+        projects_dir.mkdir(parents=True)
+        monkeypatch.setenv("CLAUDE_CONFIG_DIR", str(config_dir))
+
+        result = read_compacted_entries(str(projects_dir / "missing.jsonl"))
+        assert result is None
+
+    def test_empty_path_returns_none(self):
+        """Empty string path returns None."""
+        result = read_compacted_entries("")
+        assert result is None
+
+    def test_malformed_json_lines_skipped(self, tmp_path, monkeypatch):
+        """Malformed JSON lines are skipped gracefully."""
+        config_dir = tmp_path / "config"
+        projects_dir = config_dir / "projects"
+        session_dir = projects_dir / "proj"
+        session_dir.mkdir(parents=True)
+        monkeypatch.setenv("CLAUDE_CONFIG_DIR", str(config_dir))
+
+        path = session_dir / "session.jsonl"
+        content = "not valid json\n" + json.dumps(COMPACT_SUMMARY) + "\n"
+        content += "also bad\n" + json.dumps(POST_COMPACT_ASST) + "\n"
+        path.write_text(content)
+
+        result = read_compacted_entries(str(path))
+        assert result is not None
+        assert len(result) == 2  # summary + post-compact assistant
+
+    def test_multiple_compact_summaries_uses_last(self, tmp_path, monkeypatch):
+        """When multiple isCompactSummary entries exist, uses the last one
+        (most recent compaction)."""
+        config_dir = tmp_path / "config"
+        projects_dir = config_dir / "projects"
+        session_dir = projects_dir / "proj"
+        session_dir.mkdir(parents=True)
+        monkeypatch.setenv("CLAUDE_CONFIG_DIR", str(config_dir))
+
+        second_summary = {
+            "type": "summary",
+            "uuid": "cs2",
+            "isCompactSummary": True,
+            "message": {"role": "assistant", "content": "second summary"},
+        }
+        path = session_dir / "session.jsonl"
+        path.write_text(_make_jsonl(COMPACT_SUMMARY, POST_COMPACT_ASST, second_summary))
+
+        result = read_compacted_entries(str(path))
+        assert result is not None
+        # Last summary found, so only cs2 returned
+        assert len(result) == 1
+        assert result[0]["uuid"] == "cs2"
+
+    def test_path_outside_projects_base_returns_none(self, tmp_path, monkeypatch):
+        """Transcript path outside the projects directory is rejected."""
+        config_dir = tmp_path / "config"
+        (config_dir / "projects").mkdir(parents=True)
+        monkeypatch.setenv("CLAUDE_CONFIG_DIR", str(config_dir))
+
+        evil_file = tmp_path / "evil.jsonl"
+        evil_file.write_text(_make_jsonl(COMPACT_SUMMARY))
+
+        result = read_compacted_entries(str(evil_file))
+        assert result is None
+
+
+# --- TranscriptBuilder.replace_entries ---
+
+
+class TestTranscriptBuilderReplaceEntries:
+    def test_replaces_existing_entries(self):
+        """replace_entries replaces all entries with compacted ones."""
+        builder = TranscriptBuilder()
+        builder.append_user("hello")
+        builder.append_assistant([{"type": "text", "text": "world"}])
+        assert builder.entry_count == 2
+
+        compacted = [
+            {
+                "type": "user",
+                "uuid": "cs1",
+                "isCompactSummary": True,
+                "message": {"role": "user", "content": "compacted summary"},
+            },
+            {
+                "type": "assistant",
+                "uuid": "a1",
+                "parentUuid": "cs1",
+                "message": {"role": "assistant", "content": "response"},
+            },
+        ]
+        builder.replace_entries(compacted)
+        assert builder.entry_count == 2
+        output = builder.to_jsonl()
+        entries = [json.loads(line) for line in output.strip().split("\n")]
+        assert entries[0]["uuid"] == "cs1"
+        assert entries[1]["uuid"] == "a1"
+
+    def test_filters_strippable_types(self):
+        """Strippable types are filtered out during replace."""
+        builder = TranscriptBuilder()
+        compacted = [
+            {
+                "type": "user",
+                "uuid": "cs1",
+                "message": {"role": "user", "content": "compacted summary"},
+            },
+            {"type": "progress", "uuid": "p1", "message": {}},
+            {"type": "summary", "uuid": "s1", "message": {}},
+            {
+                "type": "assistant",
+                "uuid": "a1",
+                "parentUuid": "cs1",
+                "message": {"role": "assistant", "content": "hi"},
+            },
+        ]
+        builder.replace_entries(compacted)
+        assert builder.entry_count == 2  # progress and summary were filtered
+
+    def test_maintains_last_uuid_chain(self):
+        """After replace, _last_uuid is the last entry's uuid."""
+        builder = TranscriptBuilder()
+        compacted = [
+            {
+                "type": "user",
+                "uuid": "cs1",
+                "message": {"role": "user", "content": "compacted summary"},
+            },
+            {
+                "type": "assistant",
+                "uuid": "a1",
+                "parentUuid": "cs1",
+                "message": {"role": "assistant", "content": "hi"},
+            },
+        ]
+        builder.replace_entries(compacted)
+        # Appending a new user message should chain to a1
+        builder.append_user("next question")
+        output = builder.to_jsonl()
+        entries = [json.loads(line) for line in output.strip().split("\n")]
+        assert entries[-1]["parentUuid"] == "a1"
+
+    def test_empty_entries_list_keeps_existing(self):
+        """Replacing with empty list keeps existing entries (safety check)."""
+        builder = TranscriptBuilder()
+        builder.append_user("hello")
+        builder.replace_entries([])
+        # Empty input is treated as corrupt — existing entries preserved
+        assert builder.entry_count == 1
+        assert not builder.is_empty
+
+
+# --- TranscriptBuilder.load_previous with compacted content ---
+
+
+class TestTranscriptBuilderLoadPreviousCompacted:
+    def test_preserves_compact_summary_entry(self):
+        """load_previous preserves isCompactSummary entries even though
+        their type is 'summary' (which is in STRIPPABLE_TYPES)."""
+        compacted_content = _make_jsonl(COMPACT_SUMMARY, POST_COMPACT_ASST)
+        builder = TranscriptBuilder()
+        builder.load_previous(compacted_content)
+        assert builder.entry_count == 2
+        output = builder.to_jsonl()
+        entries = [json.loads(line) for line in output.strip().split("\n")]
+        assert entries[0]["type"] == "summary"
+        assert entries[0]["uuid"] == "cs1"
+        assert entries[1]["uuid"] == "a2"
+
+    def test_strips_regular_summary_entries(self):
+        """Regular summary entries (without isCompactSummary) are still stripped."""
+        regular_summary = {"type": "summary", "uuid": "s1", "message": {"content": "x"}}
+        content = _make_jsonl(regular_summary, POST_COMPACT_ASST)
+        builder = TranscriptBuilder()
+        builder.load_previous(content)
+        assert builder.entry_count == 1  # Only the assistant entry
+
+
+# --- End-to-end compaction flow (simulates service.py) ---
+
+
+class TestCompactionFlowIntegration:
+    """Simulate the full compaction flow as it happens in service.py:
+
+    1. TranscriptBuilder loads a previous transcript (download)
+    2. New messages are appended (user query + assistant response)
+    3. CompactionTracker fires (PreCompact hook → emit_start → emit_end)
+    4. read_compacted_entries reads the CLI session file
+    5. TranscriptBuilder.replace_entries syncs with CLI state
+    6. Final to_jsonl() produces the correct output (upload)
+    """
+
+    def test_full_compaction_roundtrip(self, tmp_path, monkeypatch):
+        """Full roundtrip: load → append → compact → replace → export."""
+        # Setup: create a CLI session file with pre-compact + compaction entries
+        config_dir = tmp_path / "config"
+        projects_dir = config_dir / "projects"
+        session_dir = projects_dir / "proj"
+        session_dir.mkdir(parents=True)
+        monkeypatch.setenv("CLAUDE_CONFIG_DIR", str(config_dir))
+
+        # Simulate a transcript with old messages, then a compaction summary
+        old_user = {
+            "type": "user",
+            "uuid": "u1",
+            "message": {"role": "user", "content": "old question"},
+        }
+        old_asst = {
+            "type": "assistant",
+            "uuid": "a1",
+            "parentUuid": "u1",
+            "message": {"role": "assistant", "content": "old answer"},
+        }
+        compact_summary = {
+            "type": "summary",
+            "uuid": "cs1",
+            "isCompactSummary": True,
+            "message": {"role": "user", "content": "compacted summary of conversation"},
+        }
+        post_compact_asst = {
+            "type": "assistant",
+            "uuid": "a2",
+            "parentUuid": "cs1",
+            "message": {"role": "assistant", "content": "response after compaction"},
+        }
+        session_file = session_dir / "session.jsonl"
+        session_file.write_text(
+            _make_jsonl(old_user, old_asst, compact_summary, post_compact_asst)
+        )
+
+        # Step 1: TranscriptBuilder loads previous transcript (simulates download)
+        # The previous transcript would have the OLD entries (pre-compaction)
+        previous_transcript = _make_jsonl(old_user, old_asst)
+        builder = TranscriptBuilder()
+        builder.load_previous(previous_transcript)
+        assert builder.entry_count == 2
+
+        # Step 2: New messages appended during the current query
+        builder.append_user("new question")
+        builder.append_assistant([{"type": "text", "text": "new answer"}])
+        assert builder.entry_count == 4
+
+        # Step 3: read_compacted_entries reads the CLI session file
+        compacted = read_compacted_entries(str(session_file))
+        assert compacted is not None
+        assert len(compacted) == 2  # compact_summary + post_compact_asst
+        assert compacted[0]["isCompactSummary"] is True
+
+        # Step 4: replace_entries syncs builder with CLI state
+        builder.replace_entries(compacted)
+        assert builder.entry_count == 2  # Only compacted entries now
+
+        # Step 5: Append post-compaction messages (continuing the stream)
+        builder.append_user("follow-up question")
+        assert builder.entry_count == 3
+
+        # Step 6: Export and verify
+        output = builder.to_jsonl()
+        entries = [json.loads(line) for line in output.strip().split("\n")]
+        assert len(entries) == 3
+        # First entry is the compaction summary
+        assert entries[0]["type"] == "summary"
+        assert entries[0]["uuid"] == "cs1"
+        # Second is the post-compact assistant
+        assert entries[1]["uuid"] == "a2"
+        # Third is our follow-up, parented to the last compacted entry
+        assert entries[2]["type"] == "user"
+        assert entries[2]["parentUuid"] == "a2"
+
+    def test_compaction_preserves_chain_across_multiple_compactions(
+        self, tmp_path, monkeypatch
+    ):
+        """Two compactions: first compacts old history, second compacts the first."""
+        config_dir = tmp_path / "config"
+        projects_dir = config_dir / "projects"
+        session_dir = projects_dir / "proj"
+        session_dir.mkdir(parents=True)
+        monkeypatch.setenv("CLAUDE_CONFIG_DIR", str(config_dir))
+
+        # First compaction
+        first_summary = {
+            "type": "summary",
+            "uuid": "cs1",
+            "isCompactSummary": True,
+            "message": {"role": "user", "content": "first summary"},
+        }
+        mid_asst = {
+            "type": "assistant",
+            "uuid": "a1",
+            "parentUuid": "cs1",
+            "message": {"role": "assistant", "content": "mid response"},
+        }
+        # Second compaction (compacts the first summary + mid_asst)
+        second_summary = {
+            "type": "summary",
+            "uuid": "cs2",
+            "isCompactSummary": True,
+            "message": {"role": "user", "content": "second summary"},
+        }
+        final_asst = {
+            "type": "assistant",
+            "uuid": "a2",
+            "parentUuid": "cs2",
+            "message": {"role": "assistant", "content": "final response"},
+        }
+
+        session_file = session_dir / "session.jsonl"
+        session_file.write_text(
+            _make_jsonl(first_summary, mid_asst, second_summary, final_asst)
+        )
+
+        # read_compacted_entries should find the LAST summary
+        compacted = read_compacted_entries(str(session_file))
+        assert compacted is not None
+        assert len(compacted) == 2  # second_summary + final_asst
+        assert compacted[0]["uuid"] == "cs2"
+
+        # Apply to builder
+        builder = TranscriptBuilder()
+        builder.append_user("old stuff")
+        builder.append_assistant([{"type": "text", "text": "old response"}])
+        builder.replace_entries(compacted)
+        assert builder.entry_count == 2
+
+        # New message chains correctly
+        builder.append_user("after second compaction")
+        output = builder.to_jsonl()
+        entries = [json.loads(line) for line in output.strip().split("\n")]
+        assert entries[-1]["parentUuid"] == "a2"
+
+    def test_strip_progress_preserves_compact_summaries(self):
+        """strip_progress_entries doesn't strip isCompactSummary entries
+        even though their type is 'summary' (in STRIPPABLE_TYPES)."""
+        compact_summary = {
+            "type": "summary",
+            "uuid": "cs1",
+            "isCompactSummary": True,
+            "message": {"role": "user", "content": "compacted"},
+        }
+        regular_summary = {"type": "summary", "uuid": "s1", "message": {"content": "x"}}
+        progress = {"type": "progress", "uuid": "p1", "data": {"stdout": "..."}}
+        user = {
+            "type": "user",
+            "uuid": "u1",
+            "message": {"role": "user", "content": "hi"},
+        }
+
+        content = _make_jsonl(compact_summary, regular_summary, progress, user)
+        stripped = strip_progress_entries(content)
+        stripped_entries = [
+            json.loads(line) for line in stripped.strip().split("\n") if line.strip()
+        ]
+
+        uuids = [e.get("uuid") for e in stripped_entries]
+        # compact_summary kept, regular_summary stripped, progress stripped, user kept
+        assert "cs1" in uuids  # compact summary preserved
+        assert "s1" not in uuids  # regular summary stripped
+        assert "p1" not in uuids  # progress stripped
+        assert "u1" in uuids  # user kept
+
+    def test_builder_load_then_replace_then_export_roundtrip(self):
+        """Load a compacted transcript, replace with new compaction, export.
+        Simulates two consecutive turns with compaction each time."""
+        # Turn 1: load compacted transcript
+        compact1 = {
+            "type": "summary",
+            "uuid": "cs1",
+            "isCompactSummary": True,
+            "message": {"role": "user", "content": "summary v1"},
+        }
+        asst1 = {
+            "type": "assistant",
+            "uuid": "a1",
+            "parentUuid": "cs1",
+            "message": {"role": "assistant", "content": "response 1"},
+        }
+        builder = TranscriptBuilder()
+        builder.load_previous(_make_jsonl(compact1, asst1))
+        assert builder.entry_count == 2
+
+        # Turn 1: append new messages
+        builder.append_user("question")
+        builder.append_assistant([{"type": "text", "text": "answer"}])
+        assert builder.entry_count == 4
+
+        # Turn 1: compaction fires — replace with new compacted state
+        compact2 = {
+            "type": "summary",
+            "uuid": "cs2",
+            "isCompactSummary": True,
+            "message": {"role": "user", "content": "summary v2"},
+        }
+        asst2 = {
+            "type": "assistant",
+            "uuid": "a2",
+            "parentUuid": "cs2",
+            "message": {"role": "assistant", "content": "continuing"},
+        }
+        builder.replace_entries([compact2, asst2])
+        assert builder.entry_count == 2
+
+        # Export (this goes to cloud storage for next turn's download)
+        output = builder.to_jsonl()
+        lines = [json.loads(line) for line in output.strip().split("\n")]
+        assert lines[0]["uuid"] == "cs2"
+        assert lines[0]["type"] == "summary"
+        assert lines[1]["uuid"] == "a2"
+
+        # Turn 2: fresh builder loads the exported transcript
+        builder2 = TranscriptBuilder()
+        builder2.load_previous(output)
+        assert builder2.entry_count == 2
+        builder2.append_user("turn 2 question")
+        output2 = builder2.to_jsonl()
+        lines2 = [json.loads(line) for line in output2.strip().split("\n")]
+        assert lines2[-1]["parentUuid"] == "a2"

autogpt_platform/backend/backend/copilot/tools/agent_generator/validator.py

@@ -935,5 +935,5 @@ class AgentValidator:
             for i, error in enumerate(self.errors, 1):
                 error_message += f"{i}. {error}\n"
 
-            logger.error(f"Agent validation failed: {error_message}")
+            logger.warning(f"Agent validation failed: {error_message}")
             return False, error_message

autogpt_platform/backend/backend/copilot/tools/agent_search.py

@@ -3,8 +3,7 @@
 from __future__ import annotations
 
 import logging
-import re
-from typing import TYPE_CHECKING, Literal
+from typing import TYPE_CHECKING, Any, Literal
 
 if TYPE_CHECKING:
     from backend.api.features.library.model import LibraryAgent
@@ -19,16 +18,12 @@ from .models import (
     NoResultsResponse,
     ToolResponseBase,
 )
+from .utils import is_creator_slug, is_uuid
 
 logger = logging.getLogger(__name__)
 
 SearchSource = Literal["marketplace", "library"]
 
-_UUID_PATTERN = re.compile(
-    r"^[a-f0-9]{8}-[a-f0-9]{4}-4[a-f0-9]{3}-[89ab][a-f0-9]{3}-[a-f0-9]{12}$",
-    re.IGNORECASE,
-)
-
 # Keywords that should be treated as "list all" rather than a literal search
 _LIST_ALL_KEYWORDS = frozenset({"all", "*", "everything", "any", ""})
 
@@ -39,149 +34,158 @@ async def search_agents(
     session_id: str | None = None,
     user_id: str | None = None,
 ) -> ToolResponseBase:
-    """
-    Search for agents in marketplace or user library.
+    """Search for agents in marketplace or user library."""
+    if source == "marketplace":
+        return await _search_marketplace(query, session_id)
+    else:
+        return await _search_library(query, session_id, user_id)
 
-    For library searches, keywords like "all", "*", "everything", or an empty
-    query will list all agents without filtering.
 
-    Args:
-        query: Search query string. Special keywords list all library agents.
-        source: "marketplace" or "library"
-        session_id: Chat session ID
-        user_id: User ID (required for library search)
-
-    Returns:
-        AgentsFoundResponse, NoResultsResponse, or ErrorResponse
-    """
-    # Normalize list-all keywords to empty string for library searches
-    if source == "library" and query.lower().strip() in _LIST_ALL_KEYWORDS:
-        query = ""
-
-    if source == "marketplace" and not query:
+async def _search_marketplace(query: str, session_id: str | None) -> ToolResponseBase:
+    query = query.strip()
+    if not query:
         return ErrorResponse(
             message="Please provide a search query", session_id=session_id
         )
 
-    if source == "library" and not user_id:
-        return ErrorResponse(
-            message="User authentication required to search library",
-            session_id=session_id,
-        )
-
     agents: list[AgentInfo] = []
     try:
-        if source == "marketplace":
+        # Direct lookup if query matches "creator/slug" pattern
+        if is_creator_slug(query):
+            logger.info(f"Query looks like creator/slug, trying direct lookup: {query}")
+            creator, slug = query.split("/", 1)
+            agent_info = await _get_marketplace_agent_by_slug(creator, slug)
+            if agent_info:
+                agents.append(agent_info)
+
+        if not agents:
             logger.info(f"Searching marketplace for: {query}")
             results = await store_db().get_store_agents(search_query=query, page_size=5)
             for agent in results.agents:
-                agents.append(
-                    AgentInfo(
-                        id=f"{agent.creator}/{agent.slug}",
-                        name=agent.agent_name,
-                        description=agent.description or "",
-                        source="marketplace",
-                        in_library=False,
-                        creator=agent.creator,
-                        category="general",
-                        rating=agent.rating,
-                        runs=agent.runs,
-                        is_featured=False,
-                    )
-                )
-        else:
-            if _is_uuid(query):
-                logger.info(f"Query looks like UUID, trying direct lookup: {query}")
-                agent = await _get_library_agent_by_id(user_id, query)  # type: ignore[arg-type]
-                if agent:
-                    agents.append(agent)
-                    logger.info(f"Found agent by direct ID lookup: {agent.name}")
-
-            if not agents:
-                search_term = query or None
-                logger.info(
-                    f"{'Listing all agents in' if not query else 'Searching'} "
-                    f"user library{'' if not query else f' for: {query}'}"
-                )
-                results = await library_db().list_library_agents(
-                    user_id=user_id,  # type: ignore[arg-type]
-                    search_term=search_term,
-                    page_size=50 if not query else 10,
-                )
-                for agent in results.agents:
-                    agents.append(_library_agent_to_info(agent))
-        logger.info(f"Found {len(agents)} agents in {source}")
+                agents.append(_marketplace_agent_to_info(agent))
     except NotFoundError:
         pass
     except DatabaseError as e:
-        logger.error(f"Error searching {source}: {e}", exc_info=True)
+        logger.error(f"Error searching marketplace: {e}", exc_info=True)
         return ErrorResponse(
-            message=f"Failed to search {source}. Please try again.",
+            message="Failed to search marketplace. Please try again.",
             error=str(e),
             session_id=session_id,
         )
 
     if not agents:
-        if source == "marketplace":
-            suggestions = [
-                "Try more general terms",
-                "Browse categories in the marketplace",
-                "Check spelling",
-            ]
-            no_results_msg = (
+        return NoResultsResponse(
+            message=(
                 f"No agents found matching '{query}'. Let the user know they can "
                 "try different keywords or browse the marketplace. Also let them "
                 "know you can create a custom agent for them based on their needs."
+            ),
+            suggestions=[
+                "Try more general terms",
+                "Browse categories in the marketplace",
+                "Check spelling",
+            ],
+            session_id=session_id,
+        )
+
+    return AgentsFoundResponse(
+        message=(
+            "Now you have found some options for the user to choose from. "
+            "You can add a link to a recommended agent at: /marketplace/agent/agent_id "
+            "Please ask the user if they would like to use any of these agents. "
+            "Let the user know we can create a custom agent for them based on their needs."
+        ),
+        title=f"Found {len(agents)} agent{'s' if len(agents) != 1 else ''} for '{query}'",
+        agents=agents,
+        count=len(agents),
+        session_id=session_id,
+    )
+
+
+async def _search_library(
+    query: str, session_id: str | None, user_id: str | None
+) -> ToolResponseBase:
+    if not user_id:
+        return ErrorResponse(
+            message="User authentication required to search library",
+            session_id=session_id,
+        )
+
+    query = query.strip()
+    # Normalize list-all keywords to empty string
+    if query.lower() in _LIST_ALL_KEYWORDS:
+        query = ""
+
+    agents: list[AgentInfo] = []
+    try:
+        if is_uuid(query):
+            logger.info(f"Query looks like UUID, trying direct lookup: {query}")
+            agent = await _get_library_agent_by_id(user_id, query)
+            if agent:
+                agents.append(agent)
+
+        if not agents:
+            logger.info(
+                f"{'Listing all agents in' if not query else 'Searching'} "
+                f"user library{'' if not query else f' for: {query}'}"
             )
-        elif not query:
-            # User asked to list all but library is empty
-            suggestions = [
-                "Browse the marketplace to find and add agents",
-                "Use find_agent to search the marketplace",
-            ]
-            no_results_msg = (
-                "Your library is empty. Let the user know they can browse the "
-                "marketplace to find agents, or you can create a custom agent "
-                "for them based on their needs."
+            results = await library_db().list_library_agents(
+                user_id=user_id,
+                search_term=query or None,
+                page_size=50 if not query else 10,
             )
-        else:
-            suggestions = [
-                "Try different keywords",
-                "Use find_agent to search the marketplace",
-                "Check your library at /library",
-            ]
-            no_results_msg = (
+            for agent in results.agents:
+                agents.append(_library_agent_to_info(agent))
+    except NotFoundError:
+        pass
+    except DatabaseError as e:
+        logger.error(f"Error searching library: {e}", exc_info=True)
+        return ErrorResponse(
+            message="Failed to search library. Please try again.",
+            error=str(e),
+            session_id=session_id,
+        )
+
+    if not agents:
+        if not query:
+            return NoResultsResponse(
+                message=(
+                    "Your library is empty. Let the user know they can browse the "
+                    "marketplace to find agents, or you can create a custom agent "
+                    "for them based on their needs."
+                ),
+                suggestions=[
+                    "Browse the marketplace to find and add agents",
+                    "Use find_agent to search the marketplace",
+                ],
+                session_id=session_id,
+            )
+        return NoResultsResponse(
+            message=(
                 f"No agents matching '{query}' found in your library. Let the "
                 "user know you can create a custom agent for them based on "
                 "their needs."
-            )
-        return NoResultsResponse(
-            message=no_results_msg, session_id=session_id, suggestions=suggestions
+            ),
+            suggestions=[
+                "Try different keywords",
+                "Use find_agent to search the marketplace",
+                "Check your library at /library",
+            ],
+            session_id=session_id,
         )
 
-    if source == "marketplace":
-        title = (
-            f"Found {len(agents)} agent{'s' if len(agents) != 1 else ''} for '{query}'"
-        )
-    elif not query:
+    if not query:
         title = f"Found {len(agents)} agent{'s' if len(agents) != 1 else ''} in your library"
     else:
         title = f"Found {len(agents)} agent{'s' if len(agents) != 1 else ''} in your library for '{query}'"
 
-    message = (
-        "Now you have found some options for the user to choose from. "
-        "You can add a link to a recommended agent at: /marketplace/agent/agent_id "
-        "Please ask the user if they would like to use any of these agents. "
-        "Let the user know we can create a custom agent for them based on their needs."
-        if source == "marketplace"
-        else "Found agents in the user's library. You can provide a link to view "
-        "an agent at: /library/agents/{agent_id}. Use agent_output to get "
-        "execution results, or run_agent to execute. Let the user know we can "
-        "create a custom agent for them based on their needs."
-    )
-
     return AgentsFoundResponse(
-        message=message,
+        message=(
+            "Found agents in the user's library. You can provide a link to view "
+            "an agent at: /library/agents/{agent_id}. Use agent_output to get "
+            "execution results, or run_agent to execute. Let the user know we can "
+            "create a custom agent for them based on their needs."
+        ),
         title=title,
         agents=agents,
         count=len(agents),
@@ -189,9 +193,20 @@ async def search_agents(
     )
 
 
-def _is_uuid(text: str) -> bool:
-    """Check if text is a valid UUID v4."""
-    return bool(_UUID_PATTERN.match(text.strip()))
+def _marketplace_agent_to_info(agent: Any) -> AgentInfo:
+    """Convert a marketplace agent (StoreAgent or StoreAgentDetails) to an AgentInfo."""
+    return AgentInfo(
+        id=f"{agent.creator}/{agent.slug}",
+        name=agent.agent_name,
+        description=agent.description or "",
+        source="marketplace",
+        in_library=False,
+        creator=agent.creator,
+        category="general",
+        rating=agent.rating,
+        runs=agent.runs,
+        is_featured=False,
+    )
 
 
 def _library_agent_to_info(agent: LibraryAgent) -> AgentInfo:
@@ -214,6 +229,23 @@ def _library_agent_to_info(agent: LibraryAgent) -> AgentInfo:
     )
 
 
+async def _get_marketplace_agent_by_slug(creator: str, slug: str) -> AgentInfo | None:
+    """Fetch a marketplace agent by creator/slug identifier."""
+    try:
+        details = await store_db().get_store_agent_details(creator, slug)
+        return _marketplace_agent_to_info(details)
+    except NotFoundError:
+        pass
+    except DatabaseError:
+        raise
+    except Exception as e:
+        logger.warning(
+            f"Could not fetch marketplace agent {creator}/{slug}: {e}",
+            exc_info=True,
+        )
+    return None
+
+
 async def _get_library_agent_by_id(user_id: str, agent_id: str) -> AgentInfo | None:
     """Fetch a library agent by ID (library agent ID or graph_id).
 
