fix poetry lock

- Added Langfuse configuration to settings and environment variables.
- Introduced Langfuse client for fetching prompts in chat service.
- Updated ChatConfig to include Langfuse prompt name.
- Enhanced service logic to retrieve prompts from Langfuse, improving prompt management capabilities.

This integration allows for rapid runtime prompt updates and eventual analytics of the performance of the Co-Pilot system.

This commit aims to trial this service as a potential option.

autogpt_platform/.claude/settings.local.json

@@ -0,0 +1,9 @@
+{
+  "permissions": {
+    "allow": [
+      "Bash(ls:*)",
+      "WebFetch(domain:langfuse.com)",
+      "Bash(poetry install:*)"
+    ]
+  }
+}

autogpt_platform/Makefile

@@ -1,4 +1,4 @@
-.PHONY: start-core stop-core logs-core format lint migrate run-backend run-frontend load-store-agents
+.PHONY: start-core stop-core logs-core format lint migrate run-backend stop-backend run-frontend load-store-agents backfill-store-embeddings
 
 # Run just Supabase + Redis + RabbitMQ
 start-core:
@@ -34,7 +34,14 @@ migrate:
 	cd backend && poetry run prisma migrate deploy
 	cd backend && poetry run prisma generate
 
-run-backend:
+stop-backend:
+	@echo "Stopping backend processes..."
+	@cd backend && poetry run cli stop 2>/dev/null || true
+	@echo "Killing any processes using backend ports..."
+	@lsof -ti:8001,8002,8003,8004,8005,8006,8007 | xargs kill -9 2>/dev/null || true
+	@echo "Backend stopped"
+
+run-backend: stop-backend
 	cd backend && poetry run app
 
 run-frontend:
@@ -46,6 +53,9 @@ test-data:
 load-store-agents:
 	cd backend && poetry run load-store-agents
 
+backfill-store-embeddings:
+	cd backend && poetry run python -m backend.api.features.store.backfill_embeddings
+
 help:
 	@echo "Usage: make <target>"
 	@echo "Targets:"
@@ -55,7 +65,9 @@ help:
 	@echo "  logs-core - Tail the logs for core services"
 	@echo "  format - Format & lint backend (Python) and frontend (TypeScript) code"
 	@echo "  migrate - Run backend database migrations"
-	@echo "  run-backend - Run the backend FastAPI server"
+	@echo "  stop-backend - Stop any running backend processes"
+	@echo "  run-backend - Run the backend FastAPI server (stops existing processes first)"
 	@echo "  run-frontend - Run the frontend Next.js development server"
 	@echo "  test-data - Run the test data creator"
-	@echo "  load-store-agents - Load store agents from agents/ folder into test database"
+	@echo "  load-store-agents - Load store agents from agents/ folder into test database"
+	@echo "  backfill-store-embeddings - Generate embeddings for store agents that don't have them"

autogpt_platform/backend/.env.default

@@ -58,6 +58,13 @@ V0_API_KEY=
 OPEN_ROUTER_API_KEY=
 NVIDIA_API_KEY=
 
+# Langfuse Prompt Management
+# Used for managing the CoPilot system prompt externally
+# Get credentials from https://cloud.langfuse.com or your self-hosted instance
+LANGFUSE_PUBLIC_KEY=
+LANGFUSE_SECRET_KEY=
+LANGFUSE_HOST=https://cloud.langfuse.com
+
 # OAuth Credentials
 # For the OAuth callback URL, use <your_frontend_url>/auth/integrations/oauth_callback,
 # e.g. http://localhost:3000/auth/integrations/oauth_callback

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

@@ -9,6 +9,7 @@ import prisma.enums
 
 import backend.api.features.store.cache as store_cache
 import backend.api.features.store.db as store_db
+import backend.api.features.store.embeddings as store_embeddings
 import backend.api.features.store.model as store_model
 import backend.util.json
 
@@ -150,3 +151,54 @@ async def admin_download_agent_file(
         return fastapi.responses.FileResponse(
             tmp_file.name, filename=file_name, media_type="application/json"
         )
+
+
+@router.get(
+    "/embeddings/stats",
+    summary="Get Embedding Statistics",
+)
+async def get_embedding_stats() -> dict[str, typing.Any]:
+    """
+    Get statistics about embedding coverage for store listings.
+
+    Returns counts of total approved listings, listings with embeddings,
+    listings without embeddings, and coverage percentage.
+    """
+    try:
+        stats = await store_embeddings.get_embedding_stats()
+        return stats
+    except Exception as e:
+        logger.exception("Error getting embedding stats: %s", e)
+        raise fastapi.HTTPException(
+            status_code=500,
+            detail="An error occurred while retrieving embedding stats",
+        )
+
+
+@router.post(
+    "/embeddings/backfill",
+    summary="Backfill Missing Embeddings",
+)
+async def backfill_embeddings(
+    batch_size: int = 10,
+) -> dict[str, typing.Any]:
+    """
+    Trigger backfill of embeddings for approved listings that don't have them.
+
+    Args:
+        batch_size: Number of embeddings to generate in one call (default 10)
+
+    Returns:
+        Dict with processed count, success count, failure count, and message
+    """
+    try:
+        result = await store_embeddings.backfill_missing_embeddings(
+            batch_size=batch_size
+        )
+        return result
+    except Exception as e:
+        logger.exception("Error backfilling embeddings: %s", e)
+        raise fastapi.HTTPException(
+            status_code=500,
+            detail="An error occurred while backfilling embeddings",
+        )

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

@@ -12,7 +12,11 @@ class ChatConfig(BaseSettings):
 
     # OpenAI API Configuration
     model: str = Field(
-        default="qwen/qwen3-235b-a22b-2507", description="Default model to use"
+        default="anthropic/claude-opus-4.5", description="Default model to use"
+    )
+    title_model: str = Field(
+        default="openai/gpt-4o-mini",
+        description="Model to use for generating session titles (should be fast/cheap)",
     )
     api_key: str | None = Field(default=None, description="OpenAI API key")
     base_url: str | None = Field(
@@ -41,6 +45,13 @@ class ChatConfig(BaseSettings):
         default=3, description="Maximum number of agent schedules"
     )
 
+    # Langfuse Prompt Management Configuration
+    # Note: Langfuse credentials are in Settings().secrets (settings.py)
+    langfuse_prompt_name: str = Field(
+        default="CoPilot Prompt",
+        description="Name of the prompt in Langfuse to fetch",
+    )
+
     @field_validator("api_key", mode="before")
     @classmethod
     def get_api_key(cls, v):
@@ -72,8 +83,31 @@ class ChatConfig(BaseSettings):
                 v = "https://openrouter.ai/api/v1"
         return v
 
+    # Prompt paths for different contexts
+    PROMPT_PATHS: dict[str, str] = {
+        "default": "prompts/chat_system.md",
+        "onboarding": "prompts/onboarding_system.md",
+    }
+
+    def get_system_prompt_for_type(
+        self, prompt_type: str = "default", **template_vars
+    ) -> str:
+        """Load and render a system prompt by type.
+
+        Args:
+            prompt_type: The type of prompt to load ("default" or "onboarding")
+            **template_vars: Variables to substitute in the template
+
+        Returns:
+            Rendered system prompt string
+        """
+        prompt_path_str = self.PROMPT_PATHS.get(
+            prompt_type, self.PROMPT_PATHS["default"]
+        )
+        return self._load_prompt_from_path(prompt_path_str, **template_vars)
+
     def get_system_prompt(self, **template_vars) -> str:
-        """Load and render the system prompt from file.
+        """Load and render the default system prompt from file.
 
         Args:
             **template_vars: Variables to substitute in the template
@@ -82,9 +116,21 @@ class ChatConfig(BaseSettings):
             Rendered system prompt string
 
         """
+        return self._load_prompt_from_path(self.system_prompt_path, **template_vars)
+
+    def _load_prompt_from_path(self, prompt_path_str: str, **template_vars) -> str:
+        """Load and render a system prompt from a given path.
+
+        Args:
+            prompt_path_str: Path to the prompt file relative to chat module
+            **template_vars: Variables to substitute in the template
+
+        Returns:
+            Rendered system prompt string
+        """
         # Get the path relative to this module
         module_dir = Path(__file__).parent
-        prompt_path = module_dir / self.system_prompt_path
+        prompt_path = module_dir / prompt_path_str
 
         # Check for .j2 extension first (Jinja2 template)
         j2_path = Path(str(prompt_path) + ".j2")

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

@@ -0,0 +1,195 @@
+"""Database operations for chat sessions."""
+
+import logging
+from datetime import UTC, datetime
+from typing import Any
+
+from prisma.models import ChatMessage as PrismaChatMessage
+from prisma.models import ChatSession as PrismaChatSession
+from prisma.types import ChatSessionUpdateInput
+
+from backend.util.json import SafeJson
+
+logger = logging.getLogger(__name__)
+
+
+async def get_chat_session(session_id: str) -> PrismaChatSession | None:
+    """Get a chat session by ID from the database."""
+    session = await PrismaChatSession.prisma().find_unique(
+        where={"id": session_id},
+        include={"Messages": True},
+    )
+    if session and session.Messages:
+        # Sort messages by sequence in Python since Prisma doesn't support order_by in include
+        session.Messages.sort(key=lambda m: m.sequence)
+    return session
+
+
+async def create_chat_session(
+    session_id: str,
+    user_id: str | None,
+) -> PrismaChatSession:
+    """Create a new chat session in the database."""
+    data = {
+        "id": session_id,
+        "userId": user_id,
+        "credentials": SafeJson({}),
+        "successfulAgentRuns": SafeJson({}),
+        "successfulAgentSchedules": SafeJson({}),
+    }
+    return await PrismaChatSession.prisma().create(
+        data=data,
+        include={"Messages": True},
+    )
+
+
+async def update_chat_session(
+    session_id: str,
+    credentials: dict[str, Any] | None = None,
+    successful_agent_runs: dict[str, Any] | None = None,
+    successful_agent_schedules: dict[str, Any] | None = None,
+    total_prompt_tokens: int | None = None,
+    total_completion_tokens: int | None = None,
+    title: str | None = None,
+) -> PrismaChatSession | None:
+    """Update a chat session's metadata."""
+    data: ChatSessionUpdateInput = {"updatedAt": datetime.now(UTC)}
+
+    if credentials is not None:
+        data["credentials"] = SafeJson(credentials)
+    if successful_agent_runs is not None:
+        data["successfulAgentRuns"] = SafeJson(successful_agent_runs)
+    if successful_agent_schedules is not None:
+        data["successfulAgentSchedules"] = SafeJson(successful_agent_schedules)
+    if total_prompt_tokens is not None:
+        data["totalPromptTokens"] = total_prompt_tokens
+    if total_completion_tokens is not None:
+        data["totalCompletionTokens"] = total_completion_tokens
+    if title is not None:
+        data["title"] = title
+
+    session = await PrismaChatSession.prisma().update(
+        where={"id": session_id},
+        data=data,
+        include={"Messages": True},
+    )
+    if session and session.Messages:
+        session.Messages.sort(key=lambda m: m.sequence)
+    return session
+
+
+async def add_chat_message(
+    session_id: str,
+    role: str,
+    sequence: int,
+    content: str | None = None,
+    name: str | None = None,
+    tool_call_id: str | None = None,
+    refusal: str | None = None,
+    tool_calls: list[dict[str, Any]] | None = None,
+    function_call: dict[str, Any] | None = None,
+) -> PrismaChatMessage:
+    """Add a message to a chat session."""
+    data: dict[str, Any] = {
+        "Session": {"connect": {"id": session_id}},
+        "role": role,
+        "sequence": sequence,
+    }
+
+    if content is not None:
+        data["content"] = content
+    if name is not None:
+        data["name"] = name
+    if tool_call_id is not None:
+        data["toolCallId"] = tool_call_id
+    if refusal is not None:
+        data["refusal"] = refusal
+    if tool_calls is not None:
+        data["toolCalls"] = SafeJson(tool_calls)
+    if function_call is not None:
+        data["functionCall"] = SafeJson(function_call)
+
+    # Update session's updatedAt timestamp
+    await PrismaChatSession.prisma().update(
+        where={"id": session_id},
+        data={"updatedAt": datetime.now(UTC)},
+    )
+
+    return await PrismaChatMessage.prisma().create(data=data)
+
+
+async def add_chat_messages_batch(
+    session_id: str,
+    messages: list[dict[str, Any]],
+    start_sequence: int,
+) -> list[PrismaChatMessage]:
+    """Add multiple messages to a chat session in a batch."""
+    if not messages:
+        return []
+
+    created_messages = []
+    for i, msg in enumerate(messages):
+        data: dict[str, Any] = {
+            "Session": {"connect": {"id": session_id}},
+            "role": msg["role"],
+            "sequence": start_sequence + i,
+        }
+
+        if msg.get("content") is not None:
+            data["content"] = msg["content"]
+        if msg.get("name") is not None:
+            data["name"] = msg["name"]
+        if msg.get("tool_call_id") is not None:
+            data["toolCallId"] = msg["tool_call_id"]
+        if msg.get("refusal") is not None:
+            data["refusal"] = msg["refusal"]
+        if msg.get("tool_calls") is not None:
+            data["toolCalls"] = SafeJson(msg["tool_calls"])
+        if msg.get("function_call") is not None:
+            data["functionCall"] = SafeJson(msg["function_call"])
+
+        created = await PrismaChatMessage.prisma().create(data=data)
+        created_messages.append(created)
+
+    # Update session's updatedAt timestamp
+    await PrismaChatSession.prisma().update(
+        where={"id": session_id},
+        data={"updatedAt": datetime.now(UTC)},
+    )
+
+    return created_messages
+
+
+async def get_user_chat_sessions(
+    user_id: str,
+    limit: int = 50,
+    offset: int = 0,
+) -> list[PrismaChatSession]:
+    """Get chat sessions for a user, ordered by most recent."""
+    return await PrismaChatSession.prisma().find_many(
+        where={"userId": user_id},
+        order={"updatedAt": "desc"},
+        take=limit,
+        skip=offset,
+    )
+
+
+async def get_user_session_count(user_id: str) -> int:
+    """Get the total number of chat sessions for a user."""
+    return await PrismaChatSession.prisma().count(where={"userId": user_id})
+
+
+async def delete_chat_session(session_id: str) -> bool:
+    """Delete a chat session and all its messages."""
+    try:
+        await PrismaChatSession.prisma().delete(where={"id": session_id})
+        return True
+    except Exception as e:
+        logger.error(f"Failed to delete chat session {session_id}: {e}")
+        return False
+
+
+async def get_chat_session_message_count(session_id: str) -> int:
+    """Get the number of messages in a chat session."""
+    count = await PrismaChatMessage.prisma().count(where={"sessionId": session_id})
+    return count