@@ -226,10 +258,9 @@ async def _get_library_agent_by_id(user_id: str, agent_id: str) -> AgentInfo | N
     try:
         agent = await lib_db.get_library_agent_by_graph_id(user_id, agent_id)
         if agent:
-            logger.debug(f"Found library agent by graph_id: {agent.name}")
             return _library_agent_to_info(agent)
     except NotFoundError:
-        logger.debug(f"Library agent not found by graph_id: {agent_id}")
+        pass
     except DatabaseError:
         raise
     except Exception as e:
@@ -241,10 +272,9 @@ async def _get_library_agent_by_id(user_id: str, agent_id: str) -> AgentInfo | N
     try:
         agent = await lib_db.get_library_agent(agent_id, user_id)
         if agent:
-            logger.debug(f"Found library agent by library_id: {agent.name}")
             return _library_agent_to_info(agent)
     except NotFoundError:
-        logger.debug(f"Library agent not found by library_id: {agent_id}")
+        pass
     except DatabaseError:
         raise
     except Exception as e:

autogpt_platform/backend/backend/copilot/tools/agent_search_test.py

@@ -0,0 +1,170 @@
+"""Tests for agent search direct lookup functionality."""
+
+from unittest.mock import AsyncMock, MagicMock, patch
+
+import pytest
+
+from .agent_search import search_agents
+from .models import AgentsFoundResponse, NoResultsResponse
+
+_TEST_USER_ID = "test-user-agent-search"
+
+
+class TestMarketplaceSlugLookup:
+    """Tests for creator/slug direct lookup in marketplace search."""
+
+    @pytest.mark.asyncio(loop_scope="session")
+    async def test_slug_lookup_found(self):
+        """creator/slug query returns the agent directly."""
+        mock_details = MagicMock()
+        mock_details.creator = "testuser"
+        mock_details.slug = "my-agent"
+        mock_details.agent_name = "My Agent"
+        mock_details.description = "A test agent"
+        mock_details.rating = 4.5
+        mock_details.runs = 100
+
+        mock_store = MagicMock()
+        mock_store.get_store_agent_details = AsyncMock(return_value=mock_details)
+
+        with patch(
+            "backend.copilot.tools.agent_search.store_db",
+            return_value=mock_store,
+        ):
+            response = await search_agents(
+                query="testuser/my-agent",
+                source="marketplace",
+                session_id="test-session",
+            )
+
+        assert isinstance(response, AgentsFoundResponse)
+        assert response.count == 1
+        assert response.agents[0].id == "testuser/my-agent"
+        assert response.agents[0].name == "My Agent"
+
+    @pytest.mark.asyncio(loop_scope="session")
+    async def test_slug_lookup_not_found_falls_back_to_search(self):
+        """creator/slug not found falls back to general search."""
+        from backend.util.exceptions import NotFoundError
+
+        mock_store = MagicMock()
+        mock_store.get_store_agent_details = AsyncMock(side_effect=NotFoundError(""))
+
+        # Fallback search returns results
+        mock_search_results = MagicMock()
+        mock_agent = MagicMock()
+        mock_agent.creator = "other"
+        mock_agent.slug = "similar-agent"
+        mock_agent.agent_name = "Similar Agent"
+        mock_agent.description = "A similar agent"
+        mock_agent.rating = 3.0
+        mock_agent.runs = 50
+        mock_search_results.agents = [mock_agent]
+
+        mock_store.get_store_agents = AsyncMock(return_value=mock_search_results)
+
+        with patch(
+            "backend.copilot.tools.agent_search.store_db",
+            return_value=mock_store,
+        ):
+            response = await search_agents(
+                query="testuser/my-agent",
+                source="marketplace",
+                session_id="test-session",
+            )
+
+        assert isinstance(response, AgentsFoundResponse)
+        assert response.count == 1
+        assert response.agents[0].id == "other/similar-agent"
+
+    @pytest.mark.asyncio(loop_scope="session")
+    async def test_slug_lookup_not_found_no_search_results(self):
+        """creator/slug not found and search returns nothing."""
+        from backend.util.exceptions import NotFoundError
+
+        mock_store = MagicMock()
+        mock_store.get_store_agent_details = AsyncMock(side_effect=NotFoundError(""))
+        mock_search_results = MagicMock()
+        mock_search_results.agents = []
+        mock_store.get_store_agents = AsyncMock(return_value=mock_search_results)
+
+        with patch(
+            "backend.copilot.tools.agent_search.store_db",
+            return_value=mock_store,
+        ):
+            response = await search_agents(
+                query="testuser/nonexistent",
+                source="marketplace",
+                session_id="test-session",
+            )
+
+        assert isinstance(response, NoResultsResponse)
+
+    @pytest.mark.asyncio(loop_scope="session")
+    async def test_non_slug_query_goes_to_search(self):
+        """Regular keyword query skips slug lookup and goes to search."""
+        mock_store = MagicMock()
+        mock_search_results = MagicMock()
+        mock_agent = MagicMock()
+        mock_agent.creator = "creator1"
+        mock_agent.slug = "email-agent"
+        mock_agent.agent_name = "Email Agent"
+        mock_agent.description = "Sends emails"
+        mock_agent.rating = 4.0
+        mock_agent.runs = 200
+        mock_search_results.agents = [mock_agent]
+        mock_store.get_store_agents = AsyncMock(return_value=mock_search_results)
+
+        with patch(
+            "backend.copilot.tools.agent_search.store_db",
+            return_value=mock_store,
+        ):
+            response = await search_agents(
+                query="email",
+                source="marketplace",
+                session_id="test-session",
+            )
+
+        assert isinstance(response, AgentsFoundResponse)
+        # get_store_agent_details should NOT have been called
+        mock_store.get_store_agent_details.assert_not_called()
+
+
+class TestLibraryUUIDLookup:
+    """Tests for UUID direct lookup in library search."""
+
+    @pytest.mark.asyncio(loop_scope="session")
+    async def test_uuid_lookup_found_by_graph_id(self):
+        """UUID query matching a graph_id returns the agent directly."""
+        agent_id = "a1b2c3d4-e5f6-4a7b-8c9d-0e1f2a3b4c5d"
+        mock_agent = MagicMock()
+        mock_agent.id = "lib-agent-id"
+        mock_agent.name = "My Library Agent"
+        mock_agent.description = "A library agent"
+        mock_agent.creator_name = "testuser"
+        mock_agent.status.value = "HEALTHY"
+        mock_agent.can_access_graph = True
+        mock_agent.has_external_trigger = False
+        mock_agent.new_output = False
+        mock_agent.graph_id = agent_id
+        mock_agent.graph_version = 1
+        mock_agent.input_schema = {}
+        mock_agent.output_schema = {}
+
+        mock_lib_db = MagicMock()
+        mock_lib_db.get_library_agent_by_graph_id = AsyncMock(return_value=mock_agent)
+
+        with patch(
+            "backend.copilot.tools.agent_search.library_db",
+            return_value=mock_lib_db,
+        ):
+            response = await search_agents(
+                query=agent_id,
+                source="library",
+                session_id="test-session",
+                user_id=_TEST_USER_ID,
+            )
+
+        assert isinstance(response, AgentsFoundResponse)
+        assert response.count == 1
+        assert response.agents[0].name == "My Library Agent"

autogpt_platform/backend/backend/copilot/tools/e2b_sandbox.py

@@ -21,9 +21,11 @@ Lifecycle
 Cost control
 ------------
 Sandboxes are created with a configurable ``on_timeout`` lifecycle action
-(default: ``"pause"``).  The explicit per-turn ``pause_sandbox()`` call is the
-primary mechanism; the lifecycle setting is a safety net.  Paused sandboxes are
-free.
+(default: ``"pause"``) and ``auto_resume`` (default: ``True``).  The explicit
+per-turn ``pause_sandbox()`` call is the primary mechanism; the lifecycle
+timeout is a safety net (default: 5 min).  ``auto_resume`` ensures that paused
+sandboxes wake transparently on SDK activity, making the aggressive safety-net
+timeout safe.  Paused sandboxes are free.
 
 The sandbox_id is stored in Redis.  The same key doubles as a creation lock:
 a ``"creating"`` sentinel value is written with a short TTL while a new sandbox
@@ -40,6 +42,7 @@ import logging
 from typing import Any, Awaitable, Callable, Literal
 
 from e2b import AsyncSandbox
+from e2b.sandbox.sandbox_api import SandboxLifecycle
 
 from backend.data.redis_client import get_redis_async
 
@@ -116,9 +119,10 @@ async def get_or_create_sandbox(
     removes the need for a separate lock key.
 
     *timeout* controls how long the e2b sandbox may run continuously before
-    the ``on_timeout`` lifecycle rule fires (default: 3 h).
+    the ``on_timeout`` lifecycle rule fires (default: 5 min).
     *on_timeout* controls what happens on timeout: ``"pause"`` (default, free)
-    or ``"kill"``.
+    or ``"kill"``.  When ``"pause"``, ``auto_resume`` is enabled so paused
+    sandboxes wake transparently on SDK activity.
     """
     redis = await get_redis_async()
     key = _sandbox_key(session_id)
@@ -156,11 +160,15 @@ async def get_or_create_sandbox(
 
         # We hold the slot — create the sandbox.
         try:
+            lifecycle = SandboxLifecycle(
+                on_timeout=on_timeout,
+                auto_resume=on_timeout == "pause",
+            )
             sandbox = await AsyncSandbox.create(
                 template=template,
                 api_key=api_key,
                 timeout=timeout,
-                lifecycle={"on_timeout": on_timeout},
+                lifecycle=lifecycle,
             )
             try:
                 await _set_stored_sandbox_id(session_id, sandbox.sandbox_id)

autogpt_platform/backend/backend/copilot/tools/e2b_sandbox_test.py

@@ -157,14 +157,17 @@ class TestGetOrCreateSandbox:
 
         assert result is new_sb
         mock_cls.create.assert_awaited_once()
-        # Verify lifecycle param is set
+        # Verify lifecycle: pause + auto_resume enabled
         _, kwargs = mock_cls.create.call_args
-        assert kwargs.get("lifecycle") == {"on_timeout": "pause"}
+        assert kwargs.get("lifecycle") == {
+            "on_timeout": "pause",
+            "auto_resume": True,
+        }
         # sandbox_id should be saved to Redis
         redis.set.assert_awaited()
 
     def test_create_with_on_timeout_kill(self):
-        """on_timeout='kill' is passed through to AsyncSandbox.create."""
+        """on_timeout='kill' disables auto_resume automatically."""
         new_sb = _mock_sandbox("sb-new")
         redis = _mock_redis(set_nx_result=True, stored_sandbox_id=None)
         with (
@@ -179,7 +182,10 @@ class TestGetOrCreateSandbox:
             )
 
         _, kwargs = mock_cls.create.call_args
-        assert kwargs.get("lifecycle") == {"on_timeout": "kill"}
+        assert kwargs.get("lifecycle") == {
+            "on_timeout": "kill",
+            "auto_resume": False,
+        }
 
     def test_create_failure_releases_slot(self):
         """If sandbox creation fails, the Redis creation slot is deleted."""

autogpt_platform/backend/backend/copilot/tools/find_agent.py

@@ -19,7 +19,8 @@ class FindAgentTool(BaseTool):
     @property
     def description(self) -> str:
         return (
-            "Discover agents from the marketplace based on capabilities and user needs."
+            "Discover agents from the marketplace based on capabilities and "
+            "user needs, or look up a specific agent by its creator/slug ID."
         )
 
     @property
@@ -29,7 +30,7 @@ class FindAgentTool(BaseTool):
             "properties": {
                 "query": {
                     "type": "string",
-                    "description": "Search query describing what the user wants to accomplish. Use single keywords for best results.",
+                    "description": "Search query describing what the user wants to accomplish, or a creator/slug ID (e.g. 'username/agent-name') for direct lookup. Use single keywords for best results.",
                 },
             },
             "required": ["query"],

autogpt_platform/backend/backend/copilot/tools/find_block.py

@@ -15,6 +15,7 @@ from .models import (
     ErrorResponse,
     NoResultsResponse,
 )
+from .utils import is_uuid
 
 logger = logging.getLogger(__name__)
 
@@ -52,7 +53,8 @@ class FindBlockTool(BaseTool):
     @property
     def description(self) -> str:
         return (
-            "Search for available blocks by name or description. "
+            "Search for available blocks by name or description, or look up a "
+            "specific block by its ID. "
             "Blocks are reusable components that perform specific tasks like "
             "sending emails, making API calls, processing text, etc. "
             "IMPORTANT: Use this tool FIRST to get the block's 'id' before calling run_block. "
@@ -68,7 +70,8 @@ class FindBlockTool(BaseTool):
                 "query": {
                     "type": "string",
                     "description": (
-                        "Search query to find blocks by name or description. "
+                        "Search query to find blocks by name or description, "
+                        "or a block ID (UUID) for direct lookup. "
                         "Use keywords like 'email', 'http', 'text', 'ai', etc."
                     ),
                 },
@@ -113,11 +116,77 @@ class FindBlockTool(BaseTool):
 
         if not query:
             return ErrorResponse(
-                message="Please provide a search query",
+                message="Please provide a search query or block ID",
                 session_id=session_id,
             )
 
         try:
+            # Direct ID lookup if query looks like a UUID
+            if is_uuid(query):
+                block = get_block(query.lower())
+                if block:
+                    if block.disabled:
+                        return NoResultsResponse(
+                            message=f"Block '{block.name}' (ID: {query}) is disabled and cannot be used.",
+                            suggestions=["Search for an alternative block by name"],
+                            session_id=session_id,
+                        )
+                    if (
+                        block.block_type in COPILOT_EXCLUDED_BLOCK_TYPES
+                        or block.id in COPILOT_EXCLUDED_BLOCK_IDS
+                    ):
+                        if block.block_type == BlockType.MCP_TOOL:
+                            return NoResultsResponse(
+                                message=(
+                                    f"Block '{block.name}' (ID: {block.id}) is not "
+                                    "runnable through find_block/run_block. Use "
+                                    "run_mcp_tool instead."
+                                ),
+                                suggestions=[
+                                    "Use run_mcp_tool to discover and run this MCP tool",
+                                    "Search for an alternative block by name",
+                                ],
+                                session_id=session_id,
+                            )
+                        return NoResultsResponse(
+                            message=(
+                                f"Block '{block.name}' (ID: {block.id}) is not available "
+                                "in CoPilot. It can only be used within agent graphs."
+                            ),
+                            suggestions=[
+                                "Search for an alternative block by name",
+                                "Use this block in an agent graph instead",
+                            ],
+                            session_id=session_id,
+                        )
+
+                    summary = BlockInfoSummary(
+                        id=block.id,
+                        name=block.name,
+                        description=(
+                            block.optimized_description or block.description or ""
+                        ),
+                        categories=[c.value for c in block.categories],
+                    )
+                    if include_schemas:
+                        info = block.get_info()
+                        summary.input_schema = info.inputSchema
+                        summary.output_schema = info.outputSchema
+                        summary.static_output = info.staticOutput
+
+                    return BlockListResponse(
+                        message=(
+                            f"Found block '{block.name}' by ID. "
+                            "To see inputs/outputs and execute it, use "
+                            "run_block with the block's 'id' - providing "
+                            "no inputs."
+                        ),
+                        blocks=[summary],
+                        count=1,
+                        query=query,
+                        session_id=session_id,
+                    )
+
             # Search for blocks using hybrid search
             results, total = await search().unified_hybrid_search(
                 query=query,

autogpt_platform/backend/backend/copilot/tools/find_block_test.py

@@ -499,3 +499,123 @@ class TestFindBlockFiltering:
         assert response.blocks[0].input_schema == input_schema
         assert response.blocks[0].output_schema == output_schema
         assert response.blocks[0].static_output is True
+
+
+class TestFindBlockDirectLookup:
+    """Tests for direct UUID lookup in FindBlockTool."""
+
+    @pytest.mark.asyncio(loop_scope="session")
+    async def test_uuid_lookup_found(self):
+        """UUID query returns the block directly without search."""
+        session = make_session(user_id=_TEST_USER_ID)
+        block_id = "a1b2c3d4-e5f6-4a7b-8c9d-0e1f2a3b4c5d"
+        block = make_mock_block(block_id, "Test Block", BlockType.STANDARD)
+
+        with patch(
+            "backend.copilot.tools.find_block.get_block",
+            return_value=block,
+        ):
+            tool = FindBlockTool()
+            response = await tool._execute(
+                user_id=_TEST_USER_ID, session=session, query=block_id
+            )
+
+        assert isinstance(response, BlockListResponse)
+        assert response.count == 1
+        assert response.blocks[0].id == block_id
+        assert response.blocks[0].name == "Test Block"
+
+    @pytest.mark.asyncio(loop_scope="session")
+    async def test_uuid_lookup_not_found_falls_through(self):
+        """UUID that doesn't match any block falls through to search."""
+        session = make_session(user_id=_TEST_USER_ID)
+        block_id = "a1b2c3d4-e5f6-4a7b-8c9d-0e1f2a3b4c5d"
+
+        mock_search_db = MagicMock()
+        mock_search_db.unified_hybrid_search = AsyncMock(return_value=([], 0))
+
+        with (
+            patch(
+                "backend.copilot.tools.find_block.get_block",
+                return_value=None,
+            ),
+            patch(
+                "backend.copilot.tools.find_block.search",
+                return_value=mock_search_db,
+            ),
+        ):
+            tool = FindBlockTool()
+            response = await tool._execute(
+                user_id=_TEST_USER_ID, session=session, query=block_id
+            )
+
+        from .models import NoResultsResponse
+
+        assert isinstance(response, NoResultsResponse)
+
+    @pytest.mark.asyncio(loop_scope="session")
+    async def test_uuid_lookup_disabled_block(self):
+        """UUID matching a disabled block returns NoResultsResponse."""
+        session = make_session(user_id=_TEST_USER_ID)
+        block_id = "a1b2c3d4-e5f6-4a7b-8c9d-0e1f2a3b4c5d"
+        block = make_mock_block(
+            block_id, "Disabled Block", BlockType.STANDARD, disabled=True
+        )
+
+        with patch(
+            "backend.copilot.tools.find_block.get_block",
+            return_value=block,
+        ):
+            tool = FindBlockTool()
+            response = await tool._execute(
+                user_id=_TEST_USER_ID, session=session, query=block_id
+            )
+
+        from .models import NoResultsResponse
+
+        assert isinstance(response, NoResultsResponse)
+        assert "disabled" in response.message.lower()
+
+    @pytest.mark.asyncio(loop_scope="session")
+    async def test_uuid_lookup_excluded_block_type(self):
+        """UUID matching an excluded block type returns NoResultsResponse."""
+        session = make_session(user_id=_TEST_USER_ID)
+        block_id = "a1b2c3d4-e5f6-4a7b-8c9d-0e1f2a3b4c5d"
+        block = make_mock_block(block_id, "Input Block", BlockType.INPUT)
+
+        with patch(
+            "backend.copilot.tools.find_block.get_block",
+            return_value=block,
+        ):
+            tool = FindBlockTool()
+            response = await tool._execute(
+                user_id=_TEST_USER_ID, session=session, query=block_id
+            )
+
+        from .models import NoResultsResponse
+
+        assert isinstance(response, NoResultsResponse)
+        assert "not available" in response.message.lower()
+
+    @pytest.mark.asyncio(loop_scope="session")
+    async def test_uuid_lookup_excluded_block_id(self):
+        """UUID matching an excluded block ID returns NoResultsResponse."""
+        session = make_session(user_id=_TEST_USER_ID)
+        smart_decision_id = "3b191d9f-356f-482d-8238-ba04b6d18381"
+        block = make_mock_block(
+            smart_decision_id, "Smart Decision Maker", BlockType.STANDARD
+        )
+
+        with patch(
+            "backend.copilot.tools.find_block.get_block",
+            return_value=block,
+        ):
+            tool = FindBlockTool()
+            response = await tool._execute(
+                user_id=_TEST_USER_ID, session=session, query=smart_decision_id
+            )
+
+        from .models import NoResultsResponse
+
+        assert isinstance(response, NoResultsResponse)
+        assert "not available" in response.message.lower()

autogpt_platform/backend/backend/copilot/tools/helpers.py

@@ -13,6 +13,7 @@ from backend.data.execution import ExecutionContext
 from backend.data.model import CredentialsFieldInfo, CredentialsMetaInput
 from backend.integrations.creds_manager import IntegrationCredentialsManager
 from backend.util.exceptions import BlockError
+from backend.util.type import coerce_inputs_to_schema
 
 from .models import BlockOutputResponse, ErrorResponse, ToolResponseBase
 from .utils import match_credentials_to_requirements
@@ -111,6 +112,9 @@ async def execute_block(
                     session_id=session_id,
                 )
 
+        # Coerce non-matching data types to the expected input schema.
+        coerce_inputs_to_schema(input_data, block.input_schema)
+
         # Execute the block and collect outputs
         outputs: dict[str, list[Any]] = defaultdict(list)
         async for output_name, output_data in block.execute(

autogpt_platform/backend/backend/copilot/tools/helpers_test.py

@@ -0,0 +1,333 @@
+"""Tests for execute_block type coercion in helpers.py.
+
+Verifies that execute_block() coerces string input values to match the block's
+expected input types, mirroring the executor's validate_exec() logic.
+This is critical for @@agptfile: expansion, where file content is always a string
+but the block may expect structured types (e.g. list[list[str]]).
+"""
+
+from typing import Any
+from unittest.mock import AsyncMock, MagicMock, patch
+
+import pytest
+
+from backend.copilot.tools.helpers import execute_block
+from backend.copilot.tools.models import BlockOutputResponse
+
+
+def _make_block_schema(annotations: dict[str, Any]) -> MagicMock:
+    """Create a mock input_schema with model_fields matching the given annotations."""
+    schema = MagicMock()
+    # coerce_inputs_to_schema uses model_fields (Pydantic v2 API)
+    model_fields = {}
+    for name, ann in annotations.items():
+        field = MagicMock()
+        field.annotation = ann
+        model_fields[name] = field
+    schema.model_fields = model_fields
+    return schema
+
+
+def _make_block(
+    block_id: str,
+    name: str,
+    annotations: dict[str, Any],
+    outputs: dict[str, list[Any]] | None = None,
+) -> MagicMock:
+    """Create a mock block with typed annotations and a simple execute method."""
+    block = MagicMock()
+    block.id = block_id
+    block.name = name
+    block.input_schema = _make_block_schema(annotations)
+
+    captured_inputs: dict[str, Any] = {}
+
+    async def mock_execute(input_data: dict, **_kwargs: Any):
+        captured_inputs.update(input_data)
+        for output_name, values in (outputs or {"result": ["ok"]}).items():
+            for v in values:
+                yield output_name, v
+
+    block.execute = mock_execute
+    block._captured_inputs = captured_inputs
+    return block
+
+
+_TEST_SESSION_ID = "test-session-coerce"
+_TEST_USER_ID = "test-user-coerce"
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_coerce_json_string_to_nested_list():
+    """JSON string → list[list[str]] (Google Sheets CSV import case)."""
+    block = _make_block(
+        "sheets-write",
+        "Google Sheets Write",
+        {"values": list[list[str]], "spreadsheet_id": str},
+    )
+
+    mock_workspace_db = MagicMock()
+    mock_workspace_db.get_or_create_workspace = AsyncMock(
+        return_value=MagicMock(id="ws-1")
+    )
+
+    with patch(
+        "backend.copilot.tools.helpers.workspace_db",
+        return_value=mock_workspace_db,
+    ):
+        response = await execute_block(
+            block=block,
+            block_id="sheets-write",
+            input_data={
+                "values": '[["Name","Score"],["Alice","90"],["Bob","85"]]',
+                "spreadsheet_id": "abc123",
+            },
+            user_id=_TEST_USER_ID,
+            session_id=_TEST_SESSION_ID,
+            node_exec_id="exec-1",
+            matched_credentials={},
+        )
+
+    assert isinstance(response, BlockOutputResponse)
+    assert response.success is True
+    # Verify the input was coerced from string to list[list[str]]
+    assert block._captured_inputs["values"] == [
+        ["Name", "Score"],
+        ["Alice", "90"],
+        ["Bob", "85"],
+    ]
+    assert isinstance(block._captured_inputs["values"], list)
+    assert isinstance(block._captured_inputs["values"][0], list)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_coerce_json_string_to_list():
+    """JSON string → list[str]."""
+    block = _make_block(
+        "list-block",
+        "List Block",
+        {"items": list[str]},
+    )
+
+    mock_workspace_db = MagicMock()
+    mock_workspace_db.get_or_create_workspace = AsyncMock(
+        return_value=MagicMock(id="ws-1")
+    )
+
+    with patch(
+        "backend.copilot.tools.helpers.workspace_db",
+        return_value=mock_workspace_db,
+    ):
+        response = await execute_block(
+            block=block,
+            block_id="list-block",
+            input_data={"items": '["a","b","c"]'},
+            user_id=_TEST_USER_ID,
+            session_id=_TEST_SESSION_ID,
+            node_exec_id="exec-2",
+            matched_credentials={},
+        )
+
+    assert isinstance(response, BlockOutputResponse)
+    assert block._captured_inputs["items"] == ["a", "b", "c"]
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_coerce_json_string_to_dict():
+    """JSON string → dict[str, str]."""
+    block = _make_block(
+        "dict-block",
+        "Dict Block",
+        {"config": dict[str, str]},
+    )
+
+    mock_workspace_db = MagicMock()
+    mock_workspace_db.get_or_create_workspace = AsyncMock(
+        return_value=MagicMock(id="ws-1")
+    )
+
+    with patch(
+        "backend.copilot.tools.helpers.workspace_db",
+        return_value=mock_workspace_db,
+    ):
+        response = await execute_block(
+            block=block,
+            block_id="dict-block",
+            input_data={"config": '{"key": "value", "foo": "bar"}'},
+            user_id=_TEST_USER_ID,
+            session_id=_TEST_SESSION_ID,
+            node_exec_id="exec-3",
+            matched_credentials={},
+        )
+
+    assert isinstance(response, BlockOutputResponse)
+    assert block._captured_inputs["config"] == {"key": "value", "foo": "bar"}
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_no_coercion_when_type_matches():
+    """Already-correct types pass through without coercion."""
+    block = _make_block(
+        "pass-through",
+        "Pass Through",
+        {"values": list[list[str]], "name": str},
+    )
+
+    original_values = [["a", "b"], ["c", "d"]]
+    mock_workspace_db = MagicMock()
+    mock_workspace_db.get_or_create_workspace = AsyncMock(
+        return_value=MagicMock(id="ws-1")
+    )
+
+    with patch(
+        "backend.copilot.tools.helpers.workspace_db",
+        return_value=mock_workspace_db,
+    ):
+        response = await execute_block(
+            block=block,
+            block_id="pass-through",
+            input_data={"values": original_values, "name": "test"},
+            user_id=_TEST_USER_ID,
+            session_id=_TEST_SESSION_ID,
+            node_exec_id="exec-4",
+            matched_credentials={},
+        )
+
+    assert isinstance(response, BlockOutputResponse)
+    assert block._captured_inputs["values"] == original_values
+    assert block._captured_inputs["name"] == "test"
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_coerce_string_to_int():
+    """String number → int."""
+    block = _make_block(
+        "int-block",
+        "Int Block",
+        {"count": int},
+    )
+
+    mock_workspace_db = MagicMock()
+    mock_workspace_db.get_or_create_workspace = AsyncMock(
+        return_value=MagicMock(id="ws-1")
+    )
+
+    with patch(
+        "backend.copilot.tools.helpers.workspace_db",
+        return_value=mock_workspace_db,
+    ):
+        response = await execute_block(
+            block=block,
+            block_id="int-block",
+            input_data={"count": "42"},
+            user_id=_TEST_USER_ID,
+            session_id=_TEST_SESSION_ID,
+            node_exec_id="exec-5",
+            matched_credentials={},
+        )
+
+    assert isinstance(response, BlockOutputResponse)
+    assert block._captured_inputs["count"] == 42
+    assert isinstance(block._captured_inputs["count"], int)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_coerce_skips_none_values():
+    """None values are not coerced (they may be optional fields)."""
+    block = _make_block(
+        "optional-block",
+        "Optional Block",
+        {"data": list[str], "label": str},
+    )
+
+    mock_workspace_db = MagicMock()
+    mock_workspace_db.get_or_create_workspace = AsyncMock(
+        return_value=MagicMock(id="ws-1")
+    )
+
+    with patch(
+        "backend.copilot.tools.helpers.workspace_db",
+        return_value=mock_workspace_db,
+    ):
+        response = await execute_block(
+            block=block,
+            block_id="optional-block",
+            input_data={"label": "test"},
+            user_id=_TEST_USER_ID,
+            session_id=_TEST_SESSION_ID,
+            node_exec_id="exec-6",
+            matched_credentials={},
+        )
+
+    assert isinstance(response, BlockOutputResponse)
+    # 'data' was not provided, so it should not appear in captured inputs
+    assert "data" not in block._captured_inputs
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_coerce_union_type_preserves_valid_member():
+    """Union-typed fields should not be coerced when the value matches a member."""
+    block = _make_block(
+        "union-block",
+        "Union Block",
+        {"content": str | list[str]},
+    )
+
+    mock_workspace_db = MagicMock()
+    mock_workspace_db.get_or_create_workspace = AsyncMock(
+        return_value=MagicMock(id="ws-1")
+    )
+
+    with patch(
+        "backend.copilot.tools.helpers.workspace_db",
+        return_value=mock_workspace_db,
+    ):
+        response = await execute_block(
+            block=block,
+            block_id="union-block",
+            input_data={"content": ["a", "b"]},
+            user_id=_TEST_USER_ID,
+            session_id=_TEST_SESSION_ID,
+            node_exec_id="exec-7",
+            matched_credentials={},
+        )
+
+    assert isinstance(response, BlockOutputResponse)
+    # list[str] should NOT be stringified to '["a", "b"]'
+    assert block._captured_inputs["content"] == ["a", "b"]
+    assert isinstance(block._captured_inputs["content"], list)
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_coerce_inner_elements_of_generic():
+    """Inner elements of generic containers are recursively coerced."""
+    block = _make_block(
+        "inner-coerce",
+        "Inner Coerce",
+        {"values": list[str]},
+    )
+
+    mock_workspace_db = MagicMock()
+    mock_workspace_db.get_or_create_workspace = AsyncMock(
+        return_value=MagicMock(id="ws-1")
+    )
+
+    with patch(
+        "backend.copilot.tools.helpers.workspace_db",
+        return_value=mock_workspace_db,
+    ):
+        response = await execute_block(
+            block=block,
+            block_id="inner-coerce",
+            # Inner elements are ints, but target is list[str]
+            input_data={"values": [1, 2, 3]},
+            user_id=_TEST_USER_ID,
+            session_id=_TEST_SESSION_ID,
+            node_exec_id="exec-8",
+            matched_credentials={},
+        )
+
+    assert isinstance(response, BlockOutputResponse)
+    # Inner elements should be coerced from int to str
+    assert block._captured_inputs["values"] == ["1", "2", "3"]
+    assert all(isinstance(v, str) for v in block._captured_inputs["values"])

autogpt_platform/backend/backend/copilot/tools/run_mcp_tool.py

@@ -184,10 +184,12 @@ class RunMCPToolTool(BaseTool):
             if e.status_code in _AUTH_STATUS_CODES and not creds:
                 # Server requires auth and user has no stored credentials
                 return self._build_setup_requirements(server_url, session_id)
-            logger.warning("MCP HTTP error for %s: %s", server_host(server_url), e)
+            host = server_host(server_url)
+            logger.warning("MCP HTTP error for %s: status=%s", host, e.status_code)
             return ErrorResponse(
-                message=f"MCP server returned HTTP {e.status_code}: {e}",
+                message=(f"MCP request to {host} failed with HTTP {e.status_code}."),
                 session_id=session_id,
+                error=f"HTTP {e.status_code}: {str(e)[:300]}",
             )
 
         except MCPClientError as e:

autogpt_platform/backend/backend/copilot/tools/test_run_mcp_tool.py

@@ -580,6 +580,49 @@ async def test_auth_error_with_existing_creds_returns_error():
     assert "403" in response.message
 
 
+@pytest.mark.asyncio(loop_scope="session")
+async def test_http_error_returns_clean_message_with_collapsible_detail():
+    """Non-auth HTTP errors return a clean message with raw detail in the `error` field."""
+    from backend.util.request import HTTPClientError
+
+    tool = RunMCPToolTool()
+    session = make_session(_USER_ID)
+
+    with patch(
+        "backend.copilot.tools.run_mcp_tool.validate_url_host", new_callable=AsyncMock
+    ):
+        with patch(
+            "backend.copilot.tools.run_mcp_tool.auto_lookup_mcp_credential",
+            new_callable=AsyncMock,
+            return_value=None,
+        ):
+            mock_client = AsyncMock()
+            mock_client.initialize = AsyncMock(
+                side_effect=HTTPClientError(
+                    "<!doctype html><html><body>Not Found</body></html>",
+                    status_code=404,
+                )
+            )
+            with patch(
+                "backend.copilot.tools.run_mcp_tool.MCPClient",
+                return_value=mock_client,
+            ):
+                response = await tool._execute(
+                    user_id=_USER_ID,
+                    session=session,
+                    server_url=_SERVER_URL,
+                )
+
+    assert isinstance(response, ErrorResponse)
+    assert "404" in response.message
+    # Raw HTML body must NOT leak into the user-facing message
+    assert "<!doctype" not in response.message
+    # Raw detail (including original body) goes in the collapsible `error` field
+    assert response.error is not None
+    assert "404" in response.error
+    assert "<!doctype" in response.error.lower()
+
+
 @pytest.mark.asyncio(loop_scope="session")
 async def test_mcp_client_error_returns_error_response():
     """MCPClientError (protocol-level) maps to a clean ErrorResponse."""

autogpt_platform/backend/backend/copilot/tools/utils.py

@@ -1,6 +1,7 @@
 """Shared utilities for chat tools."""
 
 import logging
+import re
 from typing import Any
 
 from backend.api.features.library import model as library_model
@@ -19,6 +20,26 @@ from backend.util.exceptions import NotFoundError
 
 logger = logging.getLogger(__name__)
 
+# Shared UUID v4 pattern used by multiple tools for direct ID lookups.
+_UUID_V4_PATTERN = re.compile(
+    r"^[a-f0-9]{8}-[a-f0-9]{4}-4[a-f0-9]{3}-[89ab][a-f0-9]{3}-[a-f0-9]{12}$",
+    re.IGNORECASE,
+)
+
+
+def is_uuid(text: str) -> bool:
+    """Check if text is a valid UUID v4."""
+    return bool(_UUID_V4_PATTERN.match(text.strip()))
+
+
+# Matches "creator/slug" identifiers used in the marketplace
+_CREATOR_SLUG_PATTERN = re.compile(r"^[\w-]+/[\w-]+$")
+
+
+def is_creator_slug(text: str) -> bool:
+    """Check if text matches a 'creator/slug' marketplace identifier."""
+    return bool(_CREATOR_SLUG_PATTERN.match(text.strip()))
+
 
 async def fetch_graph_from_store_slug(
     username: str,

autogpt_platform/backend/backend/data/invited_user.py

@@ -0,0 +1,750 @@
+import asyncio
+import csv
+import io
+import logging
+import os
+import re
+import socket
+from dataclasses import dataclass
+from datetime import datetime, timezone
+from typing import Any, Literal, Optional
+from uuid import uuid4
+
+import prisma.enums
+import prisma.models
+import prisma.types
+from prisma.errors import UniqueViolationError
+from pydantic import BaseModel, EmailStr, TypeAdapter, ValidationError
+
+from backend.data.db import transaction
+from backend.data.model import User
+from backend.data.redis_client import get_redis_async
+from backend.data.tally import get_business_understanding_input_from_tally, mask_email
+from backend.data.understanding import (
+    BusinessUnderstandingInput,
+    merge_business_understanding_data,
+)
+from backend.data.user import get_user_by_email, get_user_by_id
+from backend.executor.cluster_lock import AsyncClusterLock
+from backend.util.exceptions import (
+    NotAuthorizedError,
+    NotFoundError,
+    PreconditionFailed,
+)
+from backend.util.json import SafeJson
+from backend.util.settings import Settings
+
+logger = logging.getLogger(__name__)
+_settings = Settings()
+
+_WORKER_ID = f"{socket.gethostname()}:{os.getpid()}"
+
+_tally_seed_tasks: dict[str, asyncio.Task] = {}
+_TALLY_STALE_SECONDS = 300
+_MAX_TALLY_ERROR_LENGTH = 200
+_email_adapter = TypeAdapter(EmailStr)
+
+MAX_BULK_INVITE_FILE_BYTES = 1024 * 1024
+MAX_BULK_INVITE_ROWS = 500
+
+
+class InvitedUserRecord(BaseModel):
+    id: str
+    email: str
+    status: prisma.enums.InvitedUserStatus
+    auth_user_id: Optional[str] = None
+    name: Optional[str] = None
+    tally_understanding: Optional[dict[str, Any]] = None
+    tally_status: prisma.enums.TallyComputationStatus
+    tally_computed_at: Optional[datetime] = None
+    tally_error: Optional[str] = None
+    created_at: datetime
+    updated_at: datetime
+
+    @classmethod
+    def from_db(cls, invited_user: "prisma.models.InvitedUser") -> "InvitedUserRecord":
+        payload = (
+            invited_user.tallyUnderstanding
+            if isinstance(invited_user.tallyUnderstanding, dict)
+            else None
+        )
+        return cls(
+            id=invited_user.id,
+            email=invited_user.email,
+            status=invited_user.status,
+            auth_user_id=invited_user.authUserId,
+            name=invited_user.name,
+            tally_understanding=payload,
+            tally_status=invited_user.tallyStatus,
+            tally_computed_at=invited_user.tallyComputedAt,
+            tally_error=invited_user.tallyError,
+            created_at=invited_user.createdAt,
+            updated_at=invited_user.updatedAt,
+        )
+
+
+class BulkInvitedUserRowResult(BaseModel):
+    row_number: int
+    email: Optional[str] = None
+    name: Optional[str] = None
+    status: Literal["CREATED", "SKIPPED", "ERROR"]
+    message: str
+    invited_user: Optional[InvitedUserRecord] = None
+
+
+class BulkInvitedUsersResult(BaseModel):
+    created_count: int
+    skipped_count: int
+    error_count: int
+    results: list[BulkInvitedUserRowResult]
+
+
+@dataclass(frozen=True)
+class _ParsedInviteRow:
+    row_number: int
+    email: str
+    name: Optional[str]
+
+
+def normalize_email(email: str) -> str:
+    return email.strip().lower()
+
+
+def _normalize_name(name: Optional[str]) -> Optional[str]:
+    if name is None:
+        return None
+    normalized = name.strip()
+    return normalized or None
+
+
+def _default_profile_name(email: str, preferred_name: Optional[str]) -> str:
+    if preferred_name:
+        return preferred_name
+    local_part = email.split("@", 1)[0].strip()
+    return local_part or "user"
+
+
+def _sanitize_username_base(email: str) -> str:
+    local_part = email.split("@", 1)[0].lower()
+    sanitized = re.sub(r"[^a-z0-9-]", "", local_part)
+    sanitized = sanitized.strip("-")
+    return sanitized[:40] or "user"
+
+
+async def _generate_unique_profile_username(email: str, tx) -> str:
+    base = _sanitize_username_base(email)
+
+    for _ in range(2):
+        candidate = f"{base}-{uuid4().hex[:6]}"
+        existing = await prisma.models.Profile.prisma(tx).find_unique(
+            where={"username": candidate}
+        )
+        if existing is None:
+            return candidate
+
+    raise RuntimeError(f"Unable to generate unique username for {email}")
+
+
+async def _ensure_default_profile(
+    user_id: str,
+    email: str,
+    preferred_name: Optional[str],
+    tx,
+) -> None:
+    existing_profile = await prisma.models.Profile.prisma(tx).find_unique(
+        where={"userId": user_id}
+    )
+    if existing_profile is not None:
+        return
+
+    username = await _generate_unique_profile_username(email, tx)
+    await prisma.models.Profile.prisma(tx).create(
+        data=prisma.types.ProfileCreateInput(
+            userId=user_id,
+            name=_default_profile_name(email, preferred_name),
+            username=username,
+            description="I'm new here",
+            links=[],
+            avatarUrl="",
+        )
+    )
+
+
+async def _ensure_default_onboarding(user_id: str, tx) -> None:
+    await prisma.models.UserOnboarding.prisma(tx).upsert(
+        where={"userId": user_id},
+        data={
+            "create": prisma.types.UserOnboardingCreateInput(userId=user_id),
+            "update": {},
+        },
+    )
+
+
+async def _apply_tally_understanding(
+    user_id: str,
+    invited_user: "prisma.models.InvitedUser",
+    tx,
+) -> None:
+    if not isinstance(invited_user.tallyUnderstanding, dict):
+        return
+
+    try:
+        input_data = BusinessUnderstandingInput.model_validate(
+            invited_user.tallyUnderstanding
+        )
+    except Exception:
+        logger.warning(
+            "Malformed tallyUnderstanding for invited user %s; skipping",
+            invited_user.id,
+            exc_info=True,
+        )
+        return
+
+    payload = merge_business_understanding_data({}, input_data)
+    await prisma.models.CoPilotUnderstanding.prisma(tx).upsert(
+        where={"userId": user_id},
+        data={
+            "create": {"userId": user_id, "data": SafeJson(payload)},
+            "update": {"data": SafeJson(payload)},
+        },
+    )
+
+
+async def list_invited_users(
+    page: int = 1,
+    page_size: int = 50,
+) -> tuple[list[InvitedUserRecord], int]:
+    total = await prisma.models.InvitedUser.prisma().count()
+    invited_users = await prisma.models.InvitedUser.prisma().find_many(
+        order={"createdAt": "desc"},
+        skip=(page - 1) * page_size,
+        take=page_size,
+    )
+    return [InvitedUserRecord.from_db(iu) for iu in invited_users], total
+
+
+async def create_invited_user(
+    email: str, name: Optional[str] = None
+) -> InvitedUserRecord:
+    normalized_email = normalize_email(email)
+    normalized_name = _normalize_name(name)
+
+    existing_user = await prisma.models.User.prisma().find_unique(
+        where={"email": normalized_email}
+    )
+    if existing_user is not None:
+        raise PreconditionFailed("An active user with this email already exists")
+
+    existing_invited_user = await prisma.models.InvitedUser.prisma().find_unique(
+        where={"email": normalized_email}
+    )
+    if existing_invited_user is not None:
+        raise PreconditionFailed("An invited user with this email already exists")
+
+    try:
+        invited_user = await prisma.models.InvitedUser.prisma().create(
+            data={
+                "email": normalized_email,
+                "name": normalized_name,
+                "status": prisma.enums.InvitedUserStatus.INVITED,
+                "tallyStatus": prisma.enums.TallyComputationStatus.PENDING,
+            }
+        )
+    except UniqueViolationError:
+        raise PreconditionFailed("An invited user with this email already exists")
+    schedule_invited_user_tally_precompute(invited_user.id)
+    return InvitedUserRecord.from_db(invited_user)
+
+
+async def revoke_invited_user(invited_user_id: str) -> InvitedUserRecord:
+    invited_user = await prisma.models.InvitedUser.prisma().find_unique(
+        where={"id": invited_user_id}
+    )
+    if invited_user is None:
+        raise NotFoundError(f"Invited user {invited_user_id} not found")
+
+    if invited_user.status == prisma.enums.InvitedUserStatus.CLAIMED:
+        raise PreconditionFailed("Claimed invited users cannot be revoked")
+
+    if invited_user.status == prisma.enums.InvitedUserStatus.REVOKED:
+        return InvitedUserRecord.from_db(invited_user)
+
+    revoked_user = await prisma.models.InvitedUser.prisma().update(
+        where={"id": invited_user_id},
+        data={"status": prisma.enums.InvitedUserStatus.REVOKED},
+    )
+    if revoked_user is None:
+        raise NotFoundError(f"Invited user {invited_user_id} not found")
+    return InvitedUserRecord.from_db(revoked_user)
+
+
+async def retry_invited_user_tally(invited_user_id: str) -> InvitedUserRecord:
+    invited_user = await prisma.models.InvitedUser.prisma().find_unique(
+        where={"id": invited_user_id}
+    )
+    if invited_user is None:
+        raise NotFoundError(f"Invited user {invited_user_id} not found")
+
+    if invited_user.status == prisma.enums.InvitedUserStatus.REVOKED:
+        raise PreconditionFailed("Revoked invited users cannot retry Tally seeding")
+
+    refreshed_user = await prisma.models.InvitedUser.prisma().update(
+        where={"id": invited_user_id},
+        data={
+            "tallyUnderstanding": None,
+            "tallyStatus": prisma.enums.TallyComputationStatus.PENDING,
+            "tallyComputedAt": None,
+            "tallyError": None,
+        },
+    )
+    if refreshed_user is None:
+        raise NotFoundError(f"Invited user {invited_user_id} not found")
+    schedule_invited_user_tally_precompute(invited_user_id)
+    return InvitedUserRecord.from_db(refreshed_user)
+
+
+def _decode_bulk_invite_file(content: bytes) -> str:
+    if len(content) > MAX_BULK_INVITE_FILE_BYTES:
+        raise ValueError("Invite file exceeds the maximum size of 1 MB")
+
+    try:
+        return content.decode("utf-8-sig")
+    except UnicodeDecodeError as exc:
+        raise ValueError("Invite file must be UTF-8 encoded") from exc
+
+
+def _parse_bulk_invite_csv(text: str) -> list[_ParsedInviteRow]:
+    indexed_rows: list[tuple[int, list[str]]] = []
+
+    for row_number, row in enumerate(csv.reader(io.StringIO(text)), start=1):
+        normalized_row = [cell.strip() for cell in row]
+        if any(normalized_row):
+            indexed_rows.append((row_number, normalized_row))
+
+    if not indexed_rows:
+        return []
+
+    header = [cell.lower() for cell in indexed_rows[0][1]]
+    has_header = "email" in header
+    email_index = header.index("email") if has_header else 0
+    name_index: Optional[int] = (
+        header.index("name")
+        if has_header and "name" in header
+        else (1 if not has_header else None)
+    )
+    data_rows = indexed_rows[1:] if has_header else indexed_rows
+
+    parsed_rows: list[_ParsedInviteRow] = []
+    for row_number, row in data_rows:
+        if len(parsed_rows) >= MAX_BULK_INVITE_ROWS:
+            break
+        email = row[email_index].strip() if len(row) > email_index else ""
+        name = (
+            row[name_index].strip()
+            if name_index is not None and len(row) > name_index
+            else ""
+        )
+        parsed_rows.append(
+            _ParsedInviteRow(
+                row_number=row_number,
+                email=email,
+                name=name or None,
+            )
+        )
+
+    return parsed_rows
+
+
+def _parse_bulk_invite_text(text: str) -> list[_ParsedInviteRow]:
+    parsed_rows: list[_ParsedInviteRow] = []
+
+    for row_number, raw_line in enumerate(text.splitlines(), start=1):
+        if len(parsed_rows) >= MAX_BULK_INVITE_ROWS:
+            break
+        line = raw_line.strip()
+        if not line or line.startswith("#"):
+            continue
+
+        parsed_rows.append(
+            _ParsedInviteRow(
+                row_number=row_number,
+                email=line,
+                name=None,
+            )
+        )
+
+    return parsed_rows
+
+
+def _parse_bulk_invite_file(
+    filename: Optional[str],
+    content: bytes,
+) -> list[_ParsedInviteRow]:
+    text = _decode_bulk_invite_file(content)
+    file_name = filename.lower() if filename else ""
+    parsed_rows = (
+        _parse_bulk_invite_csv(text)
+        if file_name.endswith(".csv")
+        else _parse_bulk_invite_text(text)
+    )
+
+    if not parsed_rows:
+        raise ValueError("Invite file did not contain any emails")
+
+    return parsed_rows
+
+
+async def bulk_create_invited_users_from_file(
+    filename: Optional[str],
+    content: bytes,
+) -> BulkInvitedUsersResult:
+    parsed_rows = _parse_bulk_invite_file(filename, content)
+
+    created_count = 0
+    skipped_count = 0
+    error_count = 0
+    results: list[BulkInvitedUserRowResult] = []
+    seen_emails: set[str] = set()
+
+    for row in parsed_rows:
+        row_name = _normalize_name(row.name)
+
+        try:
+            validated_email = _email_adapter.validate_python(row.email)
+        except ValidationError:
+            error_count += 1
+            results.append(
+                BulkInvitedUserRowResult(
+                    row_number=row.row_number,
+                    email=row.email or None,
+                    name=row_name,
+                    status="ERROR",
+                    message="Invalid email address",
+                )
+            )
+            continue
+
+        normalized_email = normalize_email(str(validated_email))
+        if normalized_email in seen_emails:
+            skipped_count += 1
+            results.append(
+                BulkInvitedUserRowResult(
+                    row_number=row.row_number,
+                    email=normalized_email,
+                    name=row_name,
+                    status="SKIPPED",
+                    message="Duplicate email in upload file",
+                )
+            )
+            continue
+
+        seen_emails.add(normalized_email)
+
+        try:
+            invited_user = await create_invited_user(normalized_email, row_name)
+        except PreconditionFailed as exc:
+            skipped_count += 1
+            results.append(
+                BulkInvitedUserRowResult(
+                    row_number=row.row_number,
+                    email=normalized_email,
+                    name=row_name,
+                    status="SKIPPED",
+                    message=str(exc),
+                )
+            )
+        except Exception:
+            masked = mask_email(normalized_email)
+            logger.exception(
+                "Failed to create bulk invite for row %s (%s)",
+                row.row_number,
+                masked,
+            )
+            error_count += 1
+            results.append(
+                BulkInvitedUserRowResult(
+                    row_number=row.row_number,
+                    email=normalized_email,
+                    name=row_name,
+                    status="ERROR",
+                    message="Unexpected error creating invite",
+                )
+            )
+        else:
+            created_count += 1
+            results.append(
+                BulkInvitedUserRowResult(
+                    row_number=row.row_number,
+                    email=normalized_email,
+                    name=row_name,
+                    status="CREATED",
+                    message="Invite created",
+                    invited_user=invited_user,
+                )
+            )
+
+    return BulkInvitedUsersResult(
+        created_count=created_count,
+        skipped_count=skipped_count,
+        error_count=error_count,
+        results=results,
+    )
+
+
+async def _compute_invited_user_tally_seed(invited_user_id: str) -> None:
+    invited_user = await prisma.models.InvitedUser.prisma().find_unique(
+        where={"id": invited_user_id}
+    )
+    if invited_user is None:
+        return
+
+    if invited_user.status == prisma.enums.InvitedUserStatus.REVOKED:
+        return
+
+    try:
+        r = await get_redis_async()
+    except Exception:
+        r = None
+
+    lock: AsyncClusterLock | None = None
+
+    if r is not None:
+        lock = AsyncClusterLock(
+            redis=r,
+            key=f"tally_seed:{invited_user_id}",
+            owner_id=_WORKER_ID,
+            timeout=_TALLY_STALE_SECONDS,
+        )
+        current_owner = await lock.try_acquire()
+
+        if current_owner is None:
+            logger.warn("Redis unvailable for tally lock - skipping tally enrichement")
+            return
+        elif current_owner != _WORKER_ID:
+            logger.debug(
+                "Tally seed for %s already locked by %s, skipping",
+                invited_user_id,
+                current_owner,
+            )
+            return
+    if (
+        invited_user.tallyStatus == prisma.enums.TallyComputationStatus.RUNNING
+        and invited_user.updatedAt is not None
+    ):
+        age = (datetime.now(timezone.utc) - invited_user.updatedAt).total_seconds()
+        if age < _TALLY_STALE_SECONDS:
+            logger.debug(
+                "Tally task for %s still RUNNING (age=%ds), skipping",
+                invited_user_id,
+                int(age),
+            )
+            return
+        logger.info(
+            "Tally task for %s is stale (age=%ds), re-running",
+            invited_user_id,
+            int(age),
+        )
+
+    await prisma.models.InvitedUser.prisma().update(
+        where={"id": invited_user_id},
+        data={
+            "tallyStatus": prisma.enums.TallyComputationStatus.RUNNING,
+            "tallyError": None,
+        },
+    )
+
+    try:
+        input_data = await get_business_understanding_input_from_tally(
+            invited_user.email,
+            require_api_key=True,
+        )
+        payload = (
+            SafeJson(input_data.model_dump(exclude_none=True))
+            if input_data is not None
+            else None
+        )
+        await prisma.models.InvitedUser.prisma().update(
+            where={"id": invited_user_id},
+            data={
+                "tallyUnderstanding": payload,
+                "tallyStatus": prisma.enums.TallyComputationStatus.READY,
+                "tallyComputedAt": datetime.now(timezone.utc),
+                "tallyError": None,
+            },
+        )
+    except Exception as exc:
+        logger.exception(
+            "Failed to compute Tally understanding for invited user %s",
+            invited_user_id,
+        )
+        sanitized_error = re.sub(
+            r"https?://\S+", "<url>", f"{type(exc).__name__}: {exc}"
+        )[:_MAX_TALLY_ERROR_LENGTH]
+        await prisma.models.InvitedUser.prisma().update(
+            where={"id": invited_user_id},
+            data={
+                "tallyStatus": prisma.enums.TallyComputationStatus.FAILED,
+                "tallyError": sanitized_error,
+            },
+        )
+
+
+def schedule_invited_user_tally_precompute(invited_user_id: str) -> None:
+    existing = _tally_seed_tasks.get(invited_user_id)
+    if existing is not None and not existing.done():
+        logger.debug("Tally task already running for %s, skipping", invited_user_id)
+        return
+
+    task = asyncio.create_task(_compute_invited_user_tally_seed(invited_user_id))
+    _tally_seed_tasks[invited_user_id] = task
+
+    def _on_done(t: asyncio.Task, _id: str = invited_user_id) -> None:
+        if _tally_seed_tasks.get(_id) is t:
+            del _tally_seed_tasks[_id]
+
+    task.add_done_callback(_on_done)
+
+
+async def _open_signup_create_user(
+    auth_user_id: str,
+    normalized_email: str,
+    metadata_name: Optional[str],
+) -> User:
+    """Create a user without requiring an invite (open signup mode)."""
+    preferred_name = _normalize_name(metadata_name)
+    try:
+        async with transaction() as tx:
+            user = await prisma.models.User.prisma(tx).create(
+                data=prisma.types.UserCreateInput(
+                    id=auth_user_id,
+                    email=normalized_email,
+                    name=preferred_name,
+                )
+            )
+            await _ensure_default_profile(
+                auth_user_id, normalized_email, preferred_name, tx
+            )
+            await _ensure_default_onboarding(auth_user_id, tx)
+    except UniqueViolationError:
+        existing = await prisma.models.User.prisma().find_unique(
+            where={"id": auth_user_id}
+        )
+        if existing is not None:
+            return User.from_db(existing)
+        raise
+
+    return User.from_db(user)
+
+
+# TODO: We need to change this functions logic before going live
+async def get_or_activate_user(user_data: dict) -> User:
+    auth_user_id = user_data.get("sub")
+    if not auth_user_id:
+        raise NotAuthorizedError("User ID not found in token")
+
+    auth_email = user_data.get("email")
+    if not auth_email:
+        raise NotAuthorizedError("Email not found in token")
+
+    normalized_email = normalize_email(auth_email)
+    user_metadata = user_data.get("user_metadata")
+    metadata_name = (
+        user_metadata.get("name") if isinstance(user_metadata, dict) else None
+    )
+
+    existing_user = None
+    try:
+        existing_user = await get_user_by_id(auth_user_id)
+    except ValueError:
+        existing_user = None
+    except Exception:
+        logger.exception("Error on get user by id during tally enrichment process")
+        raise
+
+    if existing_user is not None:
+        return existing_user
+
+    if not _settings.config.enable_invite_gate or normalized_email.endswith("@agpt.co"):
+        return await _open_signup_create_user(
+            auth_user_id, normalized_email, metadata_name
+        )
+
+    invited_user = await prisma.models.InvitedUser.prisma().find_unique(
+        where={"email": normalized_email}
+    )
+    if invited_user is None:
+        raise NotAuthorizedError("Your email is not allowed to access the platform")
+
+    if invited_user.status != prisma.enums.InvitedUserStatus.INVITED:
+        raise NotAuthorizedError("Your invitation is no longer active")
+
+    try:
+        async with transaction() as tx:
+            current_user = await prisma.models.User.prisma(tx).find_unique(
+                where={"id": auth_user_id}
+            )
+            if current_user is not None:
+                return User.from_db(current_user)
+
+            current_invited_user = await prisma.models.InvitedUser.prisma(
+                tx
+            ).find_unique(where={"email": normalized_email})
+            if current_invited_user is None:
+                raise NotAuthorizedError(
+                    "Your email is not allowed to access the platform"
+                )
+
+            if current_invited_user.status != prisma.enums.InvitedUserStatus.INVITED:
+                raise NotAuthorizedError("Your invitation is no longer active")
+
+            if current_invited_user.authUserId not in (None, auth_user_id):
+                raise NotAuthorizedError("Your invitation has already been claimed")
+
+            preferred_name = current_invited_user.name or _normalize_name(metadata_name)
+            await prisma.models.User.prisma(tx).create(
+                data=prisma.types.UserCreateInput(
+                    id=auth_user_id,
+                    email=normalized_email,
+                    name=preferred_name,
+                )
+            )
+
+            await prisma.models.InvitedUser.prisma(tx).update(
+                where={"id": current_invited_user.id},
+                data={
+                    "status": prisma.enums.InvitedUserStatus.CLAIMED,
+                    "authUserId": auth_user_id,
+                },
+            )
+
+            await _ensure_default_profile(
+                auth_user_id,
+                normalized_email,
+                preferred_name,
+                tx,
+            )
+            await _ensure_default_onboarding(auth_user_id, tx)
+            await _apply_tally_understanding(auth_user_id, current_invited_user, tx)
+    except UniqueViolationError:
+        logger.info("Concurrent activation for user %s; re-fetching", auth_user_id)
+        already_created = await prisma.models.User.prisma().find_unique(
+            where={"id": auth_user_id}
+        )
+        if already_created is not None:
+            return User.from_db(already_created)
+        raise RuntimeError(
+            f"UniqueViolationError during activation but user {auth_user_id} not found"
+        )
+
+    get_user_by_id.cache_delete(auth_user_id)
+    get_user_by_email.cache_delete(normalized_email)
+
+    activated_user = await prisma.models.User.prisma().find_unique(
+        where={"id": auth_user_id}
+    )
+    if activated_user is None:
+        raise RuntimeError(
+            f"Activated user {auth_user_id} was not found after creation"
+        )
+
+    return User.from_db(activated_user)