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

@@ -16,11 +16,15 @@ from openai.types.chat.chat_completion_message_tool_call_param import (
     ChatCompletionMessageToolCallParam,
     Function,
 )
+from prisma.models import ChatMessage as PrismaChatMessage
+from prisma.models import ChatSession as PrismaChatSession
 from pydantic import BaseModel
 
 from backend.data.redis_client import get_redis_async
+from backend.util import json
 from backend.util.exceptions import RedisError
 
+from . import db as chat_db
 from .config import ChatConfig
 
 logger = logging.getLogger(__name__)
@@ -46,6 +50,7 @@ class Usage(BaseModel):
 class ChatSession(BaseModel):
     session_id: str
     user_id: str | None
+    title: str | None = None
     messages: list[ChatMessage]
     usage: list[Usage]
     credentials: dict[str, dict] = {}  # Map of provider -> credential metadata
@@ -59,6 +64,7 @@ class ChatSession(BaseModel):
         return ChatSession(
             session_id=str(uuid.uuid4()),
             user_id=user_id,
+            title=None,
             messages=[],
             usage=[],
             credentials={},
@@ -66,6 +72,85 @@ class ChatSession(BaseModel):
             updated_at=datetime.now(UTC),
         )
 
+    @staticmethod
+    def from_prisma(
+        prisma_session: PrismaChatSession,
+        prisma_messages: list[PrismaChatMessage] | None = None,
+    ) -> "ChatSession":
+        """Convert Prisma models to Pydantic ChatSession."""
+        messages = []
+        if prisma_messages:
+            for msg in prisma_messages:
+                tool_calls = None
+                if msg.toolCalls:
+                    tool_calls = (
+                        json.loads(msg.toolCalls)
+                        if isinstance(msg.toolCalls, str)
+                        else msg.toolCalls
+                    )
+
+                function_call = None
+                if msg.functionCall:
+                    function_call = (
+                        json.loads(msg.functionCall)
+                        if isinstance(msg.functionCall, str)
+                        else msg.functionCall
+                    )
+
+                messages.append(
+                    ChatMessage(
+                        role=msg.role,
+                        content=msg.content,
+                        name=msg.name,
+                        tool_call_id=msg.toolCallId,
+                        refusal=msg.refusal,
+                        tool_calls=tool_calls,
+                        function_call=function_call,
+                    )
+                )
+
+        # Parse JSON fields from Prisma
+        credentials = (
+            json.loads(prisma_session.credentials)
+            if isinstance(prisma_session.credentials, str)
+            else prisma_session.credentials or {}
+        )
+        successful_agent_runs = (
+            json.loads(prisma_session.successfulAgentRuns)
+            if isinstance(prisma_session.successfulAgentRuns, str)
+            else prisma_session.successfulAgentRuns or {}
+        )
+        successful_agent_schedules = (
+            json.loads(prisma_session.successfulAgentSchedules)
+            if isinstance(prisma_session.successfulAgentSchedules, str)
+            else prisma_session.successfulAgentSchedules or {}
+        )
+
+        # Calculate usage from token counts
+        usage = []
+        if prisma_session.totalPromptTokens or prisma_session.totalCompletionTokens:
+            usage.append(
+                Usage(
+                    prompt_tokens=prisma_session.totalPromptTokens or 0,
+                    completion_tokens=prisma_session.totalCompletionTokens or 0,
+                    total_tokens=(prisma_session.totalPromptTokens or 0)
+                    + (prisma_session.totalCompletionTokens or 0),
+                )
+            )
+
+        return ChatSession(
+            session_id=prisma_session.id,
+            user_id=prisma_session.userId,
+            title=prisma_session.title,
+            messages=messages,
+            usage=usage,
+            credentials=credentials,
+            started_at=prisma_session.createdAt,
+            updated_at=prisma_session.updatedAt,
+            successful_agent_runs=successful_agent_runs,
+            successful_agent_schedules=successful_agent_schedules,
+        )
+
     def to_openai_messages(self) -> list[ChatCompletionMessageParam]:
         messages = []
         for message in self.messages:
@@ -155,50 +240,234 @@ class ChatSession(BaseModel):
         return messages
 
 
-async def get_chat_session(
-    session_id: str,
-    user_id: str | None,
-) -> ChatSession | None:
-    """Get a chat session by ID."""
+async def _get_session_from_cache(session_id: str) -> ChatSession | None:
+    """Get a chat session from Redis cache."""
     redis_key = f"chat:session:{session_id}"
     async_redis = await get_redis_async()
-
     raw_session: bytes | None = await async_redis.get(redis_key)
 
     if raw_session is None:
-        logger.warning(f"Session {session_id} not found in Redis")
         return None
 
     try:
         session = ChatSession.model_validate_json(raw_session)
+        logger.info(
+            f"Loading session {session_id} from cache: "
+            f"message_count={len(session.messages)}, "
+            f"roles={[m.role for m in session.messages]}"
+        )
+        return session
     except Exception as e:
         logger.error(f"Failed to deserialize session {session_id}: {e}", exc_info=True)
         raise RedisError(f"Corrupted session data for {session_id}") from e
 
+
+async def _cache_session(session: ChatSession) -> None:
+    """Cache a chat session in Redis."""
+    redis_key = f"chat:session:{session.session_id}"
+    async_redis = await get_redis_async()
+    await async_redis.setex(redis_key, config.session_ttl, session.model_dump_json())
+
+
+async def _get_session_from_db(session_id: str) -> ChatSession | None:
+    """Get a chat session from the database."""
+    prisma_session = await chat_db.get_chat_session(session_id)
+    if not prisma_session:
+        return None
+
+    messages = prisma_session.Messages
+    logger.info(
+        f"Loading session {session_id} from DB: "
+        f"has_messages={messages is not None}, "
+        f"message_count={len(messages) if messages else 0}, "
+        f"roles={[m.role for m in messages] if messages else []}"
+    )
+
+    return ChatSession.from_prisma(prisma_session, messages)
+
+
+async def _save_session_to_db(
+    session: ChatSession, existing_message_count: int
+) -> None:
+    """Save or update a chat session in the database."""
+    # Check if session exists in DB
+    existing = await chat_db.get_chat_session(session.session_id)
+
+    if not existing:
+        # Create new session
+        await chat_db.create_chat_session(
+            session_id=session.session_id,
+            user_id=session.user_id,
+        )
+        existing_message_count = 0
+
+    # Calculate total tokens from usage
+    total_prompt = sum(u.prompt_tokens for u in session.usage)
+    total_completion = sum(u.completion_tokens for u in session.usage)
+
+    # Update session metadata
+    await chat_db.update_chat_session(
+        session_id=session.session_id,
+        credentials=session.credentials,
+        successful_agent_runs=session.successful_agent_runs,
+        successful_agent_schedules=session.successful_agent_schedules,
+        total_prompt_tokens=total_prompt,
+        total_completion_tokens=total_completion,
+    )
+
+    # Add new messages (only those after existing count)
+    new_messages = session.messages[existing_message_count:]
+    if new_messages:
+        messages_data = []
+        for msg in new_messages:
+            messages_data.append(
+                {
+                    "role": msg.role,
+                    "content": msg.content,
+                    "name": msg.name,
+                    "tool_call_id": msg.tool_call_id,
+                    "refusal": msg.refusal,
+                    "tool_calls": msg.tool_calls,
+                    "function_call": msg.function_call,
+                }
+            )
+        logger.info(
+            f"Saving {len(new_messages)} new messages to DB for session {session.session_id}: "
+            f"roles={[m['role'] for m in messages_data]}, "
+            f"start_sequence={existing_message_count}"
+        )
+        await chat_db.add_chat_messages_batch(
+            session_id=session.session_id,
+            messages=messages_data,
+            start_sequence=existing_message_count,
+        )
+
+
+async def get_chat_session(
+    session_id: str,
+    user_id: str | None,
+) -> ChatSession | None:
+    """Get a chat session by ID.
+
+    Checks Redis cache first, falls back to database if not found.
+    Caches database results back to Redis.
+    """
+    # Try cache first
+    try:
+        session = await _get_session_from_cache(session_id)
+        if session:
+            # Verify user ownership
+            if session.user_id is not None and session.user_id != user_id:
+                logger.warning(
+                    f"Session {session_id} user id mismatch: {session.user_id} != {user_id}"
+                )
+                return None
+            return session
+    except RedisError:
+        logger.warning(f"Cache error for session {session_id}, trying database")
+    except Exception as e:
+        logger.warning(f"Unexpected cache error for session {session_id}: {e}")
+
+    # Fall back to database
+    logger.info(f"Session {session_id} not in cache, checking database")
+    session = await _get_session_from_db(session_id)
+
+    if session is None:
+        logger.warning(f"Session {session_id} not found in cache or database")
+        return None
+
+    # Verify user ownership
     if session.user_id is not None and session.user_id != user_id:
         logger.warning(
             f"Session {session_id} user id mismatch: {session.user_id} != {user_id}"
         )
         return None
 
+    # Cache the session from DB
+    try:
+        await _cache_session(session)
+        logger.info(f"Cached session {session_id} from database")
+    except Exception as e:
+        logger.warning(f"Failed to cache session {session_id}: {e}")
+
     return session
 
 
 async def upsert_chat_session(
     session: ChatSession,
 ) -> ChatSession:
-    """Update a chat session with the given messages."""
-
-    redis_key = f"chat:session:{session.session_id}"
-
-    async_redis = await get_redis_async()
-    resp = await async_redis.setex(
-        redis_key, config.session_ttl, session.model_dump_json()
+    """Update a chat session in both cache and database."""
+    # Get existing message count from DB for incremental saves
+    existing_message_count = await chat_db.get_chat_session_message_count(
+        session.session_id
     )
 
-    if not resp:
+    # Save to database
+    try:
+        await _save_session_to_db(session, existing_message_count)
+    except Exception as e:
+        logger.error(f"Failed to save session {session.session_id} to database: {e}")
+        # Continue to cache even if DB fails
+
+    # Save to cache
+    try:
+        await _cache_session(session)
+    except Exception as e:
         raise RedisError(
-            f"Failed to persist chat session {session.session_id} to Redis: {resp}"
-        )
+            f"Failed to persist chat session {session.session_id} to Redis: {e}"
+        ) from e
 
     return session
+
+
+async def create_chat_session(user_id: str | None) -> ChatSession:
+    """Create a new chat session and persist it."""
+    session = ChatSession.new(user_id)
+
+    # Create in database first
+    try:
+        await chat_db.create_chat_session(
+            session_id=session.session_id,
+            user_id=user_id,
+        )
+    except Exception as e:
+        logger.error(f"Failed to create session in database: {e}")
+        # Continue even if DB fails - cache will still work
+
+    # Cache the session
+    try:
+        await _cache_session(session)
+    except Exception as e:
+        logger.warning(f"Failed to cache new session: {e}")
+
+    return session
+
+
+async def get_user_sessions(
+    user_id: str,
+    limit: int = 50,
+    offset: int = 0,
+) -> list[ChatSession]:
+    """Get all chat sessions for a user from the database."""
+    prisma_sessions = await chat_db.get_user_chat_sessions(user_id, limit, offset)
+
+    sessions = []
+    for prisma_session in prisma_sessions:
+        # Convert without messages for listing (lighter weight)
+        sessions.append(ChatSession.from_prisma(prisma_session, None))
+
+    return sessions
+
+
+async def delete_chat_session(session_id: str) -> bool:
+    """Delete a chat session from both cache and database."""
+    # Delete from cache
+    try:
+        redis_key = f"chat:session:{session_id}"
+        async_redis = await get_redis_async()
+        await async_redis.delete(redis_key)
+    except Exception as e:
+        logger.warning(f"Failed to delete session {session_id} from cache: {e}")
+
+    # Delete from database
+    return await chat_db.delete_chat_session(session_id)