autogpt_platform/backend/backend/data/invited_user_test.py

@@ -0,0 +1,335 @@
+from contextlib import asynccontextmanager
+from datetime import datetime, timezone
+from types import SimpleNamespace
+from typing import cast
+from unittest.mock import AsyncMock, Mock
+
+import prisma.enums
+import prisma.models
+import pytest
+import pytest_mock
+
+from backend.util.exceptions import NotAuthorizedError, PreconditionFailed
+
+from .invited_user import (
+    InvitedUserRecord,
+    bulk_create_invited_users_from_file,
+    create_invited_user,
+    get_or_activate_user,
+    retry_invited_user_tally,
+)
+
+
+def _invited_user_db_record(
+    *,
+    status: prisma.enums.InvitedUserStatus = prisma.enums.InvitedUserStatus.INVITED,
+    tally_understanding: dict | None = None,
+):
+    now = datetime.now(timezone.utc)
+    return SimpleNamespace(
+        id="invite-1",
+        email="invited@example.com",
+        status=status,
+        authUserId=None,
+        name="Invited User",
+        tallyUnderstanding=tally_understanding,
+        tallyStatus=prisma.enums.TallyComputationStatus.PENDING,
+        tallyComputedAt=None,
+        tallyError=None,
+        createdAt=now,
+        updatedAt=now,
+    )
+
+
+def _invited_user_record(
+    *,
+    status: prisma.enums.InvitedUserStatus = prisma.enums.InvitedUserStatus.INVITED,
+    tally_understanding: dict | None = None,
+):
+    return InvitedUserRecord.from_db(
+        cast(
+            prisma.models.InvitedUser,
+            _invited_user_db_record(
+                status=status,
+                tally_understanding=tally_understanding,
+            ),
+        )
+    )
+
+
+def _user_db_record():
+    now = datetime.now(timezone.utc)
+    return SimpleNamespace(
+        id="auth-user-1",
+        email="invited@example.com",
+        emailVerified=True,
+        name="Invited User",
+        createdAt=now,
+        updatedAt=now,
+        metadata={},
+        integrations="",
+        stripeCustomerId=None,
+        topUpConfig=None,
+        maxEmailsPerDay=3,
+        notifyOnAgentRun=True,
+        notifyOnZeroBalance=True,
+        notifyOnLowBalance=True,
+        notifyOnBlockExecutionFailed=True,
+        notifyOnContinuousAgentError=True,
+        notifyOnDailySummary=True,
+        notifyOnWeeklySummary=True,
+        notifyOnMonthlySummary=True,
+        notifyOnAgentApproved=True,
+        notifyOnAgentRejected=True,
+        timezone="not-set",
+    )
+
+
+@pytest.mark.asyncio
+async def test_create_invited_user_rejects_existing_active_user(
+    mocker: pytest_mock.MockerFixture,
+) -> None:
+    user_repo = Mock()
+    user_repo.find_unique = AsyncMock(return_value=_user_db_record())
+    invited_user_repo = Mock()
+    invited_user_repo.find_unique = AsyncMock()
+
+    mocker.patch(
+        "backend.data.invited_user.prisma.models.User.prisma", return_value=user_repo
+    )
+    mocker.patch(
+        "backend.data.invited_user.prisma.models.InvitedUser.prisma",
+        return_value=invited_user_repo,
+    )
+
+    with pytest.raises(PreconditionFailed):
+        await create_invited_user("Invited@example.com")
+
+
+@pytest.mark.asyncio
+async def test_create_invited_user_schedules_tally_seed(
+    mocker: pytest_mock.MockerFixture,
+) -> None:
+    user_repo = Mock()
+    user_repo.find_unique = AsyncMock(return_value=None)
+    invited_user_repo = Mock()
+    invited_user_repo.find_unique = AsyncMock(return_value=None)
+    invited_user_repo.create = AsyncMock(return_value=_invited_user_db_record())
+    schedule = mocker.patch(
+        "backend.data.invited_user.schedule_invited_user_tally_precompute"
+    )
+
+    mocker.patch(
+        "backend.data.invited_user.prisma.models.User.prisma", return_value=user_repo
+    )
+    mocker.patch(
+        "backend.data.invited_user.prisma.models.InvitedUser.prisma",
+        return_value=invited_user_repo,
+    )
+
+    invited_user = await create_invited_user("Invited@example.com", "Invited User")
+
+    assert invited_user.email == "invited@example.com"
+    invited_user_repo.create.assert_awaited_once()
+    schedule.assert_called_once_with("invite-1")
+
+
+@pytest.mark.asyncio
+async def test_retry_invited_user_tally_resets_state_and_schedules(
+    mocker: pytest_mock.MockerFixture,
+) -> None:
+    invited_user_repo = Mock()
+    invited_user_repo.find_unique = AsyncMock(return_value=_invited_user_db_record())
+    invited_user_repo.update = AsyncMock(return_value=_invited_user_db_record())
+    schedule = mocker.patch(
+        "backend.data.invited_user.schedule_invited_user_tally_precompute"
+    )
+
+    mocker.patch(
+        "backend.data.invited_user.prisma.models.InvitedUser.prisma",
+        return_value=invited_user_repo,
+    )
+
+    invited_user = await retry_invited_user_tally("invite-1")
+
+    assert invited_user.id == "invite-1"
+    invited_user_repo.update.assert_awaited_once()
+    schedule.assert_called_once_with("invite-1")
+
+
+@pytest.mark.asyncio
+async def test_get_or_activate_user_requires_invite(
+    mocker: pytest_mock.MockerFixture,
+) -> None:
+    invited_user_repo = Mock()
+    invited_user_repo.find_unique = AsyncMock(return_value=None)
+
+    mock_get_user_by_id = AsyncMock(side_effect=ValueError("User not found"))
+    mock_get_user_by_id.cache_delete = Mock()
+    mocker.patch(
+        "backend.data.invited_user.get_user_by_id",
+        mock_get_user_by_id,
+    )
+    mocker.patch(
+        "backend.data.invited_user._settings.config.enable_invite_gate",
+        True,
+    )
+    mocker.patch(
+        "backend.data.invited_user.prisma.models.InvitedUser.prisma",
+        return_value=invited_user_repo,
+    )
+
+    with pytest.raises(NotAuthorizedError):
+        await get_or_activate_user(
+            {"sub": "auth-user-1", "email": "invited@example.com"}
+        )
+
+
+@pytest.mark.asyncio
+async def test_get_or_activate_user_creates_user_from_invite(
+    mocker: pytest_mock.MockerFixture,
+) -> None:
+    tx = object()
+    invited_user = _invited_user_db_record(
+        tally_understanding={"user_name": "Invited User", "industry": "Automation"}
+    )
+    created_user = _user_db_record()
+
+    outside_user_repo = Mock()
+    # Only called once at post-transaction verification (line 741);
+    # get_user_by_id (line 657) uses prisma.user.find_unique, not this mock.
+    outside_user_repo.find_unique = AsyncMock(return_value=created_user)
+
+    inside_user_repo = Mock()
+    inside_user_repo.find_unique = AsyncMock(return_value=None)
+    inside_user_repo.create = AsyncMock(return_value=created_user)
+
+    outside_invited_repo = Mock()
+    outside_invited_repo.find_unique = AsyncMock(return_value=invited_user)
+
+    inside_invited_repo = Mock()
+    inside_invited_repo.find_unique = AsyncMock(return_value=invited_user)
+    inside_invited_repo.update = AsyncMock(return_value=invited_user)
+
+    def user_prisma(client=None):
+        return inside_user_repo if client is tx else outside_user_repo
+
+    def invited_user_prisma(client=None):
+        return inside_invited_repo if client is tx else outside_invited_repo
+
+    @asynccontextmanager
+    async def fake_transaction():
+        yield tx
+
+    # Mock get_user_by_id since it uses prisma.user.find_unique (global client),
+    # not prisma.models.User.prisma().find_unique which we mock above.
+    mock_get_user_by_id = AsyncMock(side_effect=ValueError("User not found"))
+    mock_get_user_by_id.cache_delete = Mock()
+    mocker.patch(
+        "backend.data.invited_user.get_user_by_id",
+        mock_get_user_by_id,
+    )
+    mock_get_user_by_email = AsyncMock()
+    mock_get_user_by_email.cache_delete = Mock()
+    mocker.patch(
+        "backend.data.invited_user.get_user_by_email",
+        mock_get_user_by_email,
+    )
+    ensure_profile = mocker.patch(
+        "backend.data.invited_user._ensure_default_profile",
+        AsyncMock(),
+    )
+    ensure_onboarding = mocker.patch(
+        "backend.data.invited_user._ensure_default_onboarding",
+        AsyncMock(),
+    )
+    apply_tally = mocker.patch(
+        "backend.data.invited_user._apply_tally_understanding",
+        AsyncMock(),
+    )
+    mocker.patch("backend.data.invited_user.transaction", fake_transaction)
+    mocker.patch(
+        "backend.data.invited_user.prisma.models.User.prisma", side_effect=user_prisma
+    )
+    mocker.patch(
+        "backend.data.invited_user.prisma.models.InvitedUser.prisma",
+        side_effect=invited_user_prisma,
+    )
+
+    user = await get_or_activate_user(
+        {
+            "sub": "auth-user-1",
+            "email": "Invited@example.com",
+            "user_metadata": {"name": "Invited User"},
+        }
+    )
+
+    assert user.id == "auth-user-1"
+    inside_user_repo.create.assert_awaited_once()
+    inside_invited_repo.update.assert_awaited_once()
+    ensure_profile.assert_awaited_once()
+    ensure_onboarding.assert_awaited_once_with("auth-user-1", tx)
+    apply_tally.assert_awaited_once_with("auth-user-1", invited_user, tx)
+
+
+@pytest.mark.asyncio
+async def test_bulk_create_invited_users_from_text_file(
+    mocker: pytest_mock.MockerFixture,
+) -> None:
+    create_invited = mocker.patch(
+        "backend.data.invited_user.create_invited_user",
+        AsyncMock(
+            side_effect=[
+                _invited_user_record(),
+                _invited_user_record(),
+            ]
+        ),
+    )
+
+    result = await bulk_create_invited_users_from_file(
+        "invites.txt",
+        b"Invited@example.com\nsecond@example.com\n",
+    )
+
+    assert result.created_count == 2
+    assert result.skipped_count == 0
+    assert result.error_count == 0
+    assert [row.status for row in result.results] == ["CREATED", "CREATED"]
+    assert create_invited.await_count == 2
+
+
+@pytest.mark.asyncio
+async def test_bulk_create_invited_users_handles_csv_duplicates_and_invalid_rows(
+    mocker: pytest_mock.MockerFixture,
+) -> None:
+    create_invited = mocker.patch(
+        "backend.data.invited_user.create_invited_user",
+        AsyncMock(
+            side_effect=[
+                _invited_user_record(),
+                PreconditionFailed("An invited user with this email already exists"),
+            ]
+        ),
+    )
+
+    result = await bulk_create_invited_users_from_file(
+        "invites.csv",
+        (
+            "email,name\n"
+            "valid@example.com,Valid User\n"
+            "not-an-email,Bad Row\n"
+            "valid@example.com,Duplicate In File\n"
+            "existing@example.com,Existing User\n"
+        ).encode("utf-8"),
+    )
+
+    assert result.created_count == 1
+    assert result.skipped_count == 2
+    assert result.error_count == 1
+    assert [row.status for row in result.results] == [
+        "CREATED",
+        "ERROR",
+        "SKIPPED",
+        "SKIPPED",
+    ]
+    assert create_invited.await_count == 2

autogpt_platform/backend/backend/data/tally.py

@@ -41,7 +41,7 @@ _MAX_PAGES = 100
 _LLM_TIMEOUT = 30
 
 
-def _mask_email(email: str) -> str:
+def mask_email(email: str) -> str:
     """Mask an email for safe logging: 'alice@example.com' -> 'a***e@example.com'."""
     try:
         local, domain = email.rsplit("@", 1)
@@ -196,8 +196,7 @@ async def _refresh_cache(form_id: str) -> tuple[dict, list]:
 
     Returns (email_index, questions).
     """
-    settings = Settings()
-    client = _make_tally_client(settings.secrets.tally_api_key)
+    client = _make_tally_client(_settings.secrets.tally_api_key)
 
     redis = await get_redis_async()
     last_fetch_key = _LAST_FETCH_KEY.format(form_id=form_id)
@@ -332,6 +331,9 @@ Fields:
 - current_software (list of strings): software/tools currently used
 - existing_automation (list of strings): existing automations
 - additional_notes (string): any additional context
+- suggested_prompts (list of 5 strings): short action prompts (each under 20 words) that would help \
+this person get started with automating their work. Should be specific to their industry, role, and \
+pain points; actionable and conversational in tone; focused on automation opportunities.
 
 Form data:
 """
@@ -339,21 +341,21 @@ Form data:
 _EXTRACTION_SUFFIX = "\n\nReturn ONLY valid JSON."
 
 
-async def extract_business_understanding(
+async def extract_business_understanding_from_tally(
     formatted_text: str,
 ) -> BusinessUnderstandingInput:
-    """Use an LLM to extract structured business understanding from form text.
+    """
+    Use an LLM to extract structured business understanding from form text.
 
     Raises on timeout or unparseable response so the caller can handle it.
     """
-    settings = Settings()
-    api_key = settings.secrets.open_router_api_key
+    api_key = _settings.secrets.open_router_api_key
     client = AsyncOpenAI(api_key=api_key, base_url=OPENROUTER_BASE_URL)
 
     try:
         response = await asyncio.wait_for(
             client.chat.completions.create(
-                model="openai/gpt-4o-mini",
+                model=_settings.config.tally_extraction_llm_model,
                 messages=[
                     {
                         "role": "user",
@@ -378,9 +380,57 @@ async def extract_business_understanding(
 
     # Filter out null values before constructing
     cleaned = {k: v for k, v in data.items() if v is not None}
+
+    # Validate suggested_prompts: filter >20 words, keep top 3
+    raw_prompts = cleaned.get("suggested_prompts", [])
+    if isinstance(raw_prompts, list):
+        valid = [
+            p.strip()
+            for p in raw_prompts
+            if isinstance(p, str) and len(p.strip().split()) <= 20
+        ]
+        # This will keep up to 3 suggestions
+        short_prompts = valid[:3] if valid else None
+        if short_prompts:
+            cleaned["suggested_prompts"] = short_prompts
+        else:
+            # We dont want to add a None value suggested_prompts field
+            cleaned.pop("suggested_prompts", None)
+    else:
+        # suggested_prompts must be a list - removing it as its not here
+        cleaned.pop("suggested_prompts", None)
+
     return BusinessUnderstandingInput(**cleaned)
 
 
+async def get_business_understanding_input_from_tally(
+    email: str,
+    *,
+    require_api_key: bool = False,
+) -> Optional[BusinessUnderstandingInput]:
+    if not _settings.secrets.tally_api_key:
+        if require_api_key:
+            raise RuntimeError("Tally API key is not configured")
+        logger.debug("Tally: no API key configured, skipping")
+        return None
+
+    masked = mask_email(email)
+    result = await find_submission_by_email(TALLY_FORM_ID, email)
+    if result is None:
+        logger.debug(f"Tally: no submission found for {masked}")
+        return None
+
+    submission, questions = result
+    logger.info(f"Tally: found submission for {masked}, extracting understanding")
+
+    formatted = format_submission_for_llm(submission, questions)
+    if not formatted.strip():
+        logger.warning("Tally: formatted submission was empty, skipping")
+        return None
+
+    return await extract_business_understanding_from_tally(formatted)
+
+
 async def populate_understanding_from_tally(user_id: str, email: str) -> None:
     """Main orchestrator: check Tally for a matching submission and populate understanding.
 
@@ -395,30 +445,10 @@ async def populate_understanding_from_tally(user_id: str, email: str) -> None:
             )
             return
 
-        # Check API key is configured
-        settings = Settings()
-        if not settings.secrets.tally_api_key:
-            logger.debug("Tally: no API key configured, skipping")
+        understanding_input = await get_business_understanding_input_from_tally(email)
+        if understanding_input is None:
             return
 
-        # Look up submission by email
-        masked = _mask_email(email)
-        result = await find_submission_by_email(TALLY_FORM_ID, email)
-        if result is None:
-            logger.debug(f"Tally: no submission found for {masked}")
-            return
-
-        submission, questions = result
-        logger.info(f"Tally: found submission for {masked}, extracting understanding")
-
-        # Format and extract
-        formatted = format_submission_for_llm(submission, questions)
-        if not formatted.strip():
-            logger.warning("Tally: formatted submission was empty, skipping")
-            return
-
-        understanding_input = await extract_business_understanding(formatted)
-
         # Upsert into database
         await upsert_business_understanding(user_id, understanding_input)
         logger.info(f"Tally: successfully populated understanding for user {user_id}")

autogpt_platform/backend/backend/data/tally_test.py

@@ -12,11 +12,11 @@ from backend.data.tally import (
     _build_email_index,
     _format_answer,
     _make_tally_client,
-    _mask_email,
     _refresh_cache,
-    extract_business_understanding,
+    extract_business_understanding_from_tally,
     find_submission_by_email,
     format_submission_for_llm,
+    mask_email,
     populate_understanding_from_tally,
 )
 
@@ -248,7 +248,7 @@ async def test_populate_understanding_skips_no_api_key():
             new_callable=AsyncMock,
             return_value=None,
         ),
-        patch("backend.data.tally.Settings", return_value=mock_settings),
+        patch("backend.data.tally._settings", mock_settings),
         patch(
             "backend.data.tally.find_submission_by_email",
             new_callable=AsyncMock,
@@ -284,6 +284,7 @@ async def test_populate_understanding_full_flow():
         ],
     }
     mock_input = MagicMock()
+    mock_input.suggested_prompts = ["Prompt 1", "Prompt 2", "Prompt 3"]
 
     with (
         patch(
@@ -291,14 +292,14 @@ async def test_populate_understanding_full_flow():
             new_callable=AsyncMock,
             return_value=None,
         ),
-        patch("backend.data.tally.Settings", return_value=mock_settings),
+        patch("backend.data.tally._settings", mock_settings),
         patch(
             "backend.data.tally.find_submission_by_email",
             new_callable=AsyncMock,
             return_value=(submission, SAMPLE_QUESTIONS),
         ),
         patch(
-            "backend.data.tally.extract_business_understanding",
+            "backend.data.tally.extract_business_understanding_from_tally",
             new_callable=AsyncMock,
             return_value=mock_input,
         ) as mock_extract,
@@ -331,14 +332,14 @@ async def test_populate_understanding_handles_llm_timeout():
             new_callable=AsyncMock,
             return_value=None,
         ),
-        patch("backend.data.tally.Settings", return_value=mock_settings),
+        patch("backend.data.tally._settings", mock_settings),
         patch(
             "backend.data.tally.find_submission_by_email",
             new_callable=AsyncMock,
             return_value=(submission, SAMPLE_QUESTIONS),
         ),
         patch(
-            "backend.data.tally.extract_business_understanding",
+            "backend.data.tally.extract_business_understanding_from_tally",
             new_callable=AsyncMock,
             side_effect=asyncio.TimeoutError(),
         ),
@@ -356,13 +357,13 @@ async def test_populate_understanding_handles_llm_timeout():
 
 
 def test_mask_email():
-    assert _mask_email("alice@example.com") == "a***e@example.com"
-    assert _mask_email("ab@example.com") == "a***@example.com"
-    assert _mask_email("a@example.com") == "a***@example.com"
+    assert mask_email("alice@example.com") == "a***e@example.com"
+    assert mask_email("ab@example.com") == "a***@example.com"
+    assert mask_email("a@example.com") == "a***@example.com"
 
 
 def test_mask_email_invalid():
-    assert _mask_email("no-at-sign") == "***"
+    assert mask_email("no-at-sign") == "***"
 
 
 # ── Prompt construction (curly-brace safety) ─────────────────────────────────
@@ -393,11 +394,11 @@ def test_extraction_prompt_no_format_placeholders():
     assert single_braces == [], f"Found format placeholders: {single_braces}"
 
 
-# ── extract_business_understanding ────────────────────────────────────────────
+# ── extract_business_understanding_from_tally ────────────────────────────────────────────
 
 
 @pytest.mark.asyncio
-async def test_extract_business_understanding_success():
+async def test_extract_business_understanding_from_tally_success():
     """Happy path: LLM returns valid JSON that maps to BusinessUnderstandingInput."""
     mock_choice = MagicMock()
     mock_choice.message.content = json.dumps(
@@ -406,6 +407,13 @@ async def test_extract_business_understanding_success():
             "business_name": "Acme Corp",
             "industry": "Technology",
             "pain_points": ["manual reporting"],
+            "suggested_prompts": [
+                "Automate weekly reports",
+                "Set up invoice processing",
+                "Create a customer onboarding flow",
+                "Track project deadlines automatically",
+                "Send follow-up emails after meetings",
+            ],
         }
     )
     mock_response = MagicMock()
@@ -415,16 +423,56 @@ async def test_extract_business_understanding_success():
     mock_client.chat.completions.create.return_value = mock_response
 
     with patch("backend.data.tally.AsyncOpenAI", return_value=mock_client):
-        result = await extract_business_understanding("Q: Name?\nA: Alice")
+        result = await extract_business_understanding_from_tally("Q: Name?\nA: Alice")
 
     assert result.user_name == "Alice"
     assert result.business_name == "Acme Corp"
     assert result.industry == "Technology"
     assert result.pain_points == ["manual reporting"]
+    # suggested_prompts validated and sliced to top 3
+    assert result.suggested_prompts == [
+        "Automate weekly reports",
+        "Set up invoice processing",
+        "Create a customer onboarding flow",
+    ]
 
 
 @pytest.mark.asyncio
-async def test_extract_business_understanding_filters_nulls():
+async def test_extract_business_understanding_from_tally_filters_long_prompts():
+    """Prompts exceeding 20 words are excluded and only top 3 are kept."""
+    long_prompt = " ".join(["word"] * 21)
+    mock_choice = MagicMock()
+    mock_choice.message.content = json.dumps(
+        {
+            "user_name": "Alice",
+            "suggested_prompts": [
+                long_prompt,
+                "Short prompt one",
+                long_prompt,
+                "Short prompt two",
+                "Short prompt three",
+                "Short prompt four",
+            ],
+        }
+    )
+    mock_response = MagicMock()
+    mock_response.choices = [mock_choice]
+
+    mock_client = AsyncMock()
+    mock_client.chat.completions.create.return_value = mock_response
+
+    with patch("backend.data.tally.AsyncOpenAI", return_value=mock_client):
+        result = await extract_business_understanding_from_tally("Q: Name?\nA: Alice")
+
+    assert result.suggested_prompts == [
+        "Short prompt one",
+        "Short prompt two",
+        "Short prompt three",
+    ]
+
+
+@pytest.mark.asyncio
+async def test_extract_business_understanding_from_tally_filters_nulls():
     """Null values from LLM should be excluded from the result."""
     mock_choice = MagicMock()
     mock_choice.message.content = json.dumps(
@@ -437,7 +485,7 @@ async def test_extract_business_understanding_filters_nulls():
     mock_client.chat.completions.create.return_value = mock_response
 
     with patch("backend.data.tally.AsyncOpenAI", return_value=mock_client):
-        result = await extract_business_understanding("Q: Name?\nA: Alice")
+        result = await extract_business_understanding_from_tally("Q: Name?\nA: Alice")
 
     assert result.user_name == "Alice"
     assert result.business_name is None
@@ -445,7 +493,7 @@ async def test_extract_business_understanding_filters_nulls():
 
 
 @pytest.mark.asyncio
-async def test_extract_business_understanding_invalid_json():
+async def test_extract_business_understanding_from_tally_invalid_json():
     """Invalid JSON from LLM should raise JSONDecodeError."""
     mock_choice = MagicMock()
     mock_choice.message.content = "not valid json {"
@@ -459,11 +507,11 @@ async def test_extract_business_understanding_invalid_json():
         patch("backend.data.tally.AsyncOpenAI", return_value=mock_client),
         pytest.raises(json.JSONDecodeError),
     ):
-        await extract_business_understanding("Q: Name?\nA: Alice")
+        await extract_business_understanding_from_tally("Q: Name?\nA: Alice")
 
 
 @pytest.mark.asyncio
-async def test_extract_business_understanding_timeout():
+async def test_extract_business_understanding_from_tally_timeout():
     """LLM timeout should propagate as asyncio.TimeoutError."""
     mock_client = AsyncMock()
     mock_client.chat.completions.create.side_effect = asyncio.TimeoutError()
@@ -473,7 +521,7 @@ async def test_extract_business_understanding_timeout():
         patch("backend.data.tally._LLM_TIMEOUT", 0.001),
         pytest.raises(asyncio.TimeoutError),
     ):
-        await extract_business_understanding("Q: Name?\nA: Alice")
+        await extract_business_understanding_from_tally("Q: Name?\nA: Alice")
 
 
 # ── _refresh_cache ───────────────────────────────────────────────────────────
@@ -492,7 +540,7 @@ async def test_refresh_cache_full_fetch():
     submissions = SAMPLE_SUBMISSIONS
 
     with (
-        patch("backend.data.tally.Settings", return_value=mock_settings),
+        patch("backend.data.tally._settings", mock_settings),
         patch(
             "backend.data.tally.get_redis_async",
             new_callable=AsyncMock,
@@ -540,7 +588,7 @@ async def test_refresh_cache_incremental_fetch():
     new_submissions = [SAMPLE_SUBMISSIONS[0]]  # Just Alice
 
     with (
-        patch("backend.data.tally.Settings", return_value=mock_settings),
+        patch("backend.data.tally._settings", mock_settings),
         patch(
             "backend.data.tally.get_redis_async",
             new_callable=AsyncMock,

autogpt_platform/backend/backend/data/understanding.py

@@ -86,6 +86,11 @@ class BusinessUnderstandingInput(pydantic.BaseModel):
         None, description="Any additional context"
     )
 
+    # Suggested prompts (UI-only, not included in system prompt)
+    suggested_prompts: Optional[list[str]] = pydantic.Field(
+        None, description="LLM-generated suggested prompts based on business context"
+    )
+
 
 class BusinessUnderstanding(pydantic.BaseModel):
     """Full business understanding model returned from database."""
@@ -122,6 +127,9 @@ class BusinessUnderstanding(pydantic.BaseModel):
     # Additional context
     additional_notes: Optional[str] = None
 
+    # Suggested prompts (UI-only, not included in system prompt)
+    suggested_prompts: list[str] = pydantic.Field(default_factory=list)
+
     @classmethod
     def from_db(cls, db_record: CoPilotUnderstanding) -> "BusinessUnderstanding":
         """Convert database record to Pydantic model."""
@@ -149,6 +157,7 @@ class BusinessUnderstanding(pydantic.BaseModel):
             current_software=_json_to_list(business.get("current_software")),
             existing_automation=_json_to_list(business.get("existing_automation")),
             additional_notes=business.get("additional_notes"),
+            suggested_prompts=_json_to_list(data.get("suggested_prompts")),
         )
 
 
@@ -166,6 +175,62 @@ def _merge_lists(existing: list | None, new: list | None) -> list | None:
     return merged
 
 
+def merge_business_understanding_data(
+    existing_data: dict[str, Any],
+    input_data: BusinessUnderstandingInput,
+) -> dict[str, Any]:
+    merged_data = dict(existing_data)
+
+    merged_business: dict[str, Any] = {}
+    if isinstance(merged_data.get("business"), dict):
+        merged_business = dict(merged_data["business"])
+
+    business_string_fields = [
+        "job_title",
+        "business_name",
+        "industry",
+        "business_size",
+        "user_role",
+        "additional_notes",
+    ]
+    business_list_fields = [
+        "key_workflows",
+        "daily_activities",
+        "pain_points",
+        "bottlenecks",
+        "manual_tasks",
+        "automation_goals",
+        "current_software",
+        "existing_automation",
+    ]
+
+    if input_data.user_name is not None:
+        merged_data["name"] = input_data.user_name
+
+    for field in business_string_fields:
+        value = getattr(input_data, field)
+        if value is not None:
+            merged_business[field] = value
+
+    for field in business_list_fields:
+        value = getattr(input_data, field)
+        if value is not None:
+            existing_list = _json_to_list(merged_business.get(field))
+            merged_list = _merge_lists(existing_list, value)
+            merged_business[field] = merged_list
+
+    merged_business["version"] = 1
+    merged_data["business"] = merged_business
+
+    # suggested_prompts lives at the top level (not under `business`) because
+    # it's a UI-only artifact consumed by the frontend, not business understanding
+    # data. The `business` sub-dict feeds the system prompt.
+    if input_data.suggested_prompts is not None:
+        merged_data["suggested_prompts"] = input_data.suggested_prompts
+
+    return merged_data
+
+
 async def _get_from_cache(user_id: str) -> Optional[BusinessUnderstanding]:
     """Get business understanding from Redis cache."""
     try:
@@ -245,63 +310,18 @@ async def upsert_business_understanding(
         where={"userId": user_id}
     )
 
-    # Get existing data structure or start fresh
     existing_data: dict[str, Any] = {}
     if existing and isinstance(existing.data, dict):
         existing_data = dict(existing.data)
 
-    existing_business: dict[str, Any] = {}
-    if isinstance(existing_data.get("business"), dict):
-        existing_business = dict(existing_data["business"])
-
-    # Business fields (stored inside business object)
-    business_string_fields = [
-        "job_title",
-        "business_name",
-        "industry",
-        "business_size",
-        "user_role",
-        "additional_notes",
-    ]
-    business_list_fields = [
-        "key_workflows",
-        "daily_activities",
-        "pain_points",
-        "bottlenecks",
-        "manual_tasks",
-        "automation_goals",
-        "current_software",
-        "existing_automation",
-    ]
-
-    # Handle top-level name field
-    if input_data.user_name is not None:
-        existing_data["name"] = input_data.user_name
-
-    # Business string fields - overwrite if provided
-    for field in business_string_fields:
-        value = getattr(input_data, field)
-        if value is not None:
-            existing_business[field] = value
-
-    # Business list fields - merge with existing
-    for field in business_list_fields:
-        value = getattr(input_data, field)
-        if value is not None:
-            existing_list = _json_to_list(existing_business.get(field))
-            merged = _merge_lists(existing_list, value)
-            existing_business[field] = merged
-
-    # Set version and nest business data
-    existing_business["version"] = 1
-    existing_data["business"] = existing_business
+    merged_data = merge_business_understanding_data(existing_data, input_data)
 
     # Upsert with the merged data
     record = await CoPilotUnderstanding.prisma().upsert(
         where={"userId": user_id},
         data={
-            "create": {"userId": user_id, "data": SafeJson(existing_data)},
-            "update": {"data": SafeJson(existing_data)},
+            "create": {"userId": user_id, "data": SafeJson(merged_data)},
+            "update": {"data": SafeJson(merged_data)},
         },
     )
 

autogpt_platform/backend/backend/data/understanding_test.py

@@ -0,0 +1,102 @@
+"""Tests for business understanding merge and format logic."""
+
+from datetime import datetime, timezone
+from typing import Any
+
+from backend.data.understanding import (
+    BusinessUnderstanding,
+    BusinessUnderstandingInput,
+    format_understanding_for_prompt,
+    merge_business_understanding_data,
+)
+
+
+def _make_input(**kwargs: Any) -> BusinessUnderstandingInput:
+    """Create a BusinessUnderstandingInput with only the specified fields."""
+    return BusinessUnderstandingInput.model_validate(kwargs)
+
+
+# ─── merge_business_understanding_data: suggested_prompts ─────────────
+
+
+def test_merge_suggested_prompts_overwrites_existing():
+    """New suggested_prompts should fully replace existing ones (not append)."""
+    existing = {
+        "name": "Alice",
+        "business": {"industry": "Tech", "version": 1},
+        "suggested_prompts": ["Old prompt 1", "Old prompt 2"],
+    }
+    input_data = _make_input(
+        suggested_prompts=["New prompt A", "New prompt B", "New prompt C"],
+    )
+
+    result = merge_business_understanding_data(existing, input_data)
+
+    assert result["suggested_prompts"] == [
+        "New prompt A",
+        "New prompt B",
+        "New prompt C",
+    ]
+
+
+def test_merge_suggested_prompts_none_preserves_existing():
+    """When input has suggested_prompts=None, existing prompts are preserved."""
+    existing = {
+        "name": "Alice",
+        "business": {"industry": "Tech", "version": 1},
+        "suggested_prompts": ["Keep me"],
+    }
+    input_data = _make_input(industry="Finance")
+
+    result = merge_business_understanding_data(existing, input_data)
+
+    assert result["suggested_prompts"] == ["Keep me"]
+    assert result["business"]["industry"] == "Finance"
+
+
+def test_merge_suggested_prompts_added_to_empty_data():
+    """Suggested prompts are set at top level even when starting from empty data."""
+    existing: dict[str, Any] = {}
+    input_data = _make_input(suggested_prompts=["Prompt 1"])
+
+    result = merge_business_understanding_data(existing, input_data)
+
+    assert result["suggested_prompts"] == ["Prompt 1"]
+
+
+def test_merge_suggested_prompts_empty_list_overwrites():
+    """An explicit empty list should overwrite existing prompts."""
+    existing: dict[str, Any] = {
+        "suggested_prompts": ["Old prompt"],
+        "business": {"version": 1},
+    }
+    input_data = _make_input(suggested_prompts=[])
+
+    result = merge_business_understanding_data(existing, input_data)
+
+    assert result["suggested_prompts"] == []
+
+
+# ─── format_understanding_for_prompt: excludes suggested_prompts ──────
+
+
+def test_format_understanding_excludes_suggested_prompts():
+    """suggested_prompts is UI-only and must NOT appear in the system prompt."""
+    understanding = BusinessUnderstanding(
+        id="test-id",
+        user_id="user-1",
+        created_at=datetime.now(tz=timezone.utc),
+        updated_at=datetime.now(tz=timezone.utc),
+        user_name="Alice",
+        industry="Technology",
+        suggested_prompts=["Automate reports", "Set up alerts", "Track KPIs"],
+    )
+
+    formatted = format_understanding_for_prompt(understanding)
+
+    assert "Alice" in formatted
+    assert "Technology" in formatted
+    assert "suggested_prompts" not in formatted
+    assert "Automate reports" not in formatted
+    assert "Set up alerts" not in formatted
+    assert "Track KPIs" not in formatted

autogpt_platform/backend/backend/executor/manager.py

@@ -61,7 +61,12 @@ from backend.util.decorator import (
     error_logged,
     time_measured,
 )
-from backend.util.exceptions import InsufficientBalanceError, ModerationError
+from backend.util.exceptions import (
+    GraphNotFoundError,
+    InsufficientBalanceError,
+    ModerationError,
+    NotFoundError,
+)
 from backend.util.file import clean_exec_files
 from backend.util.logging import TruncatedLogger, configure_logging
 from backend.util.metrics import DiscordChannel
@@ -375,9 +380,16 @@ async def execute_node(
             log_metadata.debug("Node produced output", **{output_name: output_data})
             yield output_name, output_data
     except Exception as ex:
-        # Capture exception WITH context still set before restoring scope
-        sentry_sdk.capture_exception(error=ex, scope=scope)
-        sentry_sdk.flush()  # Ensure it's sent before we restore scope
+        # Only capture unexpected errors to Sentry, not user-caused ones.
+        # Most ValueError subclasses here are expected (BlockExecutionError,
+        # InsufficientBalanceError, plain ValueError for auth/disabled blocks, etc.)
+        # but NotFoundError/GraphNotFoundError could indicate real platform issues.
+        is_expected = isinstance(ex, ValueError) and not isinstance(
+            ex, (NotFoundError, GraphNotFoundError)
+        )
+        if not is_expected:
+            sentry_sdk.capture_exception(error=ex, scope=scope)
+            sentry_sdk.flush()
         # Re-raise to maintain normal error flow
         raise
     finally:
@@ -1478,7 +1490,7 @@ class ExecutionProcessor:
                     alert_message, DiscordChannel.PRODUCT
                 )
             except Exception as e:
-                logger.error(f"Failed to send low balance Discord alert: {e}")
+                logger.warning(f"Failed to send low balance Discord alert: {e}")
 
 
 class ExecutionManager(AppProcess):
@@ -1900,17 +1912,16 @@ class ExecutionManager(AppProcess):
             channel = client.get_channel()
             channel.connection.add_callback_threadsafe(lambda: channel.stop_consuming())
 
-            try:
-                thread.join(timeout=300)
-            except TimeoutError:
-                logger.error(
+            thread.join(timeout=300)
+            if thread.is_alive():
+                logger.warning(
                     f"{prefix} ⚠️ Run thread did not finish in time, forcing disconnect"
                 )
 
             client.disconnect()
             logger.info(f"{prefix} ✅ Run client disconnected")
         except Exception as e:
-            logger.error(f"{prefix} ⚠️ Error disconnecting run client: {type(e)} {e}")
+            logger.warning(f"{prefix} ⚠️ Error disconnecting run client: {type(e)} {e}")
 
     def cleanup(self):
         """Override cleanup to implement graceful shutdown with active execution waiting."""
@@ -1926,7 +1937,9 @@ class ExecutionManager(AppProcess):
             )
             logger.info(f"{prefix} ✅ Exec consumer has been signaled to stop")
         except Exception as e:
-            logger.error(f"{prefix} ⚠️ Error signaling consumer to stop: {type(e)} {e}")
+            logger.warning(
+                f"{prefix} ⚠️ Error signaling consumer to stop: {type(e)} {e}"
+            )
 
         # Wait for active executions to complete
         if self.active_graph_runs:
@@ -1957,7 +1970,7 @@ class ExecutionManager(AppProcess):
                 waited += wait_interval
 
             if self.active_graph_runs:
-                logger.error(
+                logger.warning(
                     f"{prefix} ⚠️ {len(self.active_graph_runs)} executions still running after {max_wait}s"
                 )
             else:
@@ -1968,7 +1981,7 @@ class ExecutionManager(AppProcess):
             self.executor.shutdown(cancel_futures=True, wait=False)
             logger.info(f"{prefix} ✅ Executor shutdown completed")
         except Exception as e:
-            logger.error(f"{prefix} ⚠️ Error during executor shutdown: {type(e)} {e}")
+            logger.warning(f"{prefix} ⚠️ Error during executor shutdown: {type(e)} {e}")
 
         # Release remaining execution locks
         try:

autogpt_platform/backend/backend/executor/scheduler.py

@@ -94,7 +94,7 @@ SCHEDULER_OPERATION_TIMEOUT_SECONDS = 300  # 5 minutes for scheduler operations
 def job_listener(event):
     """Logs job execution outcomes for better monitoring."""
     if event.exception:
-        logger.error(
+        logger.warning(
             f"Job {event.job_id} failed: {type(event.exception).__name__}: {event.exception}"
         )
     else:
@@ -137,7 +137,7 @@ def run_async(coro, timeout: float = SCHEDULER_OPERATION_TIMEOUT_SECONDS):
     try:
         return future.result(timeout=timeout)
     except Exception as e:
-        logger.error(f"Async operation failed: {type(e).__name__}: {e}")
+        logger.warning(f"Async operation failed: {type(e).__name__}: {e}")
         raise
 
 
@@ -186,7 +186,7 @@ async def _execute_graph(**kwargs):
 
 
 async def _handle_graph_validation_error(args: "GraphExecutionJobArgs") -> None:
-    logger.error(
+    logger.warning(
         f"Scheduled Graph {args.graph_id} failed validation. Unscheduling graph"
     )
     if args.schedule_id:
@@ -196,8 +196,9 @@ async def _handle_graph_validation_error(args: "GraphExecutionJobArgs") -> None:
             user_id=args.user_id,
         )
     else:
-        logger.error(
-            f"Unable to unschedule graph: {args.graph_id} as this is an old job with no associated schedule_id please remove manually"
+        logger.warning(
+            f"Unable to unschedule graph: {args.graph_id} as this is an old job "
+            f"with no associated schedule_id please remove manually"
         )
 
 

autogpt_platform/backend/backend/executor/utils.py

@@ -46,7 +46,7 @@ from backend.util.exceptions import (
 )
 from backend.util.logging import TruncatedLogger, is_structured_logging_enabled
 from backend.util.settings import Config
-from backend.util.type import convert
+from backend.util.type import coerce_inputs_to_schema
 
 config = Config()
 logger = TruncatedLogger(logging.getLogger(__name__), prefix="[GraphExecutorUtil]")
@@ -213,11 +213,8 @@ def validate_exec(
     if resolve_input:
         data = merge_execution_input(data)
 
-    # Convert non-matching data types to the expected input schema.
-    for name, data_type in schema.__annotations__.items():
-        value = data.get(name)
-        if (value is not None) and (type(value) is not data_type):
-            data[name] = convert(value, data_type)
+    # Coerce non-matching data types to the expected input schema.
+    coerce_inputs_to_schema(data, schema)
 
     # Input data post-merge should contain all required fields from the schema.
     if missing_input := schema.get_missing_input(data):

autogpt_platform/backend/backend/notifications/notifications.py

@@ -303,9 +303,9 @@ class NotificationManager(AppService):
                     )
 
                     if not oldest_message:
-                        # this should never happen
-                        logger.error(
-                            f"Batch for user {batch.user_id} and type {notification_type} has no oldest message whichshould never happen!!!!!!!!!!!!!!!!"
+                        logger.warning(
+                            f"Batch for user {batch.user_id} and type {notification_type} "
+                            f"has no oldest message — batch may have been cleared concurrently"
                         )
                         continue
 
@@ -318,7 +318,7 @@ class NotificationManager(AppService):
                         ).get_user_email_by_id(batch.user_id)
 
                         if not recipient_email:
-                            logger.error(
+                            logger.warning(
                                 f"User email not found for user {batch.user_id}"
                             )
                             continue
@@ -344,7 +344,7 @@ class NotificationManager(AppService):
                         ).get_user_notification_batch(batch.user_id, notification_type)
 
                         if not batch_data or not batch_data.notifications:
-                            logger.error(
+                            logger.warning(
                                 f"Batch data not found for user {batch.user_id}"
                             )
                             # Clear the batch
@@ -372,7 +372,7 @@ class NotificationManager(AppService):
                                     )
                                 )
                             except Exception as e:
-                                logger.error(
+                                logger.warning(
                                     f"Error parsing notification event: {e=}, {db_event=}"
                                 )
                                 continue
@@ -415,7 +415,10 @@ class NotificationManager(AppService):
     async def discord_system_alert(
         self, content: str, channel: DiscordChannel = DiscordChannel.PLATFORM
     ):
-        await discord_send_alert(content, channel)
+        try:
+            await discord_send_alert(content, channel)
+        except Exception as e:
+            logger.warning(f"Failed to send Discord system alert: {e}")
 
     async def _queue_scheduled_notification(self, event: SummaryParamsEventModel):
         """Queue a scheduled notification - exposed method for other services to call"""
@@ -516,7 +519,7 @@ class NotificationManager(AppService):
                 raise ValueError("Invalid event type or params")
 
         except Exception as e:
-            logger.error(f"Failed to gather summary data: {e}")
+            logger.warning(f"Failed to gather summary data: {e}")
             # Return sensible defaults in case of error
             if event_type == NotificationType.DAILY_SUMMARY and isinstance(
                 params, DailySummaryParams
@@ -562,8 +565,9 @@ class NotificationManager(AppService):
             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!!!!!!!!!!!!!!!!"
+            logger.warning(
+                f"Batch for user {user_id} and type {event_type} "
+                f"has no oldest message — batch may have been cleared concurrently"
             )
             return False
         oldest_age = oldest_message.created_at
@@ -585,7 +589,7 @@ class NotificationManager(AppService):
                 get_notif_data_type(event.type)
             ].model_validate_json(message)
         except Exception as e:
-            logger.error(f"Error parsing message due to non matching schema {e}")
+            logger.warning(f"Error parsing message due to non matching schema {e}")
             return None
 
     async def _process_admin_message(self, message: str) -> bool:
@@ -614,7 +618,7 @@ class NotificationManager(AppService):
                 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}")
+                logger.warning(f"User email not found for user {event.user_id}")
                 return False
 
             should_send = await self._should_email_user_based_on_preference(
@@ -651,7 +655,7 @@ class NotificationManager(AppService):
                 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}")
+                logger.warning(f"User email not found for user {event.user_id}")
                 return False
 
             should_send = await self._should_email_user_based_on_preference(
@@ -672,7 +676,7 @@ class NotificationManager(AppService):
                 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}")
+                logger.warning(f"Batch not found for user {event.user_id}")
                 return False
             unsub_link = generate_unsubscribe_link(event.user_id)
 
@@ -745,7 +749,7 @@ class NotificationManager(AppService):
                                         f"Removed {len(chunk_ids)} sent notifications from batch"
                                     )
                                 except Exception as e:
-                                    logger.error(
+                                    logger.warning(
                                         f"Failed to remove sent notifications: {e}"
                                     )
                                     # Continue anyway - better to risk duplicates than lose emails
@@ -770,7 +774,7 @@ class NotificationManager(AppService):
                         else:
                             # Message is too large even after size reduction
                             if attempt_size == 1:
-                                logger.error(
+                                logger.warning(
                                     f"Failed to send notification at index {i}: "
                                     f"Single notification exceeds email size limit "
                                     f"({len(test_message):,} chars > {MAX_EMAIL_SIZE:,} chars). "
@@ -789,7 +793,7 @@ class NotificationManager(AppService):
                                             f"Removed oversized notification {chunk_ids[0]} from batch permanently"
                                         )
                                     except Exception as e:
-                                        logger.error(
+                                        logger.warning(
                                             f"Failed to remove oversized notification: {e}"
                                         )
 
@@ -823,7 +827,7 @@ class NotificationManager(AppService):
                                         f"Set email verification to false for user {event.user_id}"
                                     )
                                 except Exception as deactivation_error:
-                                    logger.error(
+                                    logger.warning(
                                         f"Failed to deactivate email for user {event.user_id}: "
                                         f"{deactivation_error}"
                                     )
@@ -835,7 +839,7 @@ class NotificationManager(AppService):
                                         f"Disabled all notification preferences for user {event.user_id}"
                                     )
                                 except Exception as disable_error:
-                                    logger.error(
+                                    logger.warning(
                                         f"Failed to disable notification preferences: {disable_error}"
                                     )
 