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

@@ -68,3 +68,50 @@ async def test_chatsession_redis_storage_user_id_mismatch():
     s2 = await get_chat_session(s.session_id, None)
 
     assert s2 is None
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_chatsession_db_storage():
+    """Test that messages are correctly saved to and loaded from DB (not cache)."""
+    from backend.data.redis_client import get_redis_async
+
+    # Create session with messages including assistant message
+    s = ChatSession.new(user_id=None)
+    s.messages = messages  # Contains user, assistant, and tool messages
+
+    # Upsert to save to both cache and DB
+    s = await upsert_chat_session(s)
+
+    # Clear the Redis cache to force DB load
+    redis_key = f"chat:session:{s.session_id}"
+    async_redis = await get_redis_async()
+    await async_redis.delete(redis_key)
+
+    # Load from DB (cache was cleared)
+    s2 = await get_chat_session(
+        session_id=s.session_id,
+        user_id=s.user_id,
+    )
+
+    assert s2 is not None, "Session not found after loading from DB"
+    assert len(s2.messages) == len(
+        s.messages
+    ), f"Message count mismatch: expected {len(s.messages)}, got {len(s2.messages)}"
+
+    # Verify all roles are present
+    roles = [m.role for m in s2.messages]
+    assert "user" in roles, f"User message missing. Roles found: {roles}"
+    assert "assistant" in roles, f"Assistant message missing. Roles found: {roles}"
+    assert "tool" in roles, f"Tool message missing. Roles found: {roles}"
+
+    # Verify message content
+    for orig, loaded in zip(s.messages, s2.messages):
+        assert orig.role == loaded.role, f"Role mismatch: {orig.role} != {loaded.role}"
+        assert (
+            orig.content == loaded.content
+        ), f"Content mismatch for {orig.role}: {orig.content} != {loaded.content}"
+        if orig.tool_calls:
+            assert (
+                loaded.tool_calls is not None
+            ), f"Tool calls missing for {orig.role} message"
+            assert len(orig.tool_calls) == len(loaded.tool_calls)

autogpt_platform/backend/backend/api/features/chat/prompts/chat_system.md

@@ -1,12 +1,80 @@
-You are Otto, an AI Co-Pilot and Forward Deployed Engineer for AutoGPT, an AI Business Automation tool. Your mission is to help users quickly find and set up AutoGPT agents to solve their business problems.
+You are Otto, an AI Co-Pilot and Forward Deployed Engineer for AutoGPT, an AI Business Automation tool. Your mission is to help users quickly find, create, and set up AutoGPT agents to solve their business problems.
 
 Here are the functions available to you:
 
 <functions>
-1. **find_agent** - Search for agents that solve the user's problem
-2. **run_agent** - Run or schedule an agent (automatically handles setup)
+**Understanding & Discovery:**
+1. **add_understanding** - Save information about the user's business context (use this as you learn about them)
+2. **find_agent** - Search the marketplace for pre-built agents that solve the user's problem
+3. **find_library_agent** - Search the user's personal library of saved agents
+4. **find_block** - Search for individual blocks (building components for agents)
+5. **search_platform_docs** - Search AutoGPT documentation for help
+
+**Agent Creation & Editing:**
+6. **create_agent** - Create a new custom agent from scratch based on user requirements
+7. **edit_agent** - Modify an existing agent (add/remove blocks, change configuration)
+
+**Execution & Output:**
+8. **run_agent** - Run or schedule an agent (automatically handles setup)
+9. **run_block** - Run a single block directly without creating an agent
+10. **agent_output** - Get the output/results from a running or completed agent execution
 </functions>
 
+## ALWAYS GET THE USER'S NAME
+
+**This is critical:** If you don't know the user's name, ask for it in your first response. Use a friendly, natural approach:
+- "Hi! I'm Otto. What's your name?"
+- "Hey there! Before we dive in, what should I call you?"
+
+Once you have their name, immediately save it with `add_understanding(user_name="...")` and use it throughout the conversation.
+
+## BUILDING USER UNDERSTANDING
+
+**If no User Business Context is provided below**, gather information naturally during conversation - don't interrogate them.
+
+**Key information to gather (in priority order):**
+1. Their name (ALWAYS first if unknown)
+2. Their job title and role
+3. Their business/company and industry
+4. Pain points and what they want to automate
+5. Tools they currently use
+
+**How to gather this information:**
+- Ask naturally as part of helping them (e.g., "What's your role?" or "What industry are you in?")
+- When they share information, immediately save it using `add_understanding`
+- Don't ask all questions at once - spread them across the conversation
+- Prioritize understanding their immediate problem first
+
+**Example:**
+```
+User: "I need help automating my social media"
+Otto: I can help with that! I'm Otto - what's your name?
+User: "I'm Sarah"
+Otto: [calls add_understanding with user_name="Sarah"]
+Nice to meet you, Sarah! What's your role - are you a social media manager or business owner?
+User: "I'm the marketing director at a fintech startup"
+Otto: [calls add_understanding with job_title="Marketing Director", industry="fintech", business_size="startup"]
+Great! Let me find social media automation agents for you.
+[calls find_agent with query="social media automation marketing"]
+```
+
+## WHEN TO USE WHICH TOOL
+
+**Finding existing agents:**
+- `find_agent` - Search the marketplace for pre-built agents others have created
+- `find_library_agent` - Search agents the user has already saved to their library
+
+**Creating/editing agents:**
+- `create_agent` - When user wants a custom agent that doesn't exist, or has specific requirements
+- `edit_agent` - When user wants to modify an existing agent (change inputs, add blocks, etc.)
+
+**Running agents:**
+- `run_agent` - To execute an agent (handles credentials and inputs automatically)
+- `agent_output` - To check the results of a running or completed agent execution
+
+**Direct execution:**
+- `run_block` - Run a single block directly without needing a full agent
+
 ## HOW run_agent WORKS
 
 The `run_agent` tool automatically handles the entire setup flow:
@@ -21,49 +89,61 @@ Parameters:
 - `use_defaults`: Set to `true` to run with default values (only after user confirms)
 - `schedule_name` + `cron`: For scheduled execution
 
+## HOW create_agent WORKS
+
+Use `create_agent` when the user wants to build a custom automation:
+- Describe what the agent should do
+- The tool will create the agent structure with appropriate blocks
+- Returns the agent ID for further editing or running
+
+## HOW agent_output WORKS
+
+Use `agent_output` to get results from agent executions:
+- Pass the execution_id from a run_agent response
+- Returns the current status and any outputs produced
+- Useful for checking if an agent has completed and what it produced
+
 ## WORKFLOW
 
-1. **find_agent** - Search for agents that solve the user's problem
-2. **run_agent** (first call, no inputs) - Get available inputs for the agent
-3. **Ask user** what values they want to use OR if they want to use defaults
-4. **run_agent** (second call) - Either with `inputs={...}` or `use_defaults=true`
+1. **Get their name** - If unknown, ask for it first
+2. **Understand context** - Ask 1-2 questions about their problem while helping
+3. **Find or create** - Use find_agent for existing solutions, create_agent for custom needs
+4. **Set up and run** - Use run_agent to execute, agent_output to get results
 
 ## YOUR APPROACH
 
-**Step 1: Understand the Problem**
+**Step 1: Greet and Identify**
+- If you don't know their name, ask for it
+- Be friendly and conversational
+
+**Step 2: Understand the Problem**
 - Ask maximum 1-2 targeted questions
 - Focus on: What business problem are they solving?
-- Move quickly to searching for solutions
+- If they want to create/edit an agent, understand what it should do
 
-**Step 2: Find Agents**
-- Use `find_agent` immediately with relevant keywords
-- Suggest the best option from search results
-- Explain briefly how it solves their problem
+**Step 3: Find or Create**
+- For existing solutions: Use `find_agent` with relevant keywords
+- For custom needs: Use `create_agent` with their requirements
+- For modifications: Use `edit_agent` on an existing agent
 
-**Step 3: Get Agent Inputs**
-- Call `run_agent(username_agent_slug="creator/agent-name")` without inputs
-- This returns the available inputs (required and optional)
-- Present these to the user and ask what values they want
+**Step 4: Execute**
+- Call `run_agent` without inputs first to see what's available
+- Ask user what values they want or if defaults are okay
+- Call `run_agent` again with inputs or `use_defaults=true`
+- Use `agent_output` to check results when needed
 
-**Step 4: Run with User's Choice**
-- If user provides values: `run_agent(username_agent_slug="...", inputs={...})`
-- If user says "use defaults": `run_agent(username_agent_slug="...", use_defaults=true)`
-- On success, share the agent link with the user
+## USING add_understanding
 
-**For Scheduled Execution:**
-- Add `schedule_name` and `cron` parameters
-- Example: `run_agent(username_agent_slug="...", inputs={...}, schedule_name="Daily Report", cron="0 9 * * *")`
+Call `add_understanding` whenever you learn something about the user:
 
-## FUNCTION CALL FORMAT
+**User info:** `user_name`, `job_title`
+**Business:** `business_name`, `industry`, `business_size` (1-10, 11-50, 51-200, 201-1000, 1000+), `user_role` (decision maker, implementer, end user)
+**Processes:** `key_workflows` (array), `daily_activities` (array)
+**Pain points:** `pain_points` (array), `bottlenecks` (array), `manual_tasks` (array), `automation_goals` (array)
+**Tools:** `current_software` (array), `existing_automation` (array)
+**Other:** `additional_notes`
 
-To call a function, use this exact format:
-`<function_call>function_name(parameter="value")</function_call>`
-
-Examples:
-- `<function_call>find_agent(query="social media automation")</function_call>`
-- `<function_call>run_agent(username_agent_slug="creator/agent-name")</function_call>` (get inputs)
-- `<function_call>run_agent(username_agent_slug="creator/agent-name", inputs={"topic": "AI news"})</function_call>`
-- `<function_call>run_agent(username_agent_slug="creator/agent-name", use_defaults=true)</function_call>`
+Example: `add_understanding(user_name="Sarah", job_title="Marketing Director", industry="fintech")`
 
 ## KEY RULES
 
@@ -73,8 +153,12 @@ Examples:
 - Don't run agents without first showing available inputs to the user
 - Don't use `use_defaults=true` without user explicitly confirming
 - Don't write responses longer than 3 sentences
+- Don't interrogate users with many questions - gather info naturally
 
 **What You DO:**
+- ALWAYS ask for user's name if you don't have it
+- Save user information with `add_understanding` as you learn it
+- Use their name when addressing them
 - Always call run_agent first without inputs to see what's available
 - Ask user what values they want OR if they want to use defaults
 - Keep all responses to maximum 3 sentences
@@ -87,18 +171,22 @@ Examples:
 ## RESPONSE STRUCTURE
 
 Before responding, wrap your analysis in <thinking> tags to systematically plan your approach:
+- Check if you know the user's name - if not, ask for it
+- Check if you have user context - if not, plan to gather some naturally
 - Extract the key business problem or request from the user's message
 - Determine what function call (if any) you need to make next
 - Plan your response to stay under the 3-sentence maximum
 
 Example interaction:
 ```
-User: "Run the AI news agent for me"
-Otto: <function_call>run_agent(username_agent_slug="autogpt/ai-news")</function_call>
-[Tool returns: Agent accepts inputs - Required: topic. Optional: num_articles (default: 5)]
-Otto: The AI News agent needs a topic. What topic would you like news about, or should I use the defaults?
-User: "Use defaults"
-Otto: <function_call>run_agent(username_agent_slug="autogpt/ai-news", use_defaults=true)</function_call>
+User: "Hi, I want to build an agent that monitors my competitors"
+Otto: <thinking>I don't know this user's name. I should ask for it while acknowledging their request.</thinking>
+Hi! I'm Otto and I'd love to help you build a competitor monitoring agent. What's your name?
+User: "I'm Mike"
+Otto: [calls add_understanding with user_name="Mike"]
+<thinking>Now I know Mike wants competitor monitoring. I should search for existing agents first.</thinking>
+Great to meet you, Mike! Let me search for competitor monitoring agents.
+[calls find_agent with query="competitor monitoring analysis"]
 ```
 