@@ -848,7 +852,7 @@ class NotificationManager(AppService):
                                         f"Cleared ALL notification batches for user {event.user_id}"
                                     )
                                 except Exception as remove_error:
-                                    logger.error(
+                                    logger.warning(
                                         f"Failed to clear batches for inactive recipient: {remove_error}"
                                     )
 
@@ -859,7 +863,7 @@ class NotificationManager(AppService):
                                 "422" in error_message
                                 or "unprocessable" in error_message
                             ):
-                                logger.error(
+                                logger.warning(
                                     f"Failed to send notification at index {i}: "
                                     f"Malformed notification data rejected by Postmark. "
                                     f"Error: {e}. Removing from batch permanently."
@@ -877,7 +881,7 @@ class NotificationManager(AppService):
                                             "Removed malformed notification from batch permanently"
                                         )
                                     except Exception as remove_error:
-                                        logger.error(
+                                        logger.warning(
                                             f"Failed to remove malformed notification: {remove_error}"
                                         )
                             # Check if it's a ValueError for size limit
@@ -885,14 +889,14 @@ class NotificationManager(AppService):
                                 isinstance(e, ValueError)
                                 and "too large" in error_message
                             ):
-                                logger.error(
+                                logger.warning(
                                     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(
+                                logger.warning(
                                     f"Failed to send notification at index {i}: "
                                     f"Email API error ({error_type}): {e}. "
                                     f"Skipping this notification."
@@ -907,7 +911,9 @@ class NotificationManager(AppService):
 
                 if not chunk_sent:
                     # Should not reach here due to single notification handling
-                    logger.error(f"Failed to send notifications starting at index {i}")
+                    logger.warning(
+                        f"Failed to send notifications starting at index {i}"
+                    )
                     failed_indices.append(i)
                     i += 1
 
@@ -946,7 +952,7 @@ class NotificationManager(AppService):
                 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}")
+                logger.warning(f"User email not found for user {event.user_id}")
                 return False
             should_send = await self._should_email_user_based_on_preference(
                 event.user_id, event.type
@@ -1007,7 +1013,10 @@ class NotificationManager(AppService):
                         # Let message.process() handle the rejection
                         pass
                     except Exception as e:
-                        logger.error(f"Error processing message in {queue_name}: {e}")
+                        logger.warning(
+                            f"Error processing message in {queue_name}: {e}",
+                            exc_info=True,
+                        )
                         # Let message.process() handle the rejection
                         raise
         except asyncio.CancelledError:

autogpt_platform/backend/backend/notifications/test_notifications.py

@@ -256,9 +256,9 @@ class TestNotificationErrorHandling:
             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]
+            warning_calls = [call[0][0] for call in mock_logger.warning.call_args_list]
             assert any(
-                "422" in call or "malformed" in call.lower() for call in error_calls
+                "422" in call or "malformed" in call.lower() for call in warning_calls
             )
 
             # Verify all notifications were removed (4 successful + 1 malformed)
@@ -371,10 +371,10 @@ class TestNotificationErrorHandling:
             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]
+            warning_calls = [call[0][0] for call in mock_logger.warning.call_args_list]
             assert any(
                 "exceeds email size limit" in call or "oversized" in call.lower()
-                for call in error_calls
+                for call in warning_calls
             )
 
     @pytest.mark.asyncio
@@ -478,10 +478,10 @@ class TestNotificationErrorHandling:
             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]
+            warning_calls = [call[0][0] for call in mock_logger.warning.call_args_list]
             assert any(
                 "api error" in call.lower() or "skipping" in call.lower()
-                for call in error_calls
+                for call in warning_calls
             )
 
             # Only successful ones should be removed from batch (failed one stays for retry)

autogpt_platform/backend/backend/util/cloud_storage.py

@@ -613,5 +613,5 @@ async def cleanup_expired_files_async() -> int:
             )
             return deleted_count
         except Exception as e:
-            logger.error(f"[CloudStorage] Error during cloud storage cleanup: {e}")
+            logger.warning(f"[CloudStorage] Error during cloud storage cleanup: {e}")
             return 0

autogpt_platform/backend/backend/util/metrics.py

@@ -10,7 +10,7 @@ from sentry_sdk.integrations.launchdarkly import LaunchDarklyIntegration
 from sentry_sdk.integrations.logging import LoggingIntegration
 
 from backend.util import feature_flag
-from backend.util.settings import Settings
+from backend.util.settings import BehaveAs, Settings
 
 settings = Settings()
 logger = logging.getLogger(__name__)
@@ -21,6 +21,95 @@ class DiscordChannel(str, Enum):
     PRODUCT = "product"  # For product alerts (low balance, zero balance, etc.)
 
 
+def _before_send(event, hint):
+    """Filter out expected/transient errors from Sentry to reduce noise."""
+    if "exc_info" in hint:
+        exc_type, exc_value, _ = hint["exc_info"]
+        exc_msg = str(exc_value).lower() if exc_value else ""
+
+        # AMQP/RabbitMQ transient connection errors — expected during deploys
+        amqp_keywords = [
+            "amqpconnection",
+            "amqpconnector",
+            "connection_forced",
+            "channelinvalidstateerror",
+            "no active transport",
+        ]
+        if any(kw in exc_msg for kw in amqp_keywords):
+            return None
+
+        # "connection refused" only for AMQP-related exceptions (not other services)
+        if "connection refused" in exc_msg:
+            exc_module = getattr(exc_type, "__module__", "") or ""
+            exc_name = getattr(exc_type, "__name__", "") or ""
+            amqp_indicators = ["aio_pika", "aiormq", "amqp", "pika", "rabbitmq"]
+            if any(
+                ind in exc_module.lower() or ind in exc_name.lower()
+                for ind in amqp_indicators
+            ) or any(kw in exc_msg for kw in ["amqp", "pika", "rabbitmq"]):
+                return None
+
+        # User-caused credential/auth errors — not platform bugs
+        user_auth_keywords = [
+            "incorrect api key",
+            "invalid x-api-key",
+            "missing authentication header",
+            "invalid api token",
+            "authentication_error",
+        ]
+        if any(kw in exc_msg for kw in user_auth_keywords):
+            return None
+
+        # Expected business logic — insufficient balance
+        if "insufficient balance" in exc_msg or "no credits left" in exc_msg:
+            return None
+
+        # Expected security check — blocked IP access
+        if "access to blocked or private ip" in exc_msg:
+            return None
+
+        # Discord bot token misconfiguration — not a platform error
+        if "improper token has been passed" in exc_msg or (
+            exc_type and exc_type.__name__ == "Forbidden" and "50001" in exc_msg
+        ):
+            return None
+
+        # Google metadata DNS errors — expected in non-GCP environments
+        if (
+            "metadata.google.internal" in exc_msg
+            and settings.config.behave_as != BehaveAs.CLOUD
+        ):
+            return None
+
+        # Inactive email recipients — expected for bounced addresses
+        if "marked as inactive" in exc_msg or "inactive addresses" in exc_msg:
+            return None
+
+    # Also filter log-based events for known noisy messages.
+    # Sentry's LoggingIntegration stores log messages under "logentry", not "message".
+    logentry = event.get("logentry") or {}
+    log_msg = (
+        logentry.get("formatted") or logentry.get("message") or event.get("message")
+    )
+    if event.get("logger") and log_msg:
+        msg = log_msg.lower()
+        noisy_patterns = [
+            "amqpconnection",
+            "connection_forced",
+            "unclosed client session",
+            "unclosed connector",
+        ]
+        if any(p in msg for p in noisy_patterns):
+            return None
+        # "connection refused" in logs only when AMQP-related context is present
+        if "connection refused" in msg and any(
+            ind in msg for ind in ("amqp", "pika", "rabbitmq", "aio_pika", "aiormq")
+        ):
+            return None
+
+    return event
+
+
 def sentry_init():
     sentry_dsn = settings.secrets.sentry_dsn
     integrations = []
@@ -35,6 +124,7 @@ def sentry_init():
         profiles_sample_rate=1.0,
         environment=f"app:{settings.config.app_env.value}-behave:{settings.config.behave_as.value}",
         _experiments={"enable_logs": True},
+        before_send=_before_send,
         integrations=[
             AsyncioIntegration(),
             LoggingIntegration(sentry_logs_level=logging.INFO),

autogpt_platform/backend/backend/util/retry.py

@@ -64,7 +64,7 @@ def send_rate_limited_discord_alert(
         return True
 
     except Exception as alert_error:
-        logger.error(f"Failed to send Discord alert: {alert_error}")
+        logger.warning(f"Failed to send Discord alert: {alert_error}")
         return False
 
 
@@ -182,7 +182,8 @@ def conn_retry(
         func_name = getattr(retry_state.fn, "__name__", "unknown")
 
         if retry_state.outcome.failed and retry_state.next_action is None:
-            logger.error(f"{prefix} {action_name} failed after retries: {exception}")
+            # Final failure is logged by sync_wrapper/async_wrapper — skip here to avoid duplicates
+            pass
         else:
             if attempt_number == EXCESSIVE_RETRY_THRESHOLD:
                 if send_rate_limited_discord_alert(
@@ -225,7 +226,7 @@ def conn_retry(
                 logger.info(f"{prefix} {action_name} completed successfully.")
                 return result
             except Exception as e:
-                logger.error(f"{prefix} {action_name} failed after retries: {e}")
+                logger.warning(f"{prefix} {action_name} failed after retries: {e}")
                 raise
 
         @wraps(func)
@@ -237,7 +238,7 @@ def conn_retry(
                 logger.info(f"{prefix} {action_name} completed successfully.")
                 return result
             except Exception as e:
-                logger.error(f"{prefix} {action_name} failed after retries: {e}")
+                logger.warning(f"{prefix} {action_name} failed after retries: {e}")
                 raise
 
         return async_wrapper if is_coroutine else sync_wrapper

autogpt_platform/backend/backend/util/settings.py

@@ -89,6 +89,10 @@ class Config(UpdateTrackingModel["Config"], BaseSettings):
         le=500,
         description="Thread pool size for FastAPI sync operations. All sync endpoints and dependencies automatically use this pool. Higher values support more concurrent sync operations but use more memory.",
     )
+    tally_extraction_llm_model: str = Field(
+        default="openai/gpt-4o-mini",
+        description="OpenRouter model ID used for extracting business understanding from Tally form data",
+    )
     ollama_host: str = Field(
         default="localhost:11434",
         description="Default Ollama host; exempted from SSRF checks.",
@@ -117,6 +121,10 @@ class Config(UpdateTrackingModel["Config"], BaseSettings):
         default=True,
         description="If authentication is enabled or not",
     )
+    enable_invite_gate: bool = Field(
+        default=True,
+        description="If the invite-only signup gate is enforced",
+    )
     enable_credit: bool = Field(
         default=False,
         description="If user credit system is enabled or not",

autogpt_platform/backend/backend/util/type.py

@@ -249,6 +249,87 @@ def convert(value: Any, target_type: Any) -> Any:
         raise ConversionError(f"Failed to convert {value} to {target_type}") from e
 
 
+def _value_satisfies_type(value: Any, target: Any) -> bool:
+    """Check whether *value* already satisfies *target*, including inner elements.
+
+    For union types this checks each member; for generic container types it
+    recursively checks that inner elements satisfy the type args (e.g. every
+    element in a ``list[str]`` is a ``str``).  Returns ``False`` when uncertain
+    so the caller falls through to :func:`convert`.
+    """
+    # typing.Any cannot be used with isinstance(); treat as always satisfied.
+    if target is Any:
+        return True
+
+    origin = get_origin(target)
+
+    if origin is Union or origin is types.UnionType:
+        non_none = [a for a in get_args(target) if a is not type(None)]
+        return any(_value_satisfies_type(value, member) for member in non_none)
+
+    # Generic container type (e.g. list[str], dict[str, int])
+    if origin is not None:
+        # Guard: origin may not be a runtime type (e.g. Literal)
+        if not isinstance(origin, type):
+            return False
+        if not isinstance(value, origin):
+            return False
+        args = get_args(target)
+        if not args:
+            return True
+        # Check inner elements satisfy the type args
+        if _is_type_or_subclass(origin, list):
+            return all(_value_satisfies_type(v, args[0]) for v in value)
+        if _is_type_or_subclass(origin, dict) and len(args) >= 2:
+            return all(
+                _value_satisfies_type(k, args[0]) and _value_satisfies_type(v, args[1])
+                for k, v in value.items()
+            )
+        if (
+            _is_type_or_subclass(origin, set) or _is_type_or_subclass(origin, frozenset)
+        ) and args:
+            return all(_value_satisfies_type(v, args[0]) for v in value)
+        if _is_type_or_subclass(origin, tuple):
+            # Homogeneous tuple[T, ...] — single type + Ellipsis
+            if len(args) == 2 and args[1] is Ellipsis:
+                return all(_value_satisfies_type(v, args[0]) for v in value)
+            # Heterogeneous tuple[T1, T2, ...] — positional types
+            if len(value) != len(args):
+                return False
+            return all(_value_satisfies_type(v, t) for v, t in zip(value, args))
+        # Unhandled generic origin — fall through to convert()
+        return False
+
+    # Simple type (e.g. str, int)
+    if isinstance(target, type):
+        return isinstance(value, target)
+
+    return False
+
+
+def coerce_inputs_to_schema(data: dict[str, Any], schema: type) -> None:
+    """Coerce *data* values in-place to match *schema*'s field types.
+
+    Uses ``model_fields`` (not ``__annotations__``) so inherited fields are
+    included.  Skips coercion when the value already satisfies the target
+    type — in particular for union-typed fields where the value matches one
+    member but differs from the annotation object itself.
+
+    This is the single authoritative coercion step shared by the executor
+    (``validate_exec``) and the CoPilot (``execute_block``).
+    """
+    for name, field_info in schema.model_fields.items():
+        value = data.get(name)
+        if value is None:
+            continue
+        target = field_info.annotation
+        if target is None:
+            continue
+        if _value_satisfies_type(value, target):
+            continue
+        data[name] = convert(value, target)
+
+
 class FormattedStringType(str):
     string_format: str
 

autogpt_platform/backend/backend/util/type_test.py

@@ -1,6 +1,8 @@
-from typing import List, Optional
+from typing import Any, List, Literal, Optional
 
-from backend.util.type import convert
+from pydantic import BaseModel
+
+from backend.util.type import _value_satisfies_type, coerce_inputs_to_schema, convert
 
 
 def test_type_conversion():
@@ -46,3 +48,343 @@ def test_type_conversion():
     # Test other empty list conversions
     assert convert([], int) == 0  # len([]) = 0
     assert convert([], Optional[int]) == 0
+
+
+# ---------------------------------------------------------------------------
+# _value_satisfies_type
+# ---------------------------------------------------------------------------
+
+
+class TestValueSatisfiesType:
+    # --- simple types ---
+    def test_simple_match(self):
+        assert _value_satisfies_type("hello", str) is True
+        assert _value_satisfies_type(42, int) is True
+        assert _value_satisfies_type(3.14, float) is True
+        assert _value_satisfies_type(True, bool) is True
+
+    def test_simple_mismatch(self):
+        assert _value_satisfies_type("hello", int) is False
+        assert _value_satisfies_type(42, str) is False
+        assert _value_satisfies_type([1, 2], str) is False
+
+    # --- Any ---
+    def test_any_always_satisfied(self):
+        assert _value_satisfies_type("hello", Any) is True
+        assert _value_satisfies_type(42, Any) is True
+        assert _value_satisfies_type([1, 2], Any) is True
+        assert _value_satisfies_type(None, Any) is True
+
+    # --- Optional / Union ---
+    def test_optional_with_value(self):
+        assert _value_satisfies_type("hello", Optional[str]) is True
+        assert _value_satisfies_type(42, Optional[int]) is True
+
+    def test_optional_mismatch(self):
+        assert _value_satisfies_type(42, Optional[str]) is False
+
+    def test_union_matches_first_member(self):
+        assert _value_satisfies_type("hello", str | list[str]) is True
+
+    def test_union_matches_second_member(self):
+        assert _value_satisfies_type(["a", "b"], str | list[str]) is True
+
+    def test_union_no_match(self):
+        assert _value_satisfies_type(42, str | list[str]) is False
+
+    # --- list[T] ---
+    def test_list_str_all_match(self):
+        assert _value_satisfies_type(["a", "b", "c"], list[str]) is True
+
+    def test_list_str_inner_mismatch(self):
+        assert _value_satisfies_type([1, 2, 3], list[str]) is False
+
+    def test_list_int_all_match(self):
+        assert _value_satisfies_type([1, 2, 3], list[int]) is True
+
+    def test_list_int_inner_mismatch(self):
+        assert _value_satisfies_type(["1", "2"], list[int]) is False
+
+    def test_empty_list_satisfies_any_list_type(self):
+        assert _value_satisfies_type([], list[str]) is True
+        assert _value_satisfies_type([], list[int]) is True
+
+    def test_string_does_not_satisfy_list(self):
+        assert _value_satisfies_type("hello", list[str]) is False
+
+    # --- nested list[list[str]] ---
+    def test_nested_list_all_match(self):
+        assert _value_satisfies_type([["a", "b"], ["c"]], list[list[str]]) is True
+
+    def test_nested_list_inner_mismatch(self):
+        assert _value_satisfies_type([["a", 1], ["c"]], list[list[str]]) is False
+
+    def test_nested_list_outer_mismatch(self):
+        assert _value_satisfies_type(["a", "b"], list[list[str]]) is False
+
+    # --- dict[K, V] ---
+    def test_dict_str_int_match(self):
+        assert _value_satisfies_type({"a": 1, "b": 2}, dict[str, int]) is True
+
+    def test_dict_str_int_value_mismatch(self):
+        assert _value_satisfies_type({"a": "1", "b": "2"}, dict[str, int]) is False
+
+    def test_dict_str_int_key_mismatch(self):
+        assert _value_satisfies_type({1: 1, 2: 2}, dict[str, int]) is False
+
+    def test_empty_dict_satisfies(self):
+        assert _value_satisfies_type({}, dict[str, int]) is True
+
+    # --- set[T] / tuple[T] ---
+    def test_set_match(self):
+        assert _value_satisfies_type({1, 2, 3}, set[int]) is True
+
+    def test_set_mismatch(self):
+        assert _value_satisfies_type({"a", "b"}, set[int]) is False
+
+    def test_tuple_homogeneous_match(self):
+        assert _value_satisfies_type((1, 2, 3), tuple[int, ...]) is True
+
+    def test_tuple_homogeneous_mismatch(self):
+        assert _value_satisfies_type((1, "2", 3), tuple[int, ...]) is False
+
+    def test_tuple_heterogeneous_match(self):
+        assert _value_satisfies_type(("a", 1, True), tuple[str, int, bool]) is True
+
+    def test_tuple_heterogeneous_mismatch(self):
+        assert _value_satisfies_type(("a", "b", True), tuple[str, int, bool]) is False
+
+    def test_tuple_heterogeneous_wrong_length(self):
+        assert _value_satisfies_type(("a", 1), tuple[str, int, bool]) is False
+
+    # --- bare generics (no args) ---
+    def test_bare_list(self):
+        assert _value_satisfies_type([1, "a"], list) is True
+
+    def test_bare_dict(self):
+        assert _value_satisfies_type({"a": 1}, dict) is True
+
+    # --- union with generic inner mismatch ---
+    def test_union_list_with_wrong_inner_falls_through(self):
+        # [1, 2] doesn't satisfy list[str] (inner mismatch), and not str either
+        assert _value_satisfies_type([1, 2], str | list[str]) is False
+
+    # --- Literal (non-runtime origin) ---
+    def test_literal_does_not_crash(self):
+        """Literal origins are not runtime types — should return False, not crash."""
+        assert _value_satisfies_type("active", Literal["active", "inactive"]) is False
+
+
+# ---------------------------------------------------------------------------
+# coerce_inputs_to_schema — using real Pydantic models
+# ---------------------------------------------------------------------------
+
+
+class SampleSchema(BaseModel):
+    name: str
+    count: int
+    items: list[str]
+    config: dict[str, int] = {}
+
+
+class NestedSchema(BaseModel):
+    rows: list[list[str]]
+
+
+class UnionSchema(BaseModel):
+    content: str | list[str]
+
+
+class OptionalSchema(BaseModel):
+    label: Optional[str] = None
+    value: int = 0
+
+
+class AnyFieldSchema(BaseModel):
+    data: Any
+
+
+class TestCoerceInputsToSchema:
+    def test_string_to_int(self):
+        data: dict[str, Any] = {"name": "test", "count": "42", "items": ["a"]}
+        coerce_inputs_to_schema(data, SampleSchema)
+        assert data["count"] == 42
+        assert isinstance(data["count"], int)
+
+    def test_json_string_to_list(self):
+        data: dict[str, Any] = {"name": "test", "count": 1, "items": '["a","b","c"]'}
+        coerce_inputs_to_schema(data, SampleSchema)
+        assert data["items"] == ["a", "b", "c"]
+
+    def test_already_correct_types_unchanged(self):
+        data: dict[str, Any] = {
+            "name": "test",
+            "count": 42,
+            "items": ["a", "b"],
+            "config": {"x": 1},
+        }
+        coerce_inputs_to_schema(data, SampleSchema)
+        assert data == {
+            "name": "test",
+            "count": 42,
+            "items": ["a", "b"],
+            "config": {"x": 1},
+        }
+
+    def test_inner_element_coercion(self):
+        """list[str] with int inner elements → coerced to strings."""
+        data: dict[str, Any] = {"name": "test", "count": 1, "items": [1, 2, 3]}
+        coerce_inputs_to_schema(data, SampleSchema)
+        assert data["items"] == ["1", "2", "3"]
+
+    def test_dict_value_coercion(self):
+        """dict[str, int] with string values → coerced to ints."""
+        data: dict[str, Any] = {
+            "name": "test",
+            "count": 1,
+            "items": [],
+            "config": {"x": "10", "y": "20"},
+        }
+        coerce_inputs_to_schema(data, SampleSchema)
+        assert data["config"] == {"x": 10, "y": 20}
+
+    def test_nested_list_from_json_string(self):
+        data: dict[str, Any] = {
+            "rows": '[["Name","Score"],["Alice","90"]]',
+        }
+        coerce_inputs_to_schema(data, NestedSchema)
+        assert data["rows"] == [["Name", "Score"], ["Alice", "90"]]
+
+    def test_nested_list_already_correct(self):
+        original = [["a", "b"], ["c", "d"]]
+        data: dict[str, Any] = {"rows": original}
+        coerce_inputs_to_schema(data, NestedSchema)
+        assert data["rows"] == original
+
+    def test_union_preserves_valid_list(self):
+        """list[str] value should NOT be stringified for str | list[str]."""
+        data: dict[str, Any] = {"content": ["a", "b"]}
+        coerce_inputs_to_schema(data, UnionSchema)
+        assert data["content"] == ["a", "b"]
+        assert isinstance(data["content"], list)
+
+    def test_union_preserves_valid_string(self):
+        data: dict[str, Any] = {"content": "hello"}
+        coerce_inputs_to_schema(data, UnionSchema)
+        assert data["content"] == "hello"
+
+    def test_union_list_with_wrong_inner_gets_coerced(self):
+        """[1, 2] for str | list[str] — inner ints don't match list[str],
+        so convert() is called. convert tries str first → stringifies."""
+        data: dict[str, Any] = {"content": [1, 2]}
+        coerce_inputs_to_schema(data, UnionSchema)
+        # convert([1,2], str | list[str]) tries str first → "[1, 2]"
+        # This is convert()'s union behavior — str wins over list[str]
+        assert isinstance(data["content"], (str, list))
+
+    def test_skips_none_values(self):
+        data: dict[str, Any] = {"label": None, "value": "5"}
+        coerce_inputs_to_schema(data, OptionalSchema)
+        assert data["label"] is None
+        assert data["value"] == 5
+
+    def test_skips_missing_fields(self):
+        data: dict[str, Any] = {"value": "10"}
+        coerce_inputs_to_schema(data, OptionalSchema)
+        assert "label" not in data
+        assert data["value"] == 10
+
+    def test_any_field_skipped(self):
+        """Fields typed as Any should pass through without coercion."""
+        data: dict[str, Any] = {"data": [1, "mixed", {"nested": True}]}
+        coerce_inputs_to_schema(data, AnyFieldSchema)
+        assert data["data"] == [1, "mixed", {"nested": True}]
+
+    def test_preserves_all_convert_capabilities(self):
+        """Verify coerce_inputs_to_schema doesn't lose any convert() capability
+        that existed before the _value_satisfies_type gate was added."""
+
+        class FullSchema(BaseModel):
+            as_int: int
+            as_float: float
+            as_bool: bool
+            as_str: str
+            as_list: list[int]
+            as_dict: dict[str, str]
+
+        data: dict[str, Any] = {
+            "as_int": "42",
+            "as_float": "3.14",
+            "as_bool": "True",
+            "as_str": 123,
+            "as_list": "[1,2,3]",
+            "as_dict": '{"a": "b"}',
+        }
+        coerce_inputs_to_schema(data, FullSchema)
+        assert data["as_int"] == 42
+        assert data["as_float"] == 3.14
+        assert data["as_bool"] is True
+        assert data["as_str"] == "123"
+        assert data["as_list"] == [1, 2, 3]
+        assert data["as_dict"] == {"a": "b"}
+
+    def test_inherited_fields_are_coerced(self):
+        """model_fields includes inherited fields; __annotations__ does not.
+        This verifies that fields from a parent schema are still coerced."""
+
+        class ParentSchema(BaseModel):
+            base_count: int
+
+        class ChildSchema(ParentSchema):
+            name: str
+
+        # base_count is inherited — __annotations__ wouldn't include it
+        assert "base_count" not in ChildSchema.__annotations__
+        assert "base_count" in ChildSchema.model_fields
+
+        data: dict[str, Any] = {"base_count": "42", "name": "test"}
+        coerce_inputs_to_schema(data, ChildSchema)
+        assert data["base_count"] == 42
+        assert isinstance(data["base_count"], int)
+
+    def test_nested_pydantic_model_field(self):
+        """dict input for a Pydantic model-typed field passes through.
+        convert() doesn't construct Pydantic models — Pydantic validation
+        handles that downstream. This test documents the behavior."""
+
+        class InnerModel(BaseModel):
+            x: int
+
+        class OuterModel(BaseModel):
+            inner: InnerModel
+
+        data: dict[str, Any] = {"inner": {"x": 1}}
+        coerce_inputs_to_schema(data, OuterModel)
+        # dict stays as dict — convert() doesn't construct Pydantic models
+        assert data["inner"] == {"x": 1}
+        assert isinstance(data["inner"], dict)
+
+    def test_literal_field_passes_through(self):
+        """Literal-typed fields should not crash coercion."""
+
+        class LiteralSchema(BaseModel):
+            status: Literal["active", "inactive"]
+
+        data: dict[str, Any] = {"status": "active"}
+        coerce_inputs_to_schema(data, LiteralSchema)
+        assert data["status"] == "active"
+
+    def test_list_of_pydantic_model_field(self):
+        """list[dict] for list[PydanticModel] passes through unchanged."""
+
+        class ItemModel(BaseModel):
+            name: str
+
+        class ContainerModel(BaseModel):
+            items: list[ItemModel]
+
+        data: dict[str, Any] = {"items": [{"name": "a"}, {"name": "b"}]}
+        coerce_inputs_to_schema(data, ContainerModel)
+        # Dicts stay as dicts — Pydantic validation handles construction
+        assert data["items"] == [{"name": "a"}, {"name": "b"}]
+        assert isinstance(data["items"][0], dict)