 KEEP ANSWERS TO 3 SENTENCES

autogpt_platform/backend/backend/api/features/chat/prompts/onboarding_system.md

@@ -0,0 +1,155 @@
+You are Otto, an AI Co-Pilot helping new users get started with AutoGPT, an AI Business Automation platform. Your mission is to welcome them, learn about their needs, and help them run their first successful agent.
+
+Here are the functions available to you:
+
+<functions>
+**Understanding & Discovery:**
+1. **add_understanding** - Save information about the user's business context (use this as you learn about them)
+2. **find_agent** - Search the marketplace for pre-built agents that solve the user's problem
+3. **find_library_agent** - Search the user's personal library of saved agents
+4. **find_block** - Search for individual blocks (building components for agents)
+5. **search_platform_docs** - Search AutoGPT documentation for help
+
+**Agent Creation & Editing:**
+6. **create_agent** - Create a new custom agent from scratch based on user requirements
+7. **edit_agent** - Modify an existing agent (add/remove blocks, change configuration)
+
+**Execution & Output:**
+8. **run_agent** - Run or schedule an agent (automatically handles setup)
+9. **run_block** - Run a single block directly without creating an agent
+10. **agent_output** - Get the output/results from a running or completed agent execution
+</functions>
+
+## YOUR ONBOARDING MISSION
+
+You are guiding a new user through their first experience with AutoGPT. Your goal is to:
+1. Welcome them warmly and get their name
+2. Learn about them and their business
+3. Find or create an agent that solves a real problem for them
+4. Get that agent running successfully
+5. Celebrate their success and point them to next steps
+
+## PHASE 1: WELCOME & INTRODUCTION
+
+**Start every conversation by:**
+- Giving a warm, friendly greeting
+- Introducing yourself as Otto, their AI assistant
+- Asking for their name immediately
+
+**Example opening:**
+```
+Hi! I'm Otto, your AI assistant. Welcome to AutoGPT! I'm here to help you set up your first automation. What's your name?
+```
+
+Once you have their name, save it immediately with `add_understanding(user_name="...")` and use it throughout.
+
+## PHASE 2: DISCOVERY
+
+**After getting their name, learn about them:**
+- What's their role/job title?
+- What industry/business are they in?
+- What's one thing they'd love to automate?
+
+**Keep it conversational - don't interrogate. Example:**
+```
+Nice to meet you, Sarah! What do you do for work, and what's one task you wish you could automate?
+```
+
+Save everything you learn with `add_understanding`.
+
+## PHASE 3: FIND OR CREATE AN AGENT
+
+**Once you understand their need:**
+- Search for existing agents with `find_agent`
+- Present the best match and explain how it helps them
+- If nothing fits, offer to create a custom agent with `create_agent`
+
+**Be enthusiastic about the solution:**
+```
+I found a great agent for you! The "Social Media Scheduler" can automatically post to your accounts on a schedule. Want to try it?
+```
+
+## PHASE 4: SETUP & RUN
+
+**Guide them through running the agent:**
+1. Call `run_agent` without inputs first to see what's needed
+2. Explain each input in simple terms
+3. Ask what values they want to use
+4. Run the agent with their inputs or defaults
+
+**Don't mention credentials** - the UI handles that automatically.
+
+## PHASE 5: CELEBRATE & HANDOFF
+
+**After successful execution:**
+- Congratulate them on their first automation!
+- Tell them where to find this agent (their Library)
+- Mention they can explore more agents in the Marketplace
+- Offer to help with anything else
+
+**Example:**
+```
+You did it! Your first agent is running. You can find it anytime in your Library. Ready to explore more automations?
+```
+
+## KEY RULES
+
+**What You DON'T Do:**
+- Don't help with login (frontend handles this)
+- Don't mention credentials (UI handles automatically)
+- Don't run agents without showing inputs first
+- Don't use `use_defaults=true` without explicit confirmation
+- Don't write responses longer than 3 sentences
+- Don't overwhelm with too many questions at once
+
+**What You DO:**
+- ALWAYS get the user's name first
+- Be warm, encouraging, and celebratory
+- Save info with `add_understanding` as you learn it
+- Use their name when addressing them
+- Keep responses to maximum 3 sentences
+- Make them feel successful at each step
+
+## USING add_understanding
+
+Save information as you learn it:
+
+**User info:** `user_name`, `job_title`
+**Business:** `business_name`, `industry`, `business_size`, `user_role`
+**Pain points:** `pain_points`, `manual_tasks`, `automation_goals`
+**Tools:** `current_software`
+
+Example: `add_understanding(user_name="Sarah", job_title="Marketing Manager", automation_goals=["social media scheduling"])`
+
+## HOW run_agent WORKS
+
+1. **First call** (no inputs) → Shows available inputs
+2. **Credentials** → UI handles automatically (don't mention)
+3. **Execution** → Run with `inputs={...}` or `use_defaults=true`
+
+## RESPONSE STRUCTURE
+
+Before responding, plan your approach in <thinking> tags:
+- What phase am I in? (Welcome/Discovery/Find/Setup/Celebrate)
+- Do I know their name? If not, ask for it
+- What's the next step to move them forward?
+- Keep response under 3 sentences
+
+**Example flow:**
+```
+User: "Hi"
+Otto: <thinking>Phase 1 - I need to welcome them and get their name.</thinking>
+Hi! I'm Otto, welcome to AutoGPT! I'm here to help you set up your first automation - what's your name?
+
+User: "I'm Alex"
+Otto: [calls add_understanding with user_name="Alex"]
+<thinking>Got their name. Phase 2 - learn about them.</thinking>
+Great to meet you, Alex! What do you do for work, and what's one task you'd love to automate?
+
+User: "I run an e-commerce store and spend hours on customer support emails"
+Otto: [calls add_understanding with industry="e-commerce", pain_points=["customer support emails"]]
+<thinking>Phase 3 - search for agents.</thinking>
+[calls find_agent with query="customer support email automation"]
+```
+
+KEEP ANSWERS TO 3 SENTENCES - Be warm, helpful, and focused on their success!

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

@@ -26,6 +26,14 @@ router = APIRouter(
 # ========== Request/Response Models ==========
 
 
+class StreamChatRequest(BaseModel):
+    """Request model for streaming chat with optional context."""
+
+    message: str
+    is_user_message: bool = True
+    context: dict[str, str] | None = None  # {url: str, content: str}
+
+
 class CreateSessionResponse(BaseModel):
     """Response model containing information on a newly created chat session."""
 
@@ -44,9 +52,64 @@ class SessionDetailResponse(BaseModel):
     messages: list[dict]
 
 
+class SessionSummaryResponse(BaseModel):
+    """Response model for a session summary (without messages)."""
+
+    id: str
+    created_at: str
+    updated_at: str
+    title: str | None = None
+
+
+class ListSessionsResponse(BaseModel):
+    """Response model for listing chat sessions."""
+
+    sessions: list[SessionSummaryResponse]
+    total: int
+
+
 # ========== Routes ==========
 
 
+@router.get(
+    "/sessions",
+    dependencies=[Security(auth.requires_user)],
+)
+async def list_sessions(
+    user_id: Annotated[str, Security(auth.get_user_id)],
+    limit: int = Query(default=50, ge=1, le=100),
+    offset: int = Query(default=0, ge=0),
+) -> ListSessionsResponse:
+    """
+    List chat sessions for the authenticated user.
+
+    Returns a paginated list of chat sessions belonging to the current user,
+    ordered by most recently updated.
+
+    Args:
+        user_id: The authenticated user's ID.
+        limit: Maximum number of sessions to return (1-100).
+        offset: Number of sessions to skip for pagination.
+
+    Returns:
+        ListSessionsResponse: List of session summaries and total count.
+    """
+    sessions = await chat_service.get_user_sessions(user_id, limit, offset)
+
+    return ListSessionsResponse(
+        sessions=[
+            SessionSummaryResponse(
+                id=session.session_id,
+                created_at=session.started_at.isoformat(),
+                updated_at=session.updated_at.isoformat(),
+                title=None,  # TODO: Add title support
+            )
+            for session in sessions
+        ],
+        total=len(sessions),
+    )
+
+
 @router.post(
     "/sessions",
 )
@@ -102,26 +165,89 @@ async def get_session(
     session = await chat_service.get_session(session_id, user_id)
     if not session:
         raise NotFoundError(f"Session {session_id} not found")
+
+    messages = [message.model_dump() for message in session.messages]
+    logger.info(
+        f"Returning session {session_id}: "
+        f"message_count={len(messages)}, "
+        f"roles={[m.get('role') for m in messages]}"
+    )
+
     return SessionDetailResponse(
         id=session.session_id,
         created_at=session.started_at.isoformat(),
         updated_at=session.updated_at.isoformat(),
         user_id=session.user_id or None,
-        messages=[message.model_dump() for message in session.messages],
+        messages=messages,
+    )
+
+
+@router.post(
+    "/sessions/{session_id}/stream",
+)
+async def stream_chat_post(
+    session_id: str,
+    request: StreamChatRequest,
+    user_id: str | None = Depends(auth.get_user_id),
+):
+    """
+    Stream chat responses for a session (POST with context support).
+
+    Streams the AI/completion responses in real time over Server-Sent Events (SSE), including:
+      - Text fragments as they are generated
+      - Tool call UI elements (if invoked)
+      - Tool execution results
+
+    Args:
+        session_id: The chat session identifier to associate with the streamed messages.
+        request: Request body containing message, is_user_message, and optional context.
+        user_id: Optional authenticated user ID.
+    Returns:
+        StreamingResponse: SSE-formatted response chunks.
+
+    """
+    # Validate session exists before starting the stream
+    # This prevents errors after the response has already started
+    session = await chat_service.get_session(session_id, user_id)
+
+    if not session:
+        raise NotFoundError(f"Session {session_id} not found. ")
+    if session.user_id is None and user_id is not None:
+        session = await chat_service.assign_user_to_session(session_id, user_id)
+
+    async def event_generator() -> AsyncGenerator[str, None]:
+        async for chunk in chat_service.stream_chat_completion(
+            session_id,
+            request.message,
+            is_user_message=request.is_user_message,
+            user_id=user_id,
+            session=session,  # Pass pre-fetched session to avoid double-fetch
+            context=request.context,
+        ):
+            yield chunk.to_sse()
+
+    return StreamingResponse(
+        event_generator(),
+        media_type="text/event-stream",
+        headers={
+            "Cache-Control": "no-cache",
+            "Connection": "keep-alive",
+            "X-Accel-Buffering": "no",  # Disable nginx buffering
+        },
     )
 
 
 @router.get(
     "/sessions/{session_id}/stream",
 )
-async def stream_chat(
+async def stream_chat_get(
     session_id: str,
     message: Annotated[str, Query(min_length=1, max_length=10000)],
     user_id: str | None = Depends(auth.get_user_id),
     is_user_message: bool = Query(default=True),
 ):
     """
-    Stream chat responses for a session.
+    Stream chat responses for a session (GET - legacy endpoint).
 
     Streams the AI/completion responses in real time over Server-Sent Events (SSE), including:
       - Text fragments as they are generated
@@ -193,6 +319,133 @@ async def session_assign_user(
     return {"status": "ok"}
 
 
+# ========== Onboarding Routes ==========
+# These routes use a specialized onboarding system prompt
+
+
+@router.post(
+    "/onboarding/sessions",
+)
+async def create_onboarding_session(
+    user_id: Annotated[str | None, Depends(auth.get_user_id)],
+) -> CreateSessionResponse:
+    """
+    Create a new onboarding chat session.
+
+    Initiates a new chat session specifically for user onboarding,
+    using a specialized prompt that guides users through their first
+    experience with AutoGPT.
+
+    Args:
+        user_id: The optional authenticated user ID parsed from the JWT.
+
+    Returns:
+        CreateSessionResponse: Details of the created onboarding session.
+    """
+    logger.info(
+        f"Creating onboarding session with user_id: "
+        f"...{user_id[-8:] if user_id and len(user_id) > 8 else '<redacted>'}"
+    )
+
+    session = await chat_service.create_chat_session(user_id)
+
+    return CreateSessionResponse(
+        id=session.session_id,
+        created_at=session.started_at.isoformat(),
+        user_id=session.user_id or None,
+    )
+
+
+@router.get(
+    "/onboarding/sessions/{session_id}",
+)
+async def get_onboarding_session(
+    session_id: str,
+    user_id: Annotated[str | None, Depends(auth.get_user_id)],
+) -> SessionDetailResponse:
+    """
+    Retrieve the details of an onboarding chat session.
+
+    Args:
+        session_id: The unique identifier for the onboarding session.
+        user_id: The optional authenticated user ID.
+
+    Returns:
+        SessionDetailResponse: Details for the requested session.
+    """
+    session = await chat_service.get_session(session_id, user_id)
+    if not session:
+        raise NotFoundError(f"Session {session_id} not found")
+
+    messages = [message.model_dump() for message in session.messages]
+    logger.info(
+        f"Returning onboarding session {session_id}: "
+        f"message_count={len(messages)}, "
+        f"roles={[m.get('role') for m in messages]}"
+    )
+
+    return SessionDetailResponse(
+        id=session.session_id,
+        created_at=session.started_at.isoformat(),
+        updated_at=session.updated_at.isoformat(),
+        user_id=session.user_id or None,
+        messages=messages,
+    )
+
+
+@router.post(
+    "/onboarding/sessions/{session_id}/stream",
+)
+async def stream_onboarding_chat(
+    session_id: str,
+    request: StreamChatRequest,
+    user_id: str | None = Depends(auth.get_user_id),
+):
+    """
+    Stream onboarding chat responses for a session.
+
+    Uses the specialized onboarding system prompt to guide new users
+    through their first experience with AutoGPT. Streams AI responses
+    in real time over Server-Sent Events (SSE).
+
+    Args:
+        session_id: The onboarding session identifier.
+        request: Request body containing message and optional context.
+        user_id: Optional authenticated user ID.
+
+    Returns:
+        StreamingResponse: SSE-formatted response chunks.
+    """
+    session = await chat_service.get_session(session_id, user_id)
+
+    if not session:
+        raise NotFoundError(f"Session {session_id} not found.")
+    if session.user_id is None and user_id is not None:
+        session = await chat_service.assign_user_to_session(session_id, user_id)
+
+    async def event_generator() -> AsyncGenerator[str, None]:
+        async for chunk in chat_service.stream_chat_completion(
+            session_id,
+            request.message,
+            is_user_message=request.is_user_message,
+            user_id=user_id,
+            session=session,
+            context=request.context,
+            prompt_type="onboarding",  # Use onboarding system prompt
+        ):
+            yield chunk.to_sse()
+
+    return StreamingResponse(
+        event_generator(),
+        media_type="text/event-stream",
+        headers={
+            "Cache-Control": "no-cache",
+            "Connection": "keep-alive",
+            "X-Accel-Buffering": "no",
+        },
+    )
+
+
 # ========== Health Check ==========
 
 

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

@@ -4,11 +4,18 @@ from datetime import UTC, datetime
 from typing import Any
 
 import orjson
+from langfuse import Langfuse
 from openai import AsyncOpenAI
 from openai.types.chat import ChatCompletionChunk, ChatCompletionToolParam
 
+from backend.data.understanding import (
+    format_understanding_for_prompt,
+    get_business_understanding,
+)
 from backend.util.exceptions import NotFoundError
+from backend.util.settings import Settings
 
+from . import db as chat_db
 from .config import ChatConfig
 from .model import (
     ChatMessage,
@@ -17,6 +24,9 @@ from .model import (
     get_chat_session,
     upsert_chat_session,
 )
+from .model import (
+    create_chat_session as model_create_chat_session,
+)
 from .response_model import (
     StreamBaseResponse,
     StreamEnd,
@@ -33,8 +43,154 @@ from .tools import execute_tool, tools
 logger = logging.getLogger(__name__)
 
 config = ChatConfig()
+settings = Settings()
 client = AsyncOpenAI(api_key=config.api_key, base_url=config.base_url)
 
+# Langfuse client (lazy initialization)
+_langfuse_client: Langfuse | None = None
+
+
+def _get_langfuse_client() -> Langfuse:
+    """Get or create the Langfuse client for prompt management."""
+    global _langfuse_client
+    if _langfuse_client is None:
+        if not settings.secrets.langfuse_public_key or not settings.secrets.langfuse_secret_key:
+            raise ValueError(
+                "Langfuse credentials not configured. "
+                "Set LANGFUSE_PUBLIC_KEY and LANGFUSE_SECRET_KEY environment variables."
+            )
+        _langfuse_client = Langfuse(
+            public_key=settings.secrets.langfuse_public_key,
+            secret_key=settings.secrets.langfuse_secret_key,
+            host=settings.secrets.langfuse_host or "https://cloud.langfuse.com",
+        )
+    return _langfuse_client
+
+
+def _get_langfuse_prompt() -> str:
+    """Fetch the latest production prompt from Langfuse.
+
+    Returns:
+        The compiled prompt text from Langfuse.
+
+    Raises:
+        Exception: If Langfuse is unavailable or prompt fetch fails.
+    """
+    try:
+        langfuse = _get_langfuse_client()
+        # cache_ttl_seconds=0 disables SDK caching to always get the latest prompt
+        prompt = langfuse.get_prompt(config.langfuse_prompt_name, cache_ttl_seconds=0)
+        compiled = prompt.compile()
+        logger.info(
+            f"Fetched prompt '{config.langfuse_prompt_name}' from Langfuse "
+            f"(version: {prompt.version})"
+        )
+        return compiled
+    except Exception as e:
+        logger.error(f"Failed to fetch prompt from Langfuse: {e}")
+        raise
+
+
+async def _is_first_session(user_id: str) -> bool:
+    """Check if this is the user's first chat session.
+
+    Returns True if the user has 1 or fewer sessions (meaning this is their first).
+    """
+    try:
+        session_count = await chat_db.get_user_session_count(user_id)
+        return session_count <= 1
+    except Exception as e:
+        logger.warning(f"Failed to check session count for user {user_id}: {e}")
+        return False  # Default to non-onboarding if we can't check
+
+
+async def _build_system_prompt(
+    user_id: str | None, prompt_type: str = "default"
+) -> str:
+    """Build the full system prompt including business understanding if available.
+
+    Args:
+        user_id: The user ID for fetching business understanding
+        prompt_type: The type of prompt to load ("default" or "onboarding")
+                     If "default" and this is the user's first session, will use "onboarding" instead.
+
+    Returns:
+        The full system prompt with business understanding context if available
+    """
+    # Auto-detect: if using default prompt and this is user's first session, use onboarding
+    effective_prompt_type = prompt_type
+    if prompt_type == "default" and user_id:
+        if await _is_first_session(user_id):
+            logger.info("First session detected for user, using onboarding prompt")
+            effective_prompt_type = "onboarding"
+
+    # Start with the base system prompt for the specified type
+    if effective_prompt_type == "default":
+        # Fetch from Langfuse for the default prompt
+        base_prompt = _get_langfuse_prompt()
+    else:
+        # Use local file for other prompt types (e.g., onboarding)
+        base_prompt = config.get_system_prompt_for_type(effective_prompt_type)
+
+    # If user is authenticated, try to fetch their business understanding
+    if user_id:
+        try:
+            understanding = await get_business_understanding(user_id)
+            if understanding:
+                context = format_understanding_for_prompt(understanding)
+                if context:
+                    return (
+                        f"{base_prompt}\n\n---\n\n"
+                        f"{context}\n\n"
+                        "Use this context to provide more personalized recommendations "
+                        "and to better understand the user's business needs when "
+                        "suggesting agents and automations."
+                    )
+        except Exception as e:
+            logger.warning(f"Failed to fetch business understanding: {e}")
+
+    return base_prompt
+
+
+async def _generate_session_title(message: str) -> str | None:
+    """Generate a concise title for a chat session based on the first message.
+
+    Args:
+        message: The first user message in the session
+
+    Returns:
+        A short title (3-6 words) or None if generation fails
+    """
+    try:
+        response = await client.chat.completions.create(
+            model=config.title_model,
+            messages=[
+                {
+                    "role": "system",
+                    "content": (
+                        "Generate a very short title (3-6 words) for a chat conversation "
+                        "based on the user's first message. The title should capture the "
+                        "main topic or intent. Return ONLY the title, no quotes or punctuation."
+                    ),
+                },
+                {"role": "user", "content": message[:500]},  # Limit input length
+            ],
+            max_tokens=20,
+            temperature=0.7,
+        )
+        title = response.choices[0].message.content
+        if title:
+            # Clean up the title
+            title = title.strip().strip("\"'")
+            # Limit length
+            if len(title) > 50:
+                title = title[:47] + "..."
+            return title
+        return None
+    except Exception as e:
+        logger.warning(f"Failed to generate session title: {e}")
+        return None
+
 
 async def create_chat_session(
     user_id: str | None = None,
@@ -42,9 +198,7 @@ async def create_chat_session(
     """
     Create a new chat session and persist it to the database.
     """
-    session = ChatSession.new(user_id)
-    # Persist the session immediately so it can be used for streaming
-    return await upsert_chat_session(session)
+    return await model_create_chat_session(user_id)
 
 
 async def get_session(
@@ -57,6 +211,19 @@ async def get_session(
     return await get_chat_session(session_id, user_id)
 
 
+async def get_user_sessions(
+    user_id: str,
+    limit: int = 50,
+    offset: int = 0,
+) -> list[ChatSession]:
+    """
+    Get all chat sessions for a user.
+    """
+    from .model import get_user_sessions as model_get_user_sessions
+
+    return await model_get_user_sessions(user_id, limit, offset)
+
+
 async def assign_user_to_session(
     session_id: str,
     user_id: str,
@@ -78,6 +245,8 @@ async def stream_chat_completion(
     user_id: str | None = None,
     retry_count: int = 0,
     session: ChatSession | None = None,
+    context: dict[str, str] | None = None,  # {url: str, content: str}
+    prompt_type: str = "default",
 ) -> AsyncGenerator[StreamBaseResponse, None]:
     """Main entry point for streaming chat completions with database handling.
 
@@ -89,6 +258,7 @@ async def stream_chat_completion(
         user_message: User's input message
         user_id: User ID for authentication (None for anonymous)
         session: Optional pre-loaded session object (for recursive calls to avoid Redis refetch)
+        prompt_type: The type of prompt to use ("default" or "onboarding")
 
     Yields:
         StreamBaseResponse objects formatted as SSE
@@ -121,9 +291,18 @@ async def stream_chat_completion(
         )
 
     if message:
+        # Build message content with context if provided
+        message_content = message
+        if context and context.get("url") and context.get("content"):
+            context_text = f"Page URL: {context['url']}\n\nPage Content:\n{context['content']}\n\n---\n\nUser Message: {message}"
+            message_content = context_text
+            logger.info(
+                f"Including page context: URL={context['url']}, content_length={len(context['content'])}"
+            )
+
         session.messages.append(
             ChatMessage(
-                role="user" if is_user_message else "assistant", content=message
+                role="user" if is_user_message else "assistant", content=message_content
             )
         )
         logger.info(
@@ -141,6 +320,32 @@ async def stream_chat_completion(
     session = await upsert_chat_session(session)
     assert session, "Session not found"
 
+    # Generate title for new sessions on first user message (non-blocking)
+    # Check: is_user_message, no title yet, and this is the first user message
+    if is_user_message and message and not session.title:
+        user_messages = [m for m in session.messages if m.role == "user"]
+        if len(user_messages) == 1:
+            # First user message - generate title in background
+            import asyncio
+
+            async def _update_title():
+                try:
+                    title = await _generate_session_title(message)
+                    if title:
+                        session.title = title
+                        await upsert_chat_session(session)
+                        logger.info(
+                            f"Generated title for session {session_id}: {title}"
+                        )
+                except Exception as e:
+                    logger.warning(f"Failed to update session title: {e}")
+
+            # Fire and forget - don't block the chat response
+            asyncio.create_task(_update_title())
+
+    # Build system prompt with business understanding
+    system_prompt = await _build_system_prompt(user_id, prompt_type)
+
     assistant_response = ChatMessage(
         role="assistant",
         content="",
@@ -159,6 +364,7 @@ async def stream_chat_completion(
         async for chunk in _stream_chat_chunks(
             session=session,
             tools=tools,
+            system_prompt=system_prompt,
         ):
 
             if isinstance(chunk, StreamTextChunk):
@@ -279,6 +485,7 @@ async def stream_chat_completion(
             user_id=user_id,
             retry_count=retry_count + 1,
             session=session,
+            prompt_type=prompt_type,
         ):
             yield chunk
         return  # Exit after retry to avoid double-saving in finally block
@@ -324,6 +531,7 @@ async def stream_chat_completion(
             session_id=session.session_id,
             user_id=user_id,
             session=session,  # Pass session object to avoid Redis refetch
+            prompt_type=prompt_type,
         ):
             yield chunk
 
@@ -331,6 +539,7 @@ async def stream_chat_completion(
 async def _stream_chat_chunks(
     session: ChatSession,
     tools: list[ChatCompletionToolParam],
+    system_prompt: str | None = None,
 ) -> AsyncGenerator[StreamBaseResponse, None]:
     """
     Pure streaming function for OpenAI chat completions with tool calling.
@@ -338,9 +547,9 @@ async def _stream_chat_chunks(
     This function is database-agnostic and focuses only on streaming logic.
 
     Args:
-        messages: Conversation context as ChatCompletionMessageParam list
-        session_id: Session ID
-        user_id: User ID for tool execution
+        session: Chat session with conversation history
+        tools: Available tools for the model
+        system_prompt: System prompt to prepend to messages
 
     Yields:
         SSE formatted JSON response objects
@@ -350,6 +559,17 @@ async def _stream_chat_chunks(
 
     logger.info("Starting pure chat stream")
 
+    # Build messages with system prompt prepended
+    messages = session.to_openai_messages()
+    if system_prompt:
+        from openai.types.chat import ChatCompletionSystemMessageParam
+
+        system_message = ChatCompletionSystemMessageParam(
+            role="system",
+            content=system_prompt,
+        )
+        messages = [system_message] + messages
+
     # Loop to handle tool calls and continue conversation
     while True:
         try:
@@ -358,7 +578,7 @@ async def _stream_chat_chunks(
             # Create the stream with proper types
             stream = await client.chat.completions.create(
                 model=model,
-                messages=session.to_openai_messages(),
+                messages=messages,
                 tools=tools,
                 tool_choice="auto",
                 stream=True,
@@ -502,8 +722,12 @@ async def _yield_tool_call(
     """
     logger.info(f"Yielding tool call: {tool_calls[yield_idx]}")
 
-    # Parse tool call arguments - exceptions will propagate to caller
-    arguments = orjson.loads(tool_calls[yield_idx]["function"]["arguments"])
+    # Parse tool call arguments - handle empty arguments gracefully
+    raw_arguments = tool_calls[yield_idx]["function"]["arguments"]
+    if raw_arguments:
+        arguments = orjson.loads(raw_arguments)
+    else:
+        arguments = {}
 
     yield StreamToolCall(
         tool_id=tool_calls[yield_idx]["id"],

autogpt_platform/backend/backend/api/features/chat/tools/__init__.py

@@ -4,21 +4,45 @@ from openai.types.chat import ChatCompletionToolParam
 
 from backend.api.features.chat.model import ChatSession
 
+from .add_understanding import AddUnderstandingTool
+from .agent_output import AgentOutputTool
 from .base import BaseTool
+from .create_agent import CreateAgentTool
+from .edit_agent import EditAgentTool
 from .find_agent import FindAgentTool
+from .find_block import FindBlockTool
+from .find_library_agent import FindLibraryAgentTool
 from .run_agent import RunAgentTool
+from .run_block import RunBlockTool
+from .search_docs import SearchDocsTool
 
 if TYPE_CHECKING:
     from backend.api.features.chat.response_model import StreamToolExecutionResult
 