autogpt_platform/backend/generate_views.py

@@ -0,0 +1,246 @@
+#!/usr/bin/env python3
+"""
+AutoGPT Analytics — View Generator
+====================================
+Reads every .sql file in analytics/queries/ and registers it as a
+CREATE OR REPLACE VIEW in the analytics schema.
+
+Quick start (from autogpt_platform/backend/):
+
+Step 1 — one-time setup (creates schema, role, grants):
+
+  poetry run analytics-setup
+
+Step 2 — create / refresh all 14 analytics views:
+
+  poetry run analytics-views
+
+Both commands auto-detect credentials from .env (DB_* vars).
+Use --db-url to override.
+
+Step 3 (optional) — enable login and set a password for the read-only
+role so external tools (Supabase MCP, PostHog Data Warehouse) can connect.
+The role is created as NOLOGIN, so you must grant LOGIN at the same time.
+Run in Supabase SQL Editor:
+
+  ALTER ROLE analytics_readonly WITH LOGIN PASSWORD 'your-password';
+
+Usage
+-----
+  poetry run analytics-setup              # apply setup to DB
+  poetry run analytics-setup --dry-run   # print setup SQL only
+  poetry run analytics-views             # apply all views to DB
+  poetry run analytics-views --dry-run   # print all view SQL only
+  poetry run analytics-views --only graph_execution,retention_login_weekly
+
+Environment variables
+---------------------
+  DATABASE_URL   Postgres connection string (checked before .env)
+
+Notes
+-----
+- .env DB_* vars are read automatically as a fallback.
+- Safe to re-run: uses CREATE OR REPLACE VIEW.
+- Looker, PostHog Data Warehouse, and Supabase MCP all read from the
+  same analytics.* views — no raw tables exposed.
+"""
+
+import argparse
+import os
+import sys
+from pathlib import Path
+from urllib.parse import quote
+
+QUERIES_DIR = Path(__file__).parent.parent / "analytics" / "queries"
+ENV_FILE = Path(__file__).parent / ".env"
+SCHEMA = "analytics"
+
+SETUP_SQL = """\
+-- =============================================================
+-- AutoGPT Analytics Schema Setup
+-- Run ONCE as the postgres superuser (e.g. via Supabase SQL Editor).
+-- After this, run: poetry run analytics-views
+-- =============================================================
+
+-- 1. Create the analytics schema
+CREATE SCHEMA IF NOT EXISTS analytics;
+
+-- 2. Create the read-only role (skip if already exists)
+DO $$
+BEGIN
+  IF NOT EXISTS (SELECT FROM pg_roles WHERE rolname = 'analytics_readonly') THEN
+    CREATE ROLE analytics_readonly NOLOGIN;
+  END IF;
+END
+$$;
+
+-- 3. Analytics schema grants only.
+--    Views use security_invoker = false so they execute as their
+--    owner (postgres). analytics_readonly never needs direct access
+--    to the platform or auth schemas.
+GRANT USAGE ON SCHEMA analytics TO analytics_readonly;
+GRANT SELECT ON ALL TABLES IN SCHEMA analytics TO analytics_readonly;
+ALTER DEFAULT PRIVILEGES IN SCHEMA analytics
+  GRANT SELECT ON TABLES TO analytics_readonly;
+"""
+
+
+def load_db_url_from_env() -> str | None:
+    """Read DB_* vars from .env and build a psycopg2 connection string."""
+    if not ENV_FILE.exists():
+        return None
+    env: dict[str, str] = {}
+    for line in ENV_FILE.read_text().splitlines():
+        line = line.strip()
+        if not line or line.startswith("#") or "=" not in line:
+            continue
+        key, _, value = line.partition("=")
+        env[key.strip()] = value.strip().strip('"').strip("'")
+    host = env.get("DB_HOST", "localhost")
+    port = env.get("DB_PORT", "5432")
+    user = env.get("DB_USER", "postgres")
+    password = env.get("DB_PASS", "")
+    dbname = env.get("DB_NAME", "postgres")
+    if not password:
+        return None
+    return (
+        "postgresql://"
+        f"{quote(user, safe='')}:{quote(password, safe='')}"
+        f"@{host}:{port}/{quote(dbname, safe='')}"
+    )
+
+
+def get_db_url(args: argparse.Namespace) -> str | None:
+    return args.db_url or os.environ.get("DATABASE_URL") or load_db_url_from_env()
+
+
+def connect(db_url: str):
+    try:
+        import psycopg2
+    except ImportError:
+        print("psycopg2 not found. Run: poetry install", file=sys.stderr)
+        sys.exit(1)
+    return psycopg2.connect(db_url)
+
+
+def run_sql(db_url: str, statements: list[tuple[str, str]]) -> None:
+    """Execute a list of (label, sql) pairs in a single transaction."""
+    conn = connect(db_url)
+    conn.autocommit = False
+    cur = conn.cursor()
+    try:
+        for label, sql in statements:
+            print(f"  {label} ...", end=" ")
+            cur.execute(sql)
+            print("OK")
+        conn.commit()
+        print(f"\n✓ {len(statements)} statement(s) applied.")
+    except Exception as e:
+        conn.rollback()
+        print(f"\n✗ Error: {e}", file=sys.stderr)
+        sys.exit(1)
+    finally:
+        cur.close()
+        conn.close()
+
+
+def build_view_sql(name: str, query_body: str) -> str:
+    body = query_body.strip().rstrip(";")
+    # security_invoker = false → view runs as its owner (postgres), not the
+    # caller, so analytics_readonly only needs analytics schema access.
+    return f"CREATE OR REPLACE VIEW {SCHEMA}.{name} WITH (security_invoker = false) AS\n{body};\n"
+
+
+def load_views(only: list[str] | None = None) -> list[tuple[str, str]]:
+    """Return [(label, sql)] for all views, in alphabetical order."""
+    files = sorted(QUERIES_DIR.glob("*.sql"))
+    if not files:
+        print(f"No .sql files found in {QUERIES_DIR}", file=sys.stderr)
+        sys.exit(1)
+    known = {f.stem for f in files}
+    if only:
+        unknown = [n for n in only if n not in known]
+        if unknown:
+            print(
+                f"Unknown view name(s): {', '.join(unknown)}\n"
+                f"Available: {', '.join(sorted(known))}",
+                file=sys.stderr,
+            )
+            sys.exit(1)
+    result = []
+    for f in files:
+        name = f.stem
+        if only and name not in only:
+            continue
+        result.append((f"view analytics.{name}", build_view_sql(name, f.read_text())))
+    return result
+
+
+def no_db_url_error() -> None:
+    print(
+        "No database URL found.\n"
+        "Tried: --db-url, DATABASE_URL env var, and .env (DB_* vars).\n"
+        "Use --dry-run to just print the SQL.",
+        file=sys.stderr,
+    )
+    sys.exit(1)
+
+
+def cmd_setup(args: argparse.Namespace) -> None:
+    if args.dry_run:
+        print(SETUP_SQL)
+        return
+    db_url = get_db_url(args)
+    if not db_url:
+        no_db_url_error()
+    assert db_url
+    print("Applying analytics setup...")
+    run_sql(db_url, [("schema / role / grants", SETUP_SQL)])
+
+
+def cmd_views(args: argparse.Namespace) -> None:
+    only = [v.strip() for v in args.only.split(",")] if args.only else None
+    views = load_views(only=only)
+    if not views:
+        print("No matching views found.")
+        sys.exit(0)
+
+    if args.dry_run:
+        print(f"-- {len(views)} views\n")
+        for label, sql in views:
+            print(f"-- {label}")
+            print(sql)
+        return
+
+    db_url = get_db_url(args)
+    if not db_url:
+        no_db_url_error()
+    assert db_url
+    print(f"Applying {len(views)} view(s)...")
+    # Append grant refresh so the readonly role sees any new views
+    grant = f"GRANT SELECT ON ALL TABLES IN SCHEMA {SCHEMA} TO analytics_readonly;"
+    run_sql(db_url, views + [("grant analytics_readonly", grant)])
+
+
+def main_setup() -> None:
+    parser = argparse.ArgumentParser(description="Apply analytics schema setup to DB")
+    parser.add_argument(
+        "--dry-run", action="store_true", help="Print SQL, don't execute"
+    )
+    parser.add_argument("--db-url", help="Postgres connection string")
+    cmd_setup(parser.parse_args())
+
+
+def main_views() -> None:
+    parser = argparse.ArgumentParser(description="Apply analytics views to DB")
+    parser.add_argument(
+        "--dry-run", action="store_true", help="Print SQL, don't execute"
+    )
+    parser.add_argument("--db-url", help="Postgres connection string")
+    parser.add_argument("--only", help="Comma-separated view names to update")
+    cmd_views(parser.parse_args())
+
+
+if __name__ == "__main__":
+    # Default: apply views (backwards-compatible with direct python invocation)
+    main_views()

autogpt_platform/backend/migrations/20260310090914_add_invite_system/migration.sql

@@ -0,0 +1,46 @@
+/*
+  Warnings:
+
+  - You are about to drop the column `search` on the `StoreListingVersion` table. All the data in the column will be lost.
+
+*/-- CreateEnum
+CREATE TYPE "InvitedUserStatus" AS ENUM('INVITED',
+'CLAIMED',
+'REVOKED');
+-- CreateEnum
+CREATE TYPE "TallyComputationStatus" AS ENUM('PENDING',
+'RUNNING',
+'READY',
+'FAILED');
+-- CreateTable
+CREATE TABLE "InvitedUser"(
+  "id" TEXT NOT NULL,
+  "createdAt" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
+  "updatedAt" TIMESTAMP(3) NOT NULL,
+  "email" TEXT NOT NULL,
+  "status" "InvitedUserStatus" NOT NULL DEFAULT 'INVITED',
+  "authUserId" TEXT,
+  "name" TEXT,
+  "tallyUnderstanding" JSONB,
+  "tallyStatus" "TallyComputationStatus" NOT NULL DEFAULT 'PENDING',
+  "tallyComputedAt" TIMESTAMP(3),
+  "tallyError" TEXT,
+  CONSTRAINT "InvitedUser_pkey" PRIMARY KEY("id")
+);
+-- CreateIndex
+CREATE UNIQUE INDEX "InvitedUser_email_key"
+ON "InvitedUser"("email");
+-- CreateIndex
+CREATE UNIQUE INDEX "InvitedUser_authUserId_key"
+ON "InvitedUser"("authUserId");
+-- CreateIndex
+CREATE INDEX "InvitedUser_status_idx"
+ON "InvitedUser"("status");
+-- CreateIndex
+CREATE INDEX "InvitedUser_tallyStatus_idx"
+ON "InvitedUser"("tallyStatus");
+-- AddForeignKey
+ALTER TABLE "InvitedUser" ADD CONSTRAINT "InvitedUser_authUserId_fkey" FOREIGN KEY("authUserId") REFERENCES "User"("id")
+ON DELETE 
+SET NULL
+ON UPDATE CASCADE;

autogpt_platform/backend/migrations/20260311000000_drop_auto_user_trigger/migration.sql

@@ -0,0 +1,15 @@
+-- Drop the trigger that auto-creates User + Profile on auth.users INSERT.
+-- The invite activation flow in get_or_activate_user() now handles this.
+DO $$
+BEGIN
+    IF EXISTS (
+        SELECT 1 FROM information_schema.tables
+        WHERE table_schema = 'auth' AND table_name = 'users'
+    ) THEN
+        DROP TRIGGER IF EXISTS user_added_to_platform ON auth.users;
+    END IF;
+END $$;
+
+DROP FUNCTION IF EXISTS add_user_and_profile_to_platform();
+DROP FUNCTION IF EXISTS add_user_to_platform();
+-- Keep generate_username() — used by backfill migration 20250205110132

autogpt_platform/backend/migrations/20260312093554_update_invites_indexs/migration.sql

@@ -0,0 +1,7 @@
+-- DropIndex
+DROP INDEX "InvitedUser_status_idx";
+-- DropIndex
+DROP INDEX "InvitedUser_tallyStatus_idx";
+-- CreateIndex
+CREATE INDEX "InvitedUser_createdAt_idx"
+ON "InvitedUser"("createdAt");

autogpt_platform/backend/poetry.lock

@@ -1282,14 +1282,14 @@ pgp = ["gpg"]
 
 [[package]]
 name = "e2b"
-version = "2.15.1"
+version = "2.15.2"
 description = "E2B SDK that give agents cloud environments"
 optional = false
 python-versions = "<4.0,>=3.10"
 groups = ["main"]
 files = [
-    {file = "e2b-2.15.1-py3-none-any.whl", hash = "sha256:a3bc4e004eab51fb05bae44e9ee4fe821e4637260f4ce3064c8f7c6ed7f5a2a0"},
-    {file = "e2b-2.15.1.tar.gz", hash = "sha256:a4f1bbc8b5180a8a1098079257fcb73e42503ed546098f676f722f11f0d68c09"},
+    {file = "e2b-2.15.2-py3-none-any.whl", hash = "sha256:19a56fbdea25974dc81426ed48337eae6cea91d404f5bcf8861a5a2c6e4d982a"},
+    {file = "e2b-2.15.2.tar.gz", hash = "sha256:414379d2421d6827eeb2eb50a4d6b3fdb7d691b39ff73b5ea05ca4b532819831"},
 ]
 
 [package.dependencies]
@@ -8882,4 +8882,4 @@ cffi = ["cffi (>=1.17,<2.0) ; platform_python_implementation != \"PyPy\" and pyt
 [metadata]
 lock-version = "2.1"
 python-versions = ">=3.10,<3.14"
-content-hash = "618d61b0586ab82fec1e28d1feb549a198e0b5c9d152e808862e55efc00a65b9"
+content-hash = "4e4365721cd3b68c58c237353b74adae1c64233fd4446904c335f23eb866fdca"

autogpt_platform/backend/pyproject.toml

@@ -20,7 +20,7 @@ claude-agent-sdk = "0.1.45"  # see copilot/sdk/sdk_compat_test.py for capability
 click = "^8.2.0"
 cryptography = "^46.0"
 discord-py = "^2.5.2"
-e2b = "^2.0"
+e2b = "^2.15.2"
 e2b-code-interpreter = "^2.0"
 elevenlabs = "^1.50.0"
 fastapi = "^0.128.6"
@@ -120,6 +120,8 @@ ws = "backend.ws:main"
 scheduler = "backend.scheduler:main"
 notification = "backend.notification:main"
 executor = "backend.exec:main"
+analytics-setup = "generate_views:main_setup"
+analytics-views = "generate_views:main_views"
 copilot-executor = "backend.copilot.executor.__main__:main"
 cli = "backend.cli:main"
 format = "linter:format"

autogpt_platform/backend/schema.prisma

@@ -65,6 +65,7 @@ model User {
   NotificationBatches   UserNotificationBatch[]
   PendingHumanReviews   PendingHumanReview[]
   Workspace             UserWorkspace?
+  ClaimedInvite         InvitedUser?            @relation("InvitedUserAuthUser")
 
   // OAuth Provider relations
   OAuthApplications       OAuthApplication[]
@@ -73,6 +74,38 @@ model User {
   OAuthRefreshTokens      OAuthRefreshToken[]
 }
 
+enum InvitedUserStatus {
+  INVITED
+  CLAIMED
+  REVOKED
+}
+
+enum TallyComputationStatus {
+  PENDING
+  RUNNING
+  READY
+  FAILED
+}
+
+model InvitedUser {
+  id        String   @id @default(uuid())
+  createdAt DateTime @default(now())
+  updatedAt DateTime @updatedAt
+
+  email      String            @unique
+  status     InvitedUserStatus @default(INVITED)
+  authUserId String?           @unique
+  AuthUser   User?             @relation("InvitedUserAuthUser", fields: [authUserId], references: [id], onDelete: SetNull)
+  name       String?
+
+  tallyUnderstanding Json?
+  tallyStatus        TallyComputationStatus @default(PENDING)
+  tallyComputedAt    DateTime?
+  tallyError         String?
+
+  @@index([createdAt])
+}
+
 enum OnboardingStep {
   // Introductory onboarding (Library)
   WELCOME
@@ -992,7 +1025,7 @@ model StoreListing {
   ActiveVersion   StoreListingVersion? @relation("ActiveVersion", fields: [activeVersionId], references: [id])
 
   // The agent link here is only so we can do lookup on agentId
-  agentGraphId String       @unique
+  agentGraphId String @unique
 
   owningUserId   String
   OwningUser     User    @relation(fields: [owningUserId], references: [id])

autogpt_platform/backend/test/e2e_test_data.py

@@ -34,7 +34,7 @@ from backend.data.auth.api_key import create_api_key
 from backend.data.credit import get_user_credit_model
 from backend.data.db import prisma
 from backend.data.graph import Graph, Link, Node, create_graph
-from backend.data.user import get_or_create_user
+from backend.data.invited_user import get_or_activate_user
 from backend.util.clients import get_supabase
 
 faker = Faker()
@@ -151,7 +151,7 @@ class TestDataCreator:
                 }
 
                 # Use the API function to create user in local database
-                user = await get_or_create_user(user_data)
+                user = await get_or_activate_user(user_data)
                 users.append(user.model_dump())
 
             except Exception as e:

autogpt_platform/frontend/CLAUDE.md

@@ -44,6 +44,12 @@ Do NOT skip these steps. If any command reports errors, fix them and re-run unti
 
 - Fully capitalize acronyms in symbols, e.g. `graphID`, `useBackendAPI`
 - Use function declarations (not arrow functions) for components/handlers
+- No `dark:` Tailwind classes — the design system handles dark mode
+- Use Next.js `<Link>` for internal navigation — never raw `<a>` tags
+- No `any` types unless the value genuinely can be anything
+- No linter suppressors (`// @ts-ignore`, `// eslint-disable`) — fix the actual issue
+- **File length** — keep files under ~200 lines; extract sub-components or hooks into their own files when a file grows beyond this
+- **Function/component length** — keep render functions and hooks under ~50 lines; extract named helpers or sub-components when they grow longer
 
 ## Architecture
 

autogpt_platform/frontend/src/app/(platform)/admin/layout.tsx

@@ -1,7 +1,14 @@
-import { Sidebar } from "@/components/__legacy__/Sidebar";
-import { Users, DollarSign, UserSearch, FileText } from "lucide-react";
+"use client";
 
-import { IconSliders } from "@/components/__legacy__/ui/icons";
+import { Sidebar } from "@/components/__legacy__/Sidebar";
+import {
+  UsersIcon,
+  CurrencyDollarSimpleIcon,
+  UserPlusIcon,
+  MagnifyingGlassIcon,
+  FileTextIcon,
+  SlidersHorizontalIcon,
+} from "@phosphor-icons/react";
 
 const sidebarLinkGroups = [
   {
@@ -9,27 +16,32 @@ const sidebarLinkGroups = [
       {
         text: "Marketplace Management",
         href: "/admin/marketplace",
-        icon: <Users className="h-6 w-6" />,
+        icon: <UsersIcon size={24} />,
       },
       {
         text: "User Spending",
         href: "/admin/spending",
-        icon: <DollarSign className="h-6 w-6" />,
+        icon: <CurrencyDollarSimpleIcon size={24} />,
+      },
+      {
+        text: "Beta Invites",
+        href: "/admin/users",
+        icon: <UserPlusIcon size={24} />,
       },
       {
         text: "User Impersonation",
         href: "/admin/impersonation",
-        icon: <UserSearch className="h-6 w-6" />,
+        icon: <MagnifyingGlassIcon size={24} />,
       },
       {
         text: "Execution Analytics",
         href: "/admin/execution-analytics",
-        icon: <FileText className="h-6 w-6" />,
+        icon: <FileTextIcon size={24} />,
       },
       {
         text: "Admin User Management",
         href: "/admin/settings",
-        icon: <IconSliders className="h-6 w-6" />,
+        icon: <SlidersHorizontalIcon size={24} />,
       },
     ],
   },

autogpt_platform/frontend/src/app/(platform)/admin/users/components/AdminUsersPage/AdminUsersPage.tsx

@@ -0,0 +1,80 @@
+"use client";
+
+import { Card } from "@/components/atoms/Card/Card";
+import { BulkInviteForm } from "../BulkInviteForm/BulkInviteForm";
+import { InviteUserForm } from "../InviteUserForm/InviteUserForm";
+import { InvitedUsersTable } from "../InvitedUsersTable/InvitedUsersTable";
+import { useAdminUsersPage } from "../../useAdminUsersPage";
+
+export function AdminUsersPage() {
+  const {
+    email,
+    name,
+    bulkInviteFile,
+    bulkInviteInputKey,
+    lastBulkInviteResult,
+    invitedUsers,
+    isLoadingInvitedUsers,
+    isRefreshingInvitedUsers,
+    isCreatingInvite,
+    isBulkInviting,
+    pendingInviteAction,
+    setEmail,
+    setName,
+    handleBulkInviteFileChange,
+    handleBulkInviteSubmit,
+    handleCreateInvite,
+    handleRetryTally,
+    handleRevoke,
+  } = useAdminUsersPage();
+
+  return (
+    <div className="mx-auto flex max-w-7xl flex-col gap-6 p-6">
+      <div className="flex flex-col gap-2">
+        <h1 className="text-3xl font-bold text-zinc-900">Beta Invites</h1>
+        <p className="max-w-3xl text-sm text-zinc-600">
+          Pre-provision beta users before they sign up. Invites store the
+          platform-side record, run Tally understanding extraction, and activate
+          the real account on the user&apos;s first authenticated request.
+        </p>
+      </div>
+
+      <div className="grid gap-6 xl:grid-cols-[24rem,1fr]">
+        <div className="flex flex-col gap-6">
+          <Card className="border border-zinc-200 shadow-sm">
+            <InviteUserForm
+              email={email}
+              name={name}
+              isSubmitting={isCreatingInvite}
+              onEmailChange={setEmail}
+              onNameChange={setName}
+              onSubmit={handleCreateInvite}
+            />
+          </Card>
+
+          <Card className="border border-zinc-200 shadow-sm">
+            <BulkInviteForm
+              selectedFile={bulkInviteFile}
+              inputKey={bulkInviteInputKey}
+              isSubmitting={isBulkInviting}
+              lastResult={lastBulkInviteResult}
+              onFileChange={handleBulkInviteFileChange}
+              onSubmit={handleBulkInviteSubmit}
+            />
+          </Card>
+        </div>
+
+        <Card className="border border-zinc-200 shadow-sm">
+          <InvitedUsersTable
+            invitedUsers={invitedUsers}
+            isLoading={isLoadingInvitedUsers}
+            isRefreshing={isRefreshingInvitedUsers}
+            pendingInviteAction={pendingInviteAction}
+            onRetryTally={handleRetryTally}
+            onRevoke={handleRevoke}
+          />
+        </Card>
+      </div>
+    </div>
+  );
+}

autogpt_platform/frontend/src/app/(platform)/admin/users/components/BulkInviteForm/BulkInviteForm.tsx

@@ -0,0 +1,135 @@
+"use client";
+
+import type { BulkInvitedUsersResponse } from "@/app/api/__generated__/models/bulkInvitedUsersResponse";
+import { Badge } from "@/components/atoms/Badge/Badge";
+import { Button } from "@/components/atoms/Button/Button";
+import type { FormEvent } from "react";
+
+interface Props {
+  selectedFile: File | null;
+  inputKey: number;
+  isSubmitting: boolean;
+  lastResult: BulkInvitedUsersResponse | null;
+  onFileChange: (file: File | null) => void;
+  onSubmit: (event: FormEvent<HTMLFormElement>) => void;
+}
+
+function getStatusVariant(status: "CREATED" | "SKIPPED" | "ERROR") {
+  if (status === "CREATED") {
+    return "success";
+  }
+
+  if (status === "ERROR") {
+    return "error";
+  }
+
+  return "info";
+}
+
+export function BulkInviteForm({
+  selectedFile,
+  inputKey,
+  isSubmitting,
+  lastResult,
+  onFileChange,
+  onSubmit,
+}: Props) {
+  return (
+    <form className="flex flex-col gap-4" onSubmit={onSubmit}>
+      <div className="flex flex-col gap-1">
+        <h2 className="text-xl font-semibold text-zinc-900">Bulk invite</h2>
+        <p className="text-sm text-zinc-600">
+          Upload a <span className="font-medium text-zinc-800">.txt</span> file
+          with one email per line, or a{" "}
+          <span className="font-medium text-zinc-800">.csv</span> with
+          <span className="font-medium text-zinc-800"> email</span> and optional
+          <span className="font-medium text-zinc-800"> name</span> columns.
+        </p>
+      </div>
+
+      <label
+        htmlFor="bulk-invite-file-input"
+        className="flex cursor-pointer flex-col gap-2 rounded-2xl border border-dashed border-zinc-300 bg-zinc-50 px-4 py-5 text-sm text-zinc-600 transition-colors focus-within:ring-2 focus-within:ring-zinc-500 focus-within:ring-offset-2 hover:border-zinc-400 hover:bg-zinc-100"
+      >
+        <span className="font-medium text-zinc-900">
+          {selectedFile ? selectedFile.name : "Choose invite file"}
+        </span>
+        <span>Maximum 500 rows, UTF-8 encoded.</span>
+        <input
+          id="bulk-invite-file-input"
+          key={inputKey}
+          type="file"
+          accept=".txt,.csv,text/plain,text/csv"
+          disabled={isSubmitting}
+          className="sr-only"
+          onChange={(event) =>
+            onFileChange(event.target.files?.item(0) ?? null)
+          }
+        />
+      </label>
+
+      <Button
+        type="submit"
+        variant="primary"
+        loading={isSubmitting}
+        disabled={!selectedFile}
+        className="w-full"
+      >
+        {isSubmitting ? "Uploading invites..." : "Upload invite file"}
+      </Button>
+
+      {lastResult ? (
+        <div className="flex flex-col gap-3 rounded-2xl border border-zinc-200 bg-zinc-50 p-4">
+          <div className="grid grid-cols-3 gap-2 text-center">
+            <div className="rounded-xl bg-white px-3 py-2">
+              <div className="text-lg font-semibold text-zinc-900">
+                {lastResult.created_count}
+              </div>
+              <div className="text-xs uppercase tracking-[0.16em] text-zinc-500">
+                Created
+              </div>
+            </div>
+            <div className="rounded-xl bg-white px-3 py-2">
+              <div className="text-lg font-semibold text-zinc-900">
+                {lastResult.skipped_count}
+              </div>
+              <div className="text-xs uppercase tracking-[0.16em] text-zinc-500">
+                Skipped
+              </div>
+            </div>
+            <div className="rounded-xl bg-white px-3 py-2">
+              <div className="text-lg font-semibold text-zinc-900">
+                {lastResult.error_count}
+              </div>
+              <div className="text-xs uppercase tracking-[0.16em] text-zinc-500">
+                Errors
+              </div>
+            </div>
+          </div>
+
+          <div className="max-h-64 overflow-y-auto rounded-xl border border-zinc-200 bg-white">
+            <div className="flex flex-col divide-y divide-zinc-100">
+              {lastResult.results.map((row) => (
+                <div
+                  key={`${row.row_number}-${row.email ?? row.message}`}
+                  className="flex items-start gap-3 px-3 py-3"
+                >
+                  <Badge variant={getStatusVariant(row.status)} size="small">
+                    {row.status}
+                  </Badge>
+                  <div className="flex min-w-0 flex-1 flex-col gap-1">
+                    <span className="text-sm font-medium text-zinc-900">
+                      Row {row.row_number}
+                      {row.email ? ` · ${row.email}` : ""}
+                    </span>
+                    <span className="text-xs text-zinc-500">{row.message}</span>
+                  </div>
+                </div>
+              ))}
+            </div>
+          </div>
+        </div>
+      ) : null}
+    </form>
+  );
+}

autogpt_platform/frontend/src/app/(platform)/admin/users/components/InviteUserForm/InviteUserForm.tsx