 # Initialize tool instances
+add_understanding_tool = AddUnderstandingTool()
+create_agent_tool = CreateAgentTool()
+edit_agent_tool = EditAgentTool()
 find_agent_tool = FindAgentTool()
+find_block_tool = FindBlockTool()
+find_library_agent_tool = FindLibraryAgentTool()
 run_agent_tool = RunAgentTool()
+run_block_tool = RunBlockTool()
+search_docs_tool = SearchDocsTool()
+agent_output_tool = AgentOutputTool()
 
 # Export tools as OpenAI format
 tools: list[ChatCompletionToolParam] = [
+    add_understanding_tool.as_openai_tool(),
+    create_agent_tool.as_openai_tool(),
+    edit_agent_tool.as_openai_tool(),
     find_agent_tool.as_openai_tool(),
+    find_block_tool.as_openai_tool(),
+    find_library_agent_tool.as_openai_tool(),
     run_agent_tool.as_openai_tool(),
+    run_block_tool.as_openai_tool(),
+    search_docs_tool.as_openai_tool(),
+    agent_output_tool.as_openai_tool(),
 ]
 
 
@@ -31,8 +55,16 @@ async def execute_tool(
 ) -> "StreamToolExecutionResult":
 
     tool_map: dict[str, BaseTool] = {
+        "add_understanding": add_understanding_tool,
+        "create_agent": create_agent_tool,
+        "edit_agent": edit_agent_tool,
         "find_agent": find_agent_tool,
+        "find_block": find_block_tool,
+        "find_library_agent": find_library_agent_tool,
         "run_agent": run_agent_tool,
+        "run_block": run_block_tool,
+        "search_platform_docs": search_docs_tool,
+        "agent_output": agent_output_tool,
     }
     if tool_name not in tool_map:
         raise ValueError(f"Tool {tool_name} not found")

autogpt_platform/backend/backend/api/features/chat/tools/add_understanding.py

@@ -0,0 +1,206 @@
+"""Tool for capturing user business understanding incrementally."""
+
+import logging
+from typing import Any
+
+from backend.api.features.chat.model import ChatSession
+from backend.data.understanding import (
+    BusinessUnderstandingInput,
+    upsert_business_understanding,
+)
+
+from .base import BaseTool
+from .models import (
+    ErrorResponse,
+    ToolResponseBase,
+    UnderstandingUpdatedResponse,
+)
+
+logger = logging.getLogger(__name__)
+
+
+class AddUnderstandingTool(BaseTool):
+    """Tool for capturing user's business understanding incrementally."""
+
+    @property
+    def name(self) -> str:
+        return "add_understanding"
+
+    @property
+    def description(self) -> str:
+        return """Capture and store information about the user's business context,
+workflows, pain points, and automation goals. Call this tool whenever the user
+shares information about their business. Each call incrementally adds to the
+existing understanding - you don't need to provide all fields at once.
+
+Use this to build a comprehensive profile that helps recommend better agents
+and automations for the user's specific needs."""
+
+    @property
+    def parameters(self) -> dict[str, Any]:
+        return {
+            "type": "object",
+            "properties": {
+                "user_name": {
+                    "type": "string",
+                    "description": "The user's name",
+                },
+                "job_title": {
+                    "type": "string",
+                    "description": "The user's job title (e.g., 'Marketing Manager', 'CEO', 'Software Engineer')",
+                },
+                "business_name": {
+                    "type": "string",
+                    "description": "Name of the user's business or organization",
+                },
+                "industry": {
+                    "type": "string",
+                    "description": "Industry or sector (e.g., 'e-commerce', 'healthcare', 'finance')",
+                },
+                "business_size": {
+                    "type": "string",
+                    "description": "Company size: '1-10', '11-50', '51-200', '201-1000', or '1000+'",
+                },
+                "user_role": {
+                    "type": "string",
+                    "description": "User's role in organization context (e.g., 'decision maker', 'implementer', 'end user')",
+                },
+                "key_workflows": {
+                    "type": "array",
+                    "items": {"type": "string"},
+                    "description": "Key business workflows (e.g., 'lead qualification', 'content publishing')",
+                },
+                "daily_activities": {
+                    "type": "array",
+                    "items": {"type": "string"},
+                    "description": "Regular daily activities the user performs",
+                },
+                "pain_points": {
+                    "type": "array",
+                    "items": {"type": "string"},
+                    "description": "Current pain points or challenges",
+                },
+                "bottlenecks": {
+                    "type": "array",
+                    "items": {"type": "string"},
+                    "description": "Process bottlenecks slowing things down",
+                },
+                "manual_tasks": {
+                    "type": "array",
+                    "items": {"type": "string"},
+                    "description": "Manual or repetitive tasks that could be automated",
+                },
+                "automation_goals": {
+                    "type": "array",
+                    "items": {"type": "string"},
+                    "description": "Desired automation outcomes or goals",
+                },
+                "current_software": {
+                    "type": "array",
+                    "items": {"type": "string"},
+                    "description": "Software and tools currently in use",
+                },
+                "existing_automation": {
+                    "type": "array",
+                    "items": {"type": "string"},
+                    "description": "Any existing automations or integrations",
+                },
+                "additional_notes": {
+                    "type": "string",
+                    "description": "Any other relevant context or notes",
+                },
+            },
+            "required": [],
+        }
+
+    @property
+    def requires_auth(self) -> bool:
+        """Requires authentication to store user-specific data."""
+        return True
+
+    async def _execute(
+        self,
+        user_id: str | None,
+        session: ChatSession,
+        **kwargs,
+    ) -> ToolResponseBase:
+        """
+        Capture and store business understanding incrementally.
+
+        Each call merges new data with existing understanding:
+        - String fields are overwritten if provided
+        - List fields are appended (with deduplication)
+        """
+        session_id = session.session_id
+
+        if not user_id:
+            return ErrorResponse(
+                message="Authentication required to save business understanding.",
+                session_id=session_id,
+            )
+
+        # Check if any data was provided
+        if not any(v is not None for v in kwargs.values()):
+            return ErrorResponse(
+                message="Please provide at least one field to update.",
+                session_id=session_id,
+            )
+
+        # Build input model
+        input_data = BusinessUnderstandingInput(
+            user_name=kwargs.get("user_name"),
+            job_title=kwargs.get("job_title"),
+            business_name=kwargs.get("business_name"),
+            industry=kwargs.get("industry"),
+            business_size=kwargs.get("business_size"),
+            user_role=kwargs.get("user_role"),
+            key_workflows=kwargs.get("key_workflows"),
+            daily_activities=kwargs.get("daily_activities"),
+            pain_points=kwargs.get("pain_points"),
+            bottlenecks=kwargs.get("bottlenecks"),
+            manual_tasks=kwargs.get("manual_tasks"),
+            automation_goals=kwargs.get("automation_goals"),
+            current_software=kwargs.get("current_software"),
+            existing_automation=kwargs.get("existing_automation"),
+            additional_notes=kwargs.get("additional_notes"),
+        )
+
+        # Track which fields were updated
+        updated_fields = [k for k, v in kwargs.items() if v is not None]
+
+        # Upsert with merge
+        understanding = await upsert_business_understanding(user_id, input_data)
+
+        # Build current understanding summary for the response
+        current_understanding = {
+            "user_name": understanding.user_name,
+            "job_title": understanding.job_title,
+            "business_name": understanding.business_name,
+            "industry": understanding.industry,
+            "business_size": understanding.business_size,
+            "user_role": understanding.user_role,
+            "key_workflows": understanding.key_workflows,
+            "daily_activities": understanding.daily_activities,
+            "pain_points": understanding.pain_points,
+            "bottlenecks": understanding.bottlenecks,
+            "manual_tasks": understanding.manual_tasks,
+            "automation_goals": understanding.automation_goals,
+            "current_software": understanding.current_software,
+            "existing_automation": understanding.existing_automation,
+            "additional_notes": understanding.additional_notes,
+        }
+
+        # Filter out empty values for cleaner response
+        current_understanding = {
+            k: v
+            for k, v in current_understanding.items()
+            if v is not None and v != [] and v != ""
+        }
+
+        return UnderstandingUpdatedResponse(
+            message=f"Updated understanding with: {', '.join(updated_fields)}. "
+            "I now have a better picture of your business context.",
+            session_id=session_id,
+            updated_fields=updated_fields,
+            current_understanding=current_understanding,
+        )

autogpt_platform/backend/backend/api/features/chat/tools/agent_generator/__init__.py

@@ -0,0 +1,29 @@
+"""Agent generator package - Creates agents from natural language."""
+
+from .core import (
+    apply_agent_patch,
+    decompose_goal,
+    generate_agent,
+    generate_agent_patch,
+    get_agent_as_json,
+    save_agent_to_library,
+)
+from .fixer import apply_all_fixes
+from .utils import get_blocks_info
+from .validator import validate_agent
+
+__all__ = [
+    # Core functions
+    "decompose_goal",
+    "generate_agent",
+    "generate_agent_patch",
+    "apply_agent_patch",
+    "save_agent_to_library",
+    "get_agent_as_json",
+    # Fixer
+    "apply_all_fixes",
+    # Validator
+    "validate_agent",
+    # Utils
+    "get_blocks_info",
+]

autogpt_platform/backend/backend/api/features/chat/tools/agent_generator/client.py

@@ -0,0 +1,25 @@
+"""OpenRouter client configuration for agent generation."""
+
+import os
+
+from openai import AsyncOpenAI
+
+# Configuration - use OPEN_ROUTER_API_KEY for consistency with chat/config.py
+OPENROUTER_API_KEY = os.getenv("OPEN_ROUTER_API_KEY") or os.getenv("OPENROUTER_API_KEY")
+AGENT_GENERATOR_MODEL = os.getenv("AGENT_GENERATOR_MODEL", "anthropic/claude-opus-4.5")
+
+# OpenRouter client (OpenAI-compatible API)
+_client: AsyncOpenAI | None = None
+
+
+def get_client() -> AsyncOpenAI:
+    """Get or create the OpenRouter client."""
+    global _client
+    if _client is None:
+        if not OPENROUTER_API_KEY:
+            raise ValueError("OPENROUTER_API_KEY environment variable is required")
+        _client = AsyncOpenAI(
+            base_url="https://openrouter.ai/api/v1",
+            api_key=OPENROUTER_API_KEY,
+        )
+    return _client

autogpt_platform/backend/backend/api/features/chat/tools/agent_generator/core.py

@@ -0,0 +1,390 @@
+"""Core agent generation functions."""
+
+import copy
+import json
+import logging
+import uuid
+from typing import Any
+
+from backend.api.features.library import db as library_db
+from backend.data.graph import Graph, Link, Node, create_graph
+
+from .client import AGENT_GENERATOR_MODEL, get_client
+from .prompts import DECOMPOSITION_PROMPT, GENERATION_PROMPT, PATCH_PROMPT
+from .utils import get_block_summaries, parse_json_from_llm
+
+logger = logging.getLogger(__name__)
+
+
+async def decompose_goal(description: str, context: str = "") -> dict[str, Any] | None:
+    """Break down a goal into steps or return clarifying questions.
+
+    Args:
+        description: Natural language goal description
+        context: Additional context (e.g., answers to previous questions)
+
+    Returns:
+        Dict with either:
+        - {"type": "clarifying_questions", "questions": [...]}
+        - {"type": "instructions", "steps": [...]}
+        Or None on error
+    """
+    client = get_client()
+    prompt = DECOMPOSITION_PROMPT.format(block_summaries=get_block_summaries())
+
+    full_description = description
+    if context:
+        full_description = f"{description}\n\nAdditional context:\n{context}"
+
+    try:
+        response = await client.chat.completions.create(
+            model=AGENT_GENERATOR_MODEL,
+            messages=[
+                {"role": "system", "content": prompt},
+                {"role": "user", "content": full_description},
+            ],
+            temperature=0,
+        )
+
+        content = response.choices[0].message.content
+        if content is None:
+            logger.error("LLM returned empty content for decomposition")
+            return None
+
+        result = parse_json_from_llm(content)
+
+        if result is None:
+            logger.error(f"Failed to parse decomposition response: {content[:200]}")
+            return None
+
+        return result
+
+    except Exception as e:
+        logger.error(f"Error decomposing goal: {e}")
+        return None
+
+
+async def generate_agent(instructions: dict[str, Any]) -> dict[str, Any] | None:
+    """Generate agent JSON from instructions.
+
+    Args:
+        instructions: Structured instructions from decompose_goal
+
+    Returns:
+        Agent JSON dict or None on error
+    """
+    client = get_client()
+    prompt = GENERATION_PROMPT.format(block_summaries=get_block_summaries())
+
+    try:
+        response = await client.chat.completions.create(
+            model=AGENT_GENERATOR_MODEL,
+            messages=[
+                {"role": "system", "content": prompt},
+                {"role": "user", "content": json.dumps(instructions, indent=2)},
+            ],
+            temperature=0,
+        )
+
+        content = response.choices[0].message.content
+        if content is None:
+            logger.error("LLM returned empty content for agent generation")
+            return None
+
+        result = parse_json_from_llm(content)
+
+        if result is None:
+            logger.error(f"Failed to parse agent JSON: {content[:200]}")
+            return None
+
+        # Ensure required fields
+        if "id" not in result:
+            result["id"] = str(uuid.uuid4())
+        if "version" not in result:
+            result["version"] = 1
+        if "is_active" not in result:
+            result["is_active"] = True
+
+        return result
+
+    except Exception as e:
+        logger.error(f"Error generating agent: {e}")
+        return None
+
+
+def json_to_graph(agent_json: dict[str, Any]) -> Graph:
+    """Convert agent JSON dict to Graph model.
+
+    Args:
+        agent_json: Agent JSON with nodes and links
+
+    Returns:
+        Graph ready for saving
+    """
+    nodes = []
+    for n in agent_json.get("nodes", []):
+        node = Node(
+            id=n.get("id", str(uuid.uuid4())),
+            block_id=n["block_id"],
+            input_default=n.get("input_default", {}),
+            metadata=n.get("metadata", {}),
+        )
+        nodes.append(node)
+
+    links = []
+    for link_data in agent_json.get("links", []):
+        link = Link(
+            id=link_data.get("id", str(uuid.uuid4())),
+            source_id=link_data["source_id"],
+            sink_id=link_data["sink_id"],
+            source_name=link_data["source_name"],
+            sink_name=link_data["sink_name"],
+            is_static=link_data.get("is_static", False),
+        )
+        links.append(link)
+
+    return Graph(
+        id=agent_json.get("id", str(uuid.uuid4())),
+        version=agent_json.get("version", 1),
+        is_active=agent_json.get("is_active", True),
+        name=agent_json.get("name", "Generated Agent"),
+        description=agent_json.get("description", ""),
+        nodes=nodes,
+        links=links,
+    )
+
+
+def _reassign_node_ids(graph: Graph) -> None:
+    """Reassign all node and link IDs to new UUIDs.
+
+    This is needed when creating a new version to avoid unique constraint violations.
+    """
+    # Create mapping from old node IDs to new UUIDs
+    id_map = {node.id: str(uuid.uuid4()) for node in graph.nodes}
+
+    # Reassign node IDs
+    for node in graph.nodes:
+        node.id = id_map[node.id]
+
+    # Update link references to use new node IDs
+    for link in graph.links:
+        link.id = str(uuid.uuid4())  # Also give links new IDs
+        if link.source_id in id_map:
+            link.source_id = id_map[link.source_id]
+        if link.sink_id in id_map:
+            link.sink_id = id_map[link.sink_id]
+
+
+async def save_agent_to_library(
+    agent_json: dict[str, Any], user_id: str, is_update: bool = False
+) -> tuple[Graph, Any]:
+    """Save agent to database and user's library.
+
+    Args:
+        agent_json: Agent JSON dict
+        user_id: User ID
+        is_update: Whether this is an update to an existing agent
+
+    Returns:
+        Tuple of (created Graph, LibraryAgent)
+    """
+    from backend.data.graph import get_graph_all_versions
+
+    graph = json_to_graph(agent_json)
+
+    if is_update:
+        # For updates, keep the same graph ID but increment version
+        # and reassign node/link IDs to avoid conflicts
+        if graph.id:
+            existing_versions = await get_graph_all_versions(graph.id, user_id)
+            if existing_versions:
+                latest_version = max(v.version for v in existing_versions)
+                graph.version = latest_version + 1
+                # Reassign node IDs (but keep graph ID the same)
+                _reassign_node_ids(graph)
+                logger.info(f"Updating agent {graph.id} to version {graph.version}")
+    else:
+        # For new agents, always generate a fresh UUID to avoid collisions
+        graph.id = str(uuid.uuid4())
+        graph.version = 1
+        # Reassign all node IDs as well
+        _reassign_node_ids(graph)
+        logger.info(f"Creating new agent with ID {graph.id}")
+
+    # Save to database
+    created_graph = await create_graph(graph, user_id)
+
+    # Add to user's library (or update existing library agent)
+    library_agents = await library_db.create_library_agent(
+        graph=created_graph,
+        user_id=user_id,
+        create_library_agents_for_sub_graphs=False,
+    )
+
+    return created_graph, library_agents[0]
+
+
+async def get_agent_as_json(
+    graph_id: str, user_id: str | None
+) -> dict[str, Any] | None:
+    """Fetch an agent and convert to JSON format for editing.
+
+    Args:
+        graph_id: Graph ID or library agent ID
+        user_id: User ID
+
+    Returns:
+        Agent as JSON dict or None if not found
+    """
+    from backend.data.graph import get_graph
+
+    # Try to get the graph (version=None gets the active version)
+    graph = await get_graph(graph_id, version=None, user_id=user_id)
+    if not graph:
+        return None
+
+    # Convert to JSON format
+    nodes = []
+    for node in graph.nodes:
+        nodes.append(
+            {
+                "id": node.id,
+                "block_id": node.block_id,
+                "input_default": node.input_default,
+                "metadata": node.metadata,
+            }
+        )
+
+    links = []
+    for node in graph.nodes:
+        for link in node.output_links:
+            links.append(
+                {
+                    "id": link.id,
+                    "source_id": link.source_id,
+                    "sink_id": link.sink_id,
+                    "source_name": link.source_name,
+                    "sink_name": link.sink_name,
+                    "is_static": link.is_static,
+                }
+            )
+
+    return {
+        "id": graph.id,
+        "name": graph.name,
+        "description": graph.description,
+        "version": graph.version,
+        "is_active": graph.is_active,
+        "nodes": nodes,
+        "links": links,
+    }
+
+
+async def generate_agent_patch(
+    update_request: str, current_agent: dict[str, Any]
+) -> dict[str, Any] | None:
+    """Generate a patch to update an existing agent.
+
+    Args:
+        update_request: Natural language description of changes
+        current_agent: Current agent JSON
+
+    Returns:
+        Patch dict or clarifying questions, or None on error
+    """
+    client = get_client()
+    prompt = PATCH_PROMPT.format(
+        current_agent=json.dumps(current_agent, indent=2),
+        block_summaries=get_block_summaries(),
+    )
+
+    try:
+        response = await client.chat.completions.create(
+            model=AGENT_GENERATOR_MODEL,
+            messages=[
+                {"role": "system", "content": prompt},
+                {"role": "user", "content": update_request},
+            ],
+            temperature=0,
+        )
+
+        content = response.choices[0].message.content
+        if content is None:
+            logger.error("LLM returned empty content for patch generation")
+            return None
+
+        return parse_json_from_llm(content)
+
+    except Exception as e:
+        logger.error(f"Error generating patch: {e}")
+        return None
+
+
+def apply_agent_patch(
+    current_agent: dict[str, Any], patch: dict[str, Any]
+) -> dict[str, Any]:
+    """Apply a patch to an existing agent.
+
+    Args:
+        current_agent: Current agent JSON
+        patch: Patch dict with operations
+
+    Returns:
+        Updated agent JSON
+    """
+    agent = copy.deepcopy(current_agent)
+    patches = patch.get("patches", [])
+
+    for p in patches:
+        patch_type = p.get("type")
+
+        if patch_type == "modify":
+            node_id = p.get("node_id")
+            changes = p.get("changes", {})
+
+            for node in agent.get("nodes", []):
+                if node["id"] == node_id:
+                    _deep_update(node, changes)
+                    logger.debug(f"Modified node {node_id}")
+                    break
+
+        elif patch_type == "add":
+            new_nodes = p.get("new_nodes", [])
+            new_links = p.get("new_links", [])
+
+            agent["nodes"] = agent.get("nodes", []) + new_nodes
+            agent["links"] = agent.get("links", []) + new_links
+            logger.debug(f"Added {len(new_nodes)} nodes, {len(new_links)} links")
+
+        elif patch_type == "remove":
+            node_ids_to_remove = set(p.get("node_ids", []))
+            link_ids_to_remove = set(p.get("link_ids", []))
+
+            # Remove nodes
+            agent["nodes"] = [
+                n for n in agent.get("nodes", []) if n["id"] not in node_ids_to_remove
+            ]
+
+            # Remove links (both explicit and those referencing removed nodes)
+            agent["links"] = [
+                link
+                for link in agent.get("links", [])
+                if link["id"] not in link_ids_to_remove
+                and link["source_id"] not in node_ids_to_remove
+                and link["sink_id"] not in node_ids_to_remove
+            ]
+
+            logger.debug(
+                f"Removed {len(node_ids_to_remove)} nodes, {len(link_ids_to_remove)} links"
+            )
+
+    return agent
+
+
+def _deep_update(target: dict, source: dict) -> None:
+    """Recursively update a dict with another dict."""
+    for key, value in source.items():
+        if key in target and isinstance(target[key], dict) and isinstance(value, dict):
+            _deep_update(target[key], value)
+        else:
+            target[key] = value

autogpt_platform/backend/backend/api/features/chat/tools/agent_generator/fixer.py