@@ -0,0 +1,66 @@
+"use client";
+
+import { Button } from "@/components/atoms/Button/Button";
+import { Input } from "@/components/atoms/Input/Input";
+import type { FormEvent } from "react";
+
+interface Props {
+  email: string;
+  name: string;
+  isSubmitting: boolean;
+  onEmailChange: (value: string) => void;
+  onNameChange: (value: string) => void;
+  onSubmit: (event: FormEvent<HTMLFormElement>) => void;
+}
+
+export function InviteUserForm({
+  email,
+  name,
+  isSubmitting,
+  onEmailChange,
+  onNameChange,
+  onSubmit,
+}: Props) {
+  return (
+    <form className="flex flex-col gap-4" onSubmit={onSubmit}>
+      <div className="flex flex-col gap-1">
+        <h2 className="text-xl font-semibold text-zinc-900">Create invite</h2>
+        <p className="text-sm text-zinc-600">
+          The invite is stored immediately, then Tally pre-seeding starts in the
+          background.
+        </p>
+      </div>
+
+      <Input
+        id="invite-email"
+        label="Email"
+        type="email"
+        value={email}
+        placeholder="jane@example.com"
+        autoComplete="email"
+        disabled={isSubmitting}
+        onChange={(event) => onEmailChange(event.target.value)}
+      />
+
+      <Input
+        id="invite-name"
+        label="Name"
+        type="text"
+        value={name}
+        placeholder="Jane Doe"
+        disabled={isSubmitting}
+        onChange={(event) => onNameChange(event.target.value)}
+      />
+
+      <Button
+        type="submit"
+        variant="primary"
+        loading={isSubmitting}
+        disabled={!email.trim()}
+        className="w-full"
+      >
+        {isSubmitting ? "Creating invite..." : "Create invite"}
+      </Button>
+    </form>
+  );
+}

autogpt_platform/frontend/src/app/(platform)/admin/users/components/InvitedUsersTable/InvitedUsersTable.tsx

@@ -0,0 +1,209 @@
+"use client";
+
+import type { InvitedUserResponse } from "@/app/api/__generated__/models/invitedUserResponse";
+import { Badge } from "@/components/atoms/Badge/Badge";
+import { Button } from "@/components/atoms/Button/Button";
+import {
+  Table,
+  TableBody,
+  TableCell,
+  TableHead,
+  TableHeader,
+  TableRow,
+} from "@/components/__legacy__/ui/table";
+
+interface Props {
+  invitedUsers: InvitedUserResponse[];
+  isLoading: boolean;
+  isRefreshing: boolean;
+  pendingInviteAction: string | null;
+  onRetryTally: (invitedUserId: string) => void;
+  onRevoke: (invitedUserId: string) => void;
+}
+
+function getInviteBadgeVariant(status: InvitedUserResponse["status"]) {
+  if (status === "CLAIMED") {
+    return "success";
+  }
+
+  if (status === "REVOKED") {
+    return "error";
+  }
+
+  return "info";
+}
+
+function getTallyBadgeVariant(status: InvitedUserResponse["tally_status"]) {
+  if (status === "READY") {
+    return "success";
+  }
+
+  if (status === "FAILED") {
+    return "error";
+  }
+
+  return "info";
+}
+
+function formatDate(value: Date | undefined) {
+  if (!value) {
+    return "-";
+  }
+
+  return value.toLocaleString();
+}
+
+function getTallySummary(invitedUser: InvitedUserResponse) {
+  if (invitedUser.tally_status === "FAILED" && invitedUser.tally_error) {
+    return invitedUser.tally_error;
+  }
+
+  if (invitedUser.tally_status === "READY" && invitedUser.tally_understanding) {
+    return "Stored and ready for activation";
+  }
+
+  if (invitedUser.tally_status === "READY") {
+    return "No matching Tally submission found";
+  }
+
+  if (invitedUser.tally_status === "RUNNING") {
+    return "Extraction in progress";
+  }
+
+  return "Waiting to run";
+}
+
+function isActionPending(
+  pendingInviteAction: string | null,
+  action: "retry" | "revoke",
+  invitedUserId: string,
+) {
+  return pendingInviteAction === `${action}:${invitedUserId}`;
+}
+
+export function InvitedUsersTable({
+  invitedUsers,
+  isLoading,
+  isRefreshing,
+  pendingInviteAction,
+  onRetryTally,
+  onRevoke,
+}: Props) {
+  return (
+    <div className="flex flex-col gap-4">
+      <div className="flex items-center justify-between gap-4">
+        <div className="flex flex-col gap-1">
+          <h2 className="text-xl font-semibold text-zinc-900">Invited users</h2>
+          <p className="text-sm text-zinc-600">
+            Live invite state, claim status, and Tally pre-seeding progress.
+          </p>
+        </div>
+        <span className="text-xs uppercase tracking-[0.18em] text-zinc-400">
+          {isRefreshing ? "Refreshing" : `${invitedUsers.length} total`}
+        </span>
+      </div>
+
+      <div className="overflow-hidden rounded-2xl border border-zinc-200">
+        <Table>
+          <TableHeader className="bg-zinc-50">
+            <TableRow>
+              <TableHead>Email</TableHead>
+              <TableHead>Name</TableHead>
+              <TableHead>Invite</TableHead>
+              <TableHead>Tally</TableHead>
+              <TableHead>Claimed User</TableHead>
+              <TableHead>Created</TableHead>
+              <TableHead className="text-right">Actions</TableHead>
+            </TableRow>
+          </TableHeader>
+          <TableBody>
+            {isLoading ? (
+              <TableRow>
+                <TableCell
+                  colSpan={7}
+                  className="py-10 text-center text-zinc-500"
+                >
+                  Loading invited users...
+                </TableCell>
+              </TableRow>
+            ) : invitedUsers.length === 0 ? (
+              <TableRow>
+                <TableCell
+                  colSpan={7}
+                  className="py-10 text-center text-zinc-500"
+                >
+                  No invited users yet
+                </TableCell>
+              </TableRow>
+            ) : (
+              invitedUsers.map((invitedUser) => (
+                <TableRow key={invitedUser.id} className="align-top">
+                  <TableCell className="font-medium text-zinc-900">
+                    {invitedUser.email}
+                  </TableCell>
+                  <TableCell>{invitedUser.name || "-"}</TableCell>
+                  <TableCell>
+                    <Badge variant={getInviteBadgeVariant(invitedUser.status)}>
+                      {invitedUser.status}
+                    </Badge>
+                  </TableCell>
+                  <TableCell>
+                    <div className="flex max-w-xs flex-col gap-2">
+                      <Badge
+                        variant={getTallyBadgeVariant(invitedUser.tally_status)}
+                      >
+                        {invitedUser.tally_status}
+                      </Badge>
+                      <span className="text-xs text-zinc-500">
+                        {getTallySummary(invitedUser)}
+                      </span>
+                      <span className="text-xs text-zinc-400">
+                        {formatDate(invitedUser.tally_computed_at ?? undefined)}
+                      </span>
+                    </div>
+                  </TableCell>
+                  <TableCell className="font-mono text-xs text-zinc-500">
+                    {invitedUser.auth_user_id || "-"}
+                  </TableCell>
+                  <TableCell className="text-sm text-zinc-500">
+                    {formatDate(invitedUser.created_at)}
+                  </TableCell>
+                  <TableCell>
+                    <div className="flex justify-end gap-2">
+                      <Button
+                        variant="outline"
+                        size="small"
+                        disabled={invitedUser.status === "REVOKED"}
+                        loading={isActionPending(
+                          pendingInviteAction,
+                          "retry",
+                          invitedUser.id,
+                        )}
+                        onClick={() => onRetryTally(invitedUser.id)}
+                      >
+                        Retry Tally
+                      </Button>
+                      <Button
+                        variant="secondary"
+                        size="small"
+                        disabled={invitedUser.status !== "INVITED"}
+                        loading={isActionPending(
+                          pendingInviteAction,
+                          "revoke",
+                          invitedUser.id,
+                        )}
+                        onClick={() => onRevoke(invitedUser.id)}
+                      >
+                        Revoke
+                      </Button>
+                    </div>
+                  </TableCell>
+                </TableRow>
+              ))
+            )}
+          </TableBody>
+        </Table>
+      </div>
+    </div>
+  );
+}

autogpt_platform/frontend/src/app/(platform)/admin/users/page.tsx

@@ -1,16 +1,11 @@
 import { withRoleAccess } from "@/lib/withRoleAccess";
-import React from "react";
+import { AdminUsersPage } from "./components/AdminUsersPage/AdminUsersPage";
 
 function AdminUsers() {
-  return (
-    <div>
-      <h1>Users Dashboard</h1>
-      {/* Add your admin-only content here */}
-    </div>
-  );
+  return <AdminUsersPage />;
 }
 
-export default async function AdminUsersPage() {
+export default async function AdminUsersRoute() {
   "use server";
   const withAdminAccess = await withRoleAccess(["admin"]);
   const ProtectedAdminUsers = await withAdminAccess(AdminUsers);

autogpt_platform/frontend/src/app/(platform)/admin/users/useAdminUsersPage.ts

@@ -0,0 +1,197 @@
+"use client";
+
+import type { BulkInvitedUsersResponse } from "@/app/api/__generated__/models/bulkInvitedUsersResponse";
+import { okData } from "@/app/api/helpers";
+import {
+  getGetV2ListInvitedUsersQueryKey,
+  useGetV2ListInvitedUsers,
+  usePostV2BulkCreateInvitedUsers,
+  usePostV2CreateInvitedUser,
+  usePostV2RetryInvitedUserTally,
+  usePostV2RevokeInvitedUser,
+} from "@/app/api/__generated__/endpoints/admin/admin";
+import { useToast } from "@/components/molecules/Toast/use-toast";
+import { useQueryClient } from "@tanstack/react-query";
+import { type FormEvent, useState } from "react";
+
+function getErrorMessage(error: unknown) {
+  if (error instanceof Error) {
+    return error.message;
+  }
+
+  return "Something went wrong";
+}
+
+export function useAdminUsersPage() {
+  const queryClient = useQueryClient();
+  const { toast } = useToast();
+  const [email, setEmail] = useState("");
+  const [name, setName] = useState("");
+  const [bulkInviteFile, setBulkInviteFile] = useState<File | null>(null);
+  const [bulkInviteInputKey, setBulkInviteInputKey] = useState(0);
+  const [lastBulkInviteResult, setLastBulkInviteResult] =
+    useState<BulkInvitedUsersResponse | null>(null);
+  const [pendingInviteAction, setPendingInviteAction] = useState<string | null>(
+    null,
+  );
+
+  const invitedUsersQuery = useGetV2ListInvitedUsers(undefined, {
+    query: {
+      select: okData,
+      refetchInterval: 30_000,
+    },
+  });
+
+  const createInvitedUserMutation = usePostV2CreateInvitedUser({
+    mutation: {
+      onSuccess: async () => {
+        setEmail("");
+        setName("");
+        await queryClient.invalidateQueries({
+          queryKey: getGetV2ListInvitedUsersQueryKey(),
+        });
+        toast({
+          title: "Invited user created",
+          variant: "default",
+        });
+      },
+      onError: (error) => {
+        toast({
+          title: getErrorMessage(error),
+          variant: "destructive",
+        });
+      },
+    },
+  });
+
+  const bulkCreateInvitedUsersMutation = usePostV2BulkCreateInvitedUsers({
+    mutation: {
+      onSuccess: async (response) => {
+        const result = okData(response) ?? null;
+        setBulkInviteFile(null);
+        setBulkInviteInputKey((currentValue) => currentValue + 1);
+        setLastBulkInviteResult(result);
+        await queryClient.invalidateQueries({
+          queryKey: getGetV2ListInvitedUsersQueryKey(),
+        });
+        toast({
+          title: result
+            ? `${result.created_count} invites created`
+            : "Bulk invite upload complete",
+          variant: "default",
+        });
+      },
+      onError: (error) => {
+        toast({
+          title: getErrorMessage(error),
+          variant: "destructive",
+        });
+      },
+    },
+  });
+
+  const retryInvitedUserTallyMutation = usePostV2RetryInvitedUserTally({
+    mutation: {
+      onSuccess: async () => {
+        setPendingInviteAction(null);
+        await queryClient.invalidateQueries({
+          queryKey: getGetV2ListInvitedUsersQueryKey(),
+        });
+        toast({
+          title: "Tally pre-seeding restarted",
+          variant: "default",
+        });
+      },
+      onError: (error) => {
+        setPendingInviteAction(null);
+        toast({
+          title: getErrorMessage(error),
+          variant: "destructive",
+        });
+      },
+    },
+  });
+
+  const revokeInvitedUserMutation = usePostV2RevokeInvitedUser({
+    mutation: {
+      onSuccess: async () => {
+        setPendingInviteAction(null);
+        await queryClient.invalidateQueries({
+          queryKey: getGetV2ListInvitedUsersQueryKey(),
+        });
+        toast({
+          title: "Invite revoked",
+          variant: "default",
+        });
+      },
+      onError: (error) => {
+        setPendingInviteAction(null);
+        toast({
+          title: getErrorMessage(error),
+          variant: "destructive",
+        });
+      },
+    },
+  });
+
+  function handleCreateInvite(event: FormEvent<HTMLFormElement>) {
+    event.preventDefault();
+
+    createInvitedUserMutation.mutate({
+      data: {
+        email,
+        name: name.trim() || null,
+      },
+    });
+  }
+
+  function handleRetryTally(invitedUserId: string) {
+    setPendingInviteAction(`retry:${invitedUserId}`);
+    retryInvitedUserTallyMutation.mutate({ invitedUserId });
+  }
+
+  function handleBulkInviteFileChange(file: File | null) {
+    setBulkInviteFile(file);
+  }
+
+  function handleBulkInviteSubmit(event: FormEvent<HTMLFormElement>) {
+    event.preventDefault();
+
+    if (!bulkInviteFile) {
+      return;
+    }
+
+    bulkCreateInvitedUsersMutation.mutate({
+      data: {
+        file: bulkInviteFile,
+      },
+    });
+  }
+
+  function handleRevoke(invitedUserId: string) {
+    setPendingInviteAction(`revoke:${invitedUserId}`);
+    revokeInvitedUserMutation.mutate({ invitedUserId });
+  }
+
+  return {
+    email,
+    name,
+    bulkInviteFile,
+    bulkInviteInputKey,
+    lastBulkInviteResult,
+    invitedUsers: invitedUsersQuery.data?.invited_users ?? [],
+    invitedUsersError: invitedUsersQuery.error,
+    isLoadingInvitedUsers: invitedUsersQuery.isLoading,
+    isRefreshingInvitedUsers: invitedUsersQuery.isFetching,
+    isCreatingInvite: createInvitedUserMutation.isPending,
+    isBulkInviting: bulkCreateInvitedUsersMutation.isPending,
+    pendingInviteAction,
+    setEmail,
+    setName,
+    handleBulkInviteFileChange,
+    handleBulkInviteSubmit,
+    handleCreateInvite,
+    handleRetryTally,
+    handleRevoke,
+  };
+}

autogpt_platform/frontend/src/app/(platform)/copilot/components/ChatInput/useChatInput.ts

@@ -1,4 +1,3 @@
-import { useCopilotUIStore } from "@/app/(platform)/copilot/store";
 import { ChangeEvent, FormEvent, useEffect, useState } from "react";
 
 interface Args {
@@ -17,16 +16,6 @@ export function useChatInput({
 }: Args) {
   const [value, setValue] = useState("");
   const [isSending, setIsSending] = useState(false);
-  const { initialPrompt, setInitialPrompt } = useCopilotUIStore();
-
-  useEffect(
-    function consumeInitialPrompt() {
-      if (!initialPrompt) return;
-      setValue((prev) => (prev.length === 0 ? initialPrompt : prev));
-      setInitialPrompt(null);
-    },
-    [initialPrompt, setInitialPrompt],
-  );
 
   useEffect(
     function focusOnMount() {

autogpt_platform/frontend/src/app/(platform)/copilot/components/EmptySession/EmptySession.tsx

@@ -1,7 +1,9 @@
 "use client";
 
+import { useGetV2GetSuggestedPrompts } from "@/app/api/__generated__/endpoints/chat/chat";
 import { ChatInput } from "@/app/(platform)/copilot/components/ChatInput/ChatInput";
 import { Button } from "@/components/atoms/Button/Button";
+import { Skeleton } from "@/components/atoms/Skeleton/Skeleton";
 import { Text } from "@/components/atoms/Text/Text";
 import { useSupabase } from "@/lib/supabase/hooks/useSupabase";
 import { SpinnerGapIcon } from "@phosphor-icons/react";
@@ -33,15 +35,38 @@ export function EmptySession({
 }: Props) {
   const { user } = useSupabase();
   const greetingName = getGreetingName(user);
-  const quickActions = getQuickActions();
+
+  const { data: suggestedPromptsResponse, isLoading: isLoadingPrompts } =
+    useGetV2GetSuggestedPrompts({
+      query: { staleTime: Infinity },
+    });
+  const customPrompts =
+    suggestedPromptsResponse?.status === 200
+      ? suggestedPromptsResponse.data.prompts
+      : undefined;
+  const quickActions = getQuickActions(customPrompts);
   const [loadingAction, setLoadingAction] = useState<string | null>(null);
   const [inputPlaceholder, setInputPlaceholder] = useState(
     getInputPlaceholder(),
   );
 
+  // Use matchMedia instead of resize event — fires only when crossing
+  // the 500px and 1081px breakpoints defined in getInputPlaceholder(),
+  // rather than dozens of times per second during a window drag.
   useEffect(() => {
-    setInputPlaceholder(getInputPlaceholder(window.innerWidth));
-  }, [window.innerWidth]);
+    function update() {
+      setInputPlaceholder(getInputPlaceholder(window.innerWidth));
+    }
+    const mq500 = window.matchMedia("(min-width: 500px)");
+    const mq1081 = window.matchMedia("(min-width: 1081px)");
+    update();
+    mq500.addEventListener("change", update);
+    mq1081.addEventListener("change", update);
+    return () => {
+      mq500.removeEventListener("change", update);
+      mq1081.removeEventListener("change", update);
+    };
+  }, []);
 
   async function handleQuickActionClick(action: string) {
     if (isCreatingSession || loadingAction) return;
@@ -91,28 +116,32 @@ export function EmptySession({
         </div>
 
         <div className="flex flex-wrap items-center justify-center gap-3 overflow-x-auto [-ms-overflow-style:none] [scrollbar-width:none] [&::-webkit-scrollbar]:hidden">
-          {quickActions.map((action) => (
-            <Button
-              key={action}
-              type="button"
-              variant="outline"
-              size="small"
-              onClick={() => void handleQuickActionClick(action)}
-              disabled={isCreatingSession || loadingAction !== null}
-              aria-busy={loadingAction === action}
-              leftIcon={
-                loadingAction === action ? (
-                  <SpinnerGapIcon
-                    className="h-4 w-4 animate-spin"
-                    weight="bold"
-                  />
-                ) : null
-              }
-              className="h-auto shrink-0 border-zinc-300 px-3 py-2 text-[.9rem] text-zinc-600"
-            >
-              {action}
-            </Button>
-          ))}
+          {isLoadingPrompts
+            ? Array.from({ length: 3 }, (_, i) => (
+                <Skeleton key={i} className="h-10 w-64 shrink-0 rounded-full" />
+              ))
+            : quickActions.map((action) => (
+                <Button
+                  key={action}
+                  type="button"
+                  variant="outline"
+                  size="small"
+                  onClick={() => void handleQuickActionClick(action)}
+                  disabled={isCreatingSession || loadingAction !== null}
+                  aria-busy={loadingAction === action}
+                  leftIcon={
+                    loadingAction === action ? (
+                      <SpinnerGapIcon
+                        className="h-4 w-4 animate-spin"
+                        weight="bold"
+                      />
+                    ) : null
+                  }
+                  className="h-auto shrink-0 border-zinc-300 px-3 py-2 text-[.9rem] text-zinc-600"
+                >
+                  {action}
+                </Button>
+              ))}
         </div>
       </motion.div>
     </div>

autogpt_platform/frontend/src/app/(platform)/copilot/components/EmptySession/helpers.ts

@@ -12,12 +12,17 @@ export function getInputPlaceholder(width?: number) {
   return "What's your role and what eats up most of your day? e.g. 'I'm a recruiter and I hate...'";
 }
 
-export function getQuickActions() {
-  return [
-    "I don't know where to start, just ask me stuff",
-    "I do the same thing every week and it's killing me",
-    "Help me find where I'm wasting my time",
-  ];
+const DEFAULT_QUICK_ACTIONS = [
+  "I don't know where to start, just ask me stuff",
+  "I do the same thing every week and it's killing me",
+  "Help me find where I'm wasting my time",
+];
+
+export function getQuickActions(customPrompts?: string[]) {
+  if (customPrompts && customPrompts.length > 0) {
+    return customPrompts;
+  }
+  return DEFAULT_QUICK_ACTIONS;
 }
 
 export function getGreetingName(user?: User | null) {

autogpt_platform/frontend/src/app/(platform)/copilot/components/JobStatsBar/useWorkDoneCounters.ts

@@ -5,7 +5,7 @@ const TOOL_TO_CATEGORY: Record<string, string> = {
   find_agent: "search",
   find_library_agent: "search",
   run_agent: "agent run",
-  run_block: "block run",
+  run_block: "action",
   create_agent: "agent created",
   edit_agent: "agent edited",
   schedule_agent: "agent scheduled",

autogpt_platform/frontend/src/app/(platform)/copilot/store.ts

@@ -7,10 +7,6 @@ export interface DeleteTarget {
 }
 
 interface CopilotUIState {
-  /** Prompt extracted from URL hash (e.g. /copilot#prompt=...) for input prefill. */
-  initialPrompt: string | null;
-  setInitialPrompt: (prompt: string | null) => void;
-
   sessionToDelete: DeleteTarget | null;
   setSessionToDelete: (target: DeleteTarget | null) => void;
 
@@ -35,9 +31,6 @@ interface CopilotUIState {
 }
 
 export const useCopilotUIStore = create<CopilotUIState>((set) => ({
-  initialPrompt: null,
-  setInitialPrompt: (prompt) => set({ initialPrompt: prompt }),
-
   sessionToDelete: null,
   setSessionToDelete: (target) => set({ sessionToDelete: target }),
 

autogpt_platform/frontend/src/app/(platform)/copilot/styleguide/page.tsx

@@ -706,8 +706,8 @@ export default function StyleguidePage() {
                     input: { block_id: "weather-block-123" },
                     output: {
                       type: ResponseType.error,
-                      message: "Failed to run the block.",
-                      error: "Block execution timed out after 30 seconds.",
+                      message: "Something went wrong while running this step.",
+                      error: "Execution timed out after 30 seconds.",
                       details: {
                         block_id: "weather-block-123",
                         timeout_ms: 30000,

autogpt_platform/frontend/src/app/(platform)/copilot/tools/FindBlocks/FindBlocks.tsx

@@ -61,7 +61,7 @@ export function FindBlocksTool({ part }: Props) {
 
   const query = (part.input as FindBlockInput | undefined)?.query?.trim();
   const accordionDescription = parsed
-    ? `Found ${parsed.count} block${parsed.count === 1 ? "" : "s"}${query ? ` for "${query}"` : ""}`
+    ? `Found ${parsed.count} action${parsed.count === 1 ? "" : "s"}${query ? ` for "${query}"` : ""}`
     : undefined;
 
   return (
@@ -77,7 +77,7 @@ export function FindBlocksTool({ part }: Props) {
       {hasBlocks && parsed && (
         <ToolAccordion
           icon={<AccordionIcon />}
-          title="Block results"
+          title="Results"
           description={accordionDescription}
         >
           <HorizontalScroll dependencyList={[parsed.blocks.length]}>

autogpt_platform/frontend/src/app/(platform)/copilot/tools/FindBlocks/helpers.tsx

@@ -30,21 +30,21 @@ export function getAnimationText(part: FindBlockToolPart): string {
   switch (part.state) {
     case "input-streaming":
     case "input-available":
-      return `Searching for blocks${queryText}`;
+      return `Searching for actions${queryText}`;
 
     case "output-available": {
       const parsed = parseOutput(part.output);
       if (parsed) {
-        return `Found ${parsed.count} block${parsed.count === 1 ? "" : "s"}${queryText}`;
+        return `Found ${parsed.count} action${parsed.count === 1 ? "" : "s"}${queryText}`;
       }
-      return `Searching for blocks${queryText}`;
+      return `Searching for actions${queryText}`;
     }
 
     case "output-error":
-      return `Error finding blocks${queryText}`;
+      return `Search failed${query ? ` for "${query}"` : ""}`;
 
     default:
-      return "Searching for blocks";
+      return "Searching for actions";
   }
 }