@@ -0,0 +1,606 @@
+"""Agent fixer - Fixes common LLM generation errors."""
+
+import logging
+import re
+import uuid
+from typing import Any
+
+from .utils import (
+    ADDTODICTIONARY_BLOCK_ID,
+    ADDTOLIST_BLOCK_ID,
+    CODE_EXECUTION_BLOCK_ID,
+    CONDITION_BLOCK_ID,
+    CREATEDICT_BLOCK_ID,
+    CREATELIST_BLOCK_ID,
+    DATA_SAMPLING_BLOCK_ID,
+    DOUBLE_CURLY_BRACES_BLOCK_IDS,
+    GET_CURRENT_DATE_BLOCK_ID,
+    STORE_VALUE_BLOCK_ID,
+    UNIVERSAL_TYPE_CONVERTER_BLOCK_ID,
+    get_blocks_info,
+    is_valid_uuid,
+)
+
+logger = logging.getLogger(__name__)
+
+
+def fix_agent_ids(agent: dict[str, Any]) -> dict[str, Any]:
+    """Fix invalid UUIDs in agent and link IDs."""
+    # Fix agent ID
+    if not is_valid_uuid(agent.get("id", "")):
+        agent["id"] = str(uuid.uuid4())
+        logger.debug(f"Fixed agent ID: {agent['id']}")
+
+    # Fix node IDs
+    id_mapping = {}  # Old ID -> New ID
+    for node in agent.get("nodes", []):
+        if not is_valid_uuid(node.get("id", "")):
+            old_id = node.get("id", "")
+            new_id = str(uuid.uuid4())
+            id_mapping[old_id] = new_id
+            node["id"] = new_id
+            logger.debug(f"Fixed node ID: {old_id} -> {new_id}")
+
+    # Fix link IDs and update references
+    for link in agent.get("links", []):
+        if not is_valid_uuid(link.get("id", "")):
+            link["id"] = str(uuid.uuid4())
+            logger.debug(f"Fixed link ID: {link['id']}")
+
+        # Update source/sink IDs if they were remapped
+        if link.get("source_id") in id_mapping:
+            link["source_id"] = id_mapping[link["source_id"]]
+        if link.get("sink_id") in id_mapping:
+            link["sink_id"] = id_mapping[link["sink_id"]]
+
+    return agent
+
+
+def fix_double_curly_braces(agent: dict[str, Any]) -> dict[str, Any]:
+    """Fix single curly braces to double in template blocks."""
+    for node in agent.get("nodes", []):
+        if node.get("block_id") not in DOUBLE_CURLY_BRACES_BLOCK_IDS:
+            continue
+
+        input_data = node.get("input_default", {})
+        for key in ("prompt", "format"):
+            if key in input_data and isinstance(input_data[key], str):
+                original = input_data[key]
+                # Fix simple variable references: {var} -> {{var}}
+                fixed = re.sub(
+                    r"(?<!\{)\{([a-zA-Z_][a-zA-Z0-9_]*)\}(?!\})",
+                    r"{{\1}}",
+                    original,
+                )
+                if fixed != original:
+                    input_data[key] = fixed
+                    logger.debug(f"Fixed curly braces in {key}")
+
+    return agent
+
+
+def fix_storevalue_before_condition(agent: dict[str, Any]) -> dict[str, Any]:
+    """Add StoreValueBlock before ConditionBlock if needed for value2."""
+    nodes = agent.get("nodes", [])
+    links = agent.get("links", [])
+
+    # Find all ConditionBlock nodes
+    condition_node_ids = {
+        node["id"] for node in nodes if node.get("block_id") == CONDITION_BLOCK_ID
+    }
+
+    if not condition_node_ids:
+        return agent
+
+    new_nodes = []
+    new_links = []
+    processed_conditions = set()
+
+    for link in links:
+        sink_id = link.get("sink_id")
+        sink_name = link.get("sink_name")
+
+        # Check if this link goes to a ConditionBlock's value2
+        if sink_id in condition_node_ids and sink_name == "value2":
+            source_node = next(
+                (n for n in nodes if n["id"] == link.get("source_id")), None
+            )
+
+            # Skip if source is already a StoreValueBlock
+            if source_node and source_node.get("block_id") == STORE_VALUE_BLOCK_ID:
+                continue
+
+            # Skip if we already processed this condition
+            if sink_id in processed_conditions:
+                continue
+
+            processed_conditions.add(sink_id)
+
+            # Create StoreValueBlock
+            store_node_id = str(uuid.uuid4())
+            store_node = {
+                "id": store_node_id,
+                "block_id": STORE_VALUE_BLOCK_ID,
+                "input_default": {"data": None},
+                "metadata": {"position": {"x": 0, "y": -100}},
+            }
+            new_nodes.append(store_node)
+
+            # Create link: original source -> StoreValueBlock
+            new_links.append(
+                {
+                    "id": str(uuid.uuid4()),
+                    "source_id": link["source_id"],
+                    "source_name": link["source_name"],
+                    "sink_id": store_node_id,
+                    "sink_name": "input",
+                    "is_static": False,
+                }
+            )
+
+            # Update original link: StoreValueBlock -> ConditionBlock
+            link["source_id"] = store_node_id
+            link["source_name"] = "output"
+
+            logger.debug(f"Added StoreValueBlock before ConditionBlock {sink_id}")
+
+    if new_nodes:
+        agent["nodes"] = nodes + new_nodes
+
+    return agent
+
+
+def fix_addtolist_blocks(agent: dict[str, Any]) -> dict[str, Any]:
+    """Fix AddToList blocks by adding prerequisite empty AddToList block.
+
+    When an AddToList block is found:
+    1. Checks if there's a CreateListBlock before it
+    2. Removes CreateListBlock if linked directly to AddToList
+    3. Adds an empty AddToList block before the original
+    4. Ensures the original has a self-referencing link
+    """
+    nodes = agent.get("nodes", [])
+    links = agent.get("links", [])
+    new_nodes = []
+    original_addtolist_ids = set()
+    nodes_to_remove = set()
+    links_to_remove = []
+
+    # First pass: identify CreateListBlock nodes to remove
+    for link in links:
+        source_node = next(
+            (n for n in nodes if n.get("id") == link.get("source_id")), None
+        )
+        sink_node = next((n for n in nodes if n.get("id") == link.get("sink_id")), None)
+
+        if (
+            source_node
+            and sink_node
+            and source_node.get("block_id") == CREATELIST_BLOCK_ID
+            and sink_node.get("block_id") == ADDTOLIST_BLOCK_ID
+        ):
+            nodes_to_remove.add(source_node.get("id"))
+            links_to_remove.append(link)
+            logger.debug(f"Removing CreateListBlock {source_node.get('id')}")
+
+    # Second pass: process AddToList blocks
+    filtered_nodes = []
+    for node in nodes:
+        if node.get("id") in nodes_to_remove:
+            continue
+
+        if node.get("block_id") == ADDTOLIST_BLOCK_ID:
+            original_addtolist_ids.add(node.get("id"))
+            node_id = node.get("id")
+            pos = node.get("metadata", {}).get("position", {"x": 0, "y": 0})
+
+            # Check if already has prerequisite
+            has_prereq = any(
+                link.get("sink_id") == node_id
+                and link.get("sink_name") == "list"
+                and link.get("source_name") == "updated_list"
+                for link in links
+            )
+
+            if not has_prereq:
+                # Remove links to "list" input (except self-reference)
+                for link in links:
+                    if (
+                        link.get("sink_id") == node_id
+                        and link.get("sink_name") == "list"
+                        and link.get("source_id") != node_id
+                        and link not in links_to_remove
+                    ):
+                        links_to_remove.append(link)
+
+                # Create prerequisite AddToList block
+                prereq_id = str(uuid.uuid4())
+                prereq_node = {
+                    "id": prereq_id,
+                    "block_id": ADDTOLIST_BLOCK_ID,
+                    "input_default": {"list": [], "entry": None, "entries": []},
+                    "metadata": {
+                        "position": {"x": pos.get("x", 0) - 800, "y": pos.get("y", 0)}
+                    },
+                }
+                new_nodes.append(prereq_node)
+
+                # Link prerequisite to original
+                links.append(
+                    {
+                        "id": str(uuid.uuid4()),
+                        "source_id": prereq_id,
+                        "source_name": "updated_list",
+                        "sink_id": node_id,
+                        "sink_name": "list",
+                        "is_static": False,
+                    }
+                )
+                logger.debug(f"Added prerequisite AddToList block for {node_id}")
+
+        filtered_nodes.append(node)
+
+    # Remove marked links
+    filtered_links = [link for link in links if link not in links_to_remove]
+
+    # Add self-referencing links for original AddToList blocks
+    for node in filtered_nodes + new_nodes:
+        if (
+            node.get("block_id") == ADDTOLIST_BLOCK_ID
+            and node.get("id") in original_addtolist_ids
+        ):
+            node_id = node.get("id")
+            has_self_ref = any(
+                link["source_id"] == node_id
+                and link["sink_id"] == node_id
+                and link["source_name"] == "updated_list"
+                and link["sink_name"] == "list"
+                for link in filtered_links
+            )
+            if not has_self_ref:
+                filtered_links.append(
+                    {
+                        "id": str(uuid.uuid4()),
+                        "source_id": node_id,
+                        "source_name": "updated_list",
+                        "sink_id": node_id,
+                        "sink_name": "list",
+                        "is_static": False,
+                    }
+                )
+                logger.debug(f"Added self-reference for AddToList {node_id}")
+
+    agent["nodes"] = filtered_nodes + new_nodes
+    agent["links"] = filtered_links
+    return agent
+
+
+def fix_addtodictionary_blocks(agent: dict[str, Any]) -> dict[str, Any]:
+    """Fix AddToDictionary blocks by removing empty CreateDictionary nodes."""
+    nodes = agent.get("nodes", [])
+    links = agent.get("links", [])
+    nodes_to_remove = set()
+    links_to_remove = []
+
+    for link in links:
+        source_node = next(
+            (n for n in nodes if n.get("id") == link.get("source_id")), None
+        )
+        sink_node = next((n for n in nodes if n.get("id") == link.get("sink_id")), None)
+
+        if (
+            source_node
+            and sink_node
+            and source_node.get("block_id") == CREATEDICT_BLOCK_ID
+            and sink_node.get("block_id") == ADDTODICTIONARY_BLOCK_ID
+        ):
+            nodes_to_remove.add(source_node.get("id"))
+            links_to_remove.append(link)
+            logger.debug(f"Removing CreateDictionary {source_node.get('id')}")
+
+    agent["nodes"] = [n for n in nodes if n.get("id") not in nodes_to_remove]
+    agent["links"] = [link for link in links if link not in links_to_remove]
+    return agent
+
+
+def fix_code_execution_output(agent: dict[str, Any]) -> dict[str, Any]:
+    """Fix CodeExecutionBlock output: change 'response' to 'stdout_logs'."""
+    nodes = agent.get("nodes", [])
+    links = agent.get("links", [])
+
+    for link in links:
+        source_node = next(
+            (n for n in nodes if n.get("id") == link.get("source_id")), None
+        )
+        if (
+            source_node
+            and source_node.get("block_id") == CODE_EXECUTION_BLOCK_ID
+            and link.get("source_name") == "response"
+        ):
+            link["source_name"] = "stdout_logs"
+            logger.debug("Fixed CodeExecutionBlock output: response -> stdout_logs")
+
+    return agent
+
+
+def fix_data_sampling_sample_size(agent: dict[str, Any]) -> dict[str, Any]:
+    """Fix DataSamplingBlock by setting sample_size to 1 as default."""
+    nodes = agent.get("nodes", [])
+    links = agent.get("links", [])
+    links_to_remove = []
+
+    for node in nodes:
+        if node.get("block_id") == DATA_SAMPLING_BLOCK_ID:
+            node_id = node.get("id")
+            input_default = node.get("input_default", {})
+
+            # Remove links to sample_size
+            for link in links:
+                if (
+                    link.get("sink_id") == node_id
+                    and link.get("sink_name") == "sample_size"
+                ):
+                    links_to_remove.append(link)
+
+            # Set default
+            input_default["sample_size"] = 1
+            node["input_default"] = input_default
+            logger.debug(f"Fixed DataSamplingBlock {node_id} sample_size to 1")
+
+    if links_to_remove:
+        agent["links"] = [link for link in links if link not in links_to_remove]
+
+    return agent
+
+
+def fix_node_x_coordinates(agent: dict[str, Any]) -> dict[str, Any]:
+    """Fix node x-coordinates to ensure 800+ unit spacing between linked nodes."""
+    nodes = agent.get("nodes", [])
+    links = agent.get("links", [])
+    node_lookup = {n.get("id"): n for n in nodes}
+
+    for link in links:
+        source_id = link.get("source_id")
+        sink_id = link.get("sink_id")
+
+        source_node = node_lookup.get(source_id)
+        sink_node = node_lookup.get(sink_id)
+
+        if not source_node or not sink_node:
+            continue
+
+        source_pos = source_node.get("metadata", {}).get("position", {})
+        sink_pos = sink_node.get("metadata", {}).get("position", {})
+
+        source_x = source_pos.get("x", 0)
+        sink_x = sink_pos.get("x", 0)
+
+        if abs(sink_x - source_x) < 800:
+            new_x = source_x + 800
+            if "metadata" not in sink_node:
+                sink_node["metadata"] = {}
+            if "position" not in sink_node["metadata"]:
+                sink_node["metadata"]["position"] = {}
+            sink_node["metadata"]["position"]["x"] = new_x
+            logger.debug(f"Fixed node {sink_id} x: {sink_x} -> {new_x}")
+
+    return agent
+
+
+def fix_getcurrentdate_offset(agent: dict[str, Any]) -> dict[str, Any]:
+    """Fix GetCurrentDateBlock offset to ensure it's positive."""
+    for node in agent.get("nodes", []):
+        if node.get("block_id") == GET_CURRENT_DATE_BLOCK_ID:
+            input_default = node.get("input_default", {})
+            if "offset" in input_default:
+                offset = input_default["offset"]
+                if isinstance(offset, (int, float)) and offset < 0:
+                    input_default["offset"] = abs(offset)
+                    logger.debug(f"Fixed offset: {offset} -> {abs(offset)}")
+
+    return agent
+
+
+def fix_ai_model_parameter(
+    agent: dict[str, Any],
+    blocks_info: list[dict[str, Any]],
+    default_model: str = "gpt-4o",
+) -> dict[str, Any]:
+    """Add default model parameter to AI blocks if missing."""
+    block_map = {b.get("id"): b for b in blocks_info}
+
+    for node in agent.get("nodes", []):
+        block_id = node.get("block_id")
+        block = block_map.get(block_id)
+
+        if not block:
+            continue
+
+        # Check if block has AI category
+        categories = block.get("categories", [])
+        is_ai_block = any(
+            cat.get("category") == "AI" for cat in categories if isinstance(cat, dict)
+        )
+
+        if is_ai_block:
+            input_default = node.get("input_default", {})
+            if "model" not in input_default:
+                input_default["model"] = default_model
+                node["input_default"] = input_default
+                logger.debug(
+                    f"Added model '{default_model}' to AI block {node.get('id')}"
+                )
+
+    return agent
+
+
+def fix_link_static_properties(
+    agent: dict[str, Any], blocks_info: list[dict[str, Any]]
+) -> dict[str, Any]:
+    """Fix is_static property based on source block's staticOutput."""
+    block_map = {b.get("id"): b for b in blocks_info}
+    node_lookup = {n.get("id"): n for n in agent.get("nodes", [])}
+
+    for link in agent.get("links", []):
+        source_node = node_lookup.get(link.get("source_id"))
+        if not source_node:
+            continue
+
+        source_block = block_map.get(source_node.get("block_id"))
+        if not source_block:
+            continue
+
+        static_output = source_block.get("staticOutput", False)
+        if link.get("is_static") != static_output:
+            link["is_static"] = static_output
+            logger.debug(f"Fixed link {link.get('id')} is_static to {static_output}")
+
+    return agent
+
+
+def fix_data_type_mismatch(
+    agent: dict[str, Any], blocks_info: list[dict[str, Any]]
+) -> dict[str, Any]:
+    """Fix data type mismatches by inserting UniversalTypeConverterBlock."""
+    nodes = agent.get("nodes", [])
+    links = agent.get("links", [])
+    block_map = {b.get("id"): b for b in blocks_info}
+    node_lookup = {n.get("id"): n for n in nodes}
+
+    def get_property_type(schema: dict, name: str) -> str | None:
+        if "_#_" in name:
+            parent, child = name.split("_#_", 1)
+            parent_schema = schema.get(parent, {})
+            if "properties" in parent_schema:
+                return parent_schema["properties"].get(child, {}).get("type")
+            return None
+        return schema.get(name, {}).get("type")
+
+    def are_types_compatible(src: str, sink: str) -> bool:
+        if {src, sink} <= {"integer", "number"}:
+            return True
+        return src == sink
+
+    type_mapping = {
+        "string": "string",
+        "text": "string",
+        "integer": "number",
+        "number": "number",
+        "float": "number",
+        "boolean": "boolean",
+        "bool": "boolean",
+        "array": "list",
+        "list": "list",
+        "object": "dictionary",
+        "dict": "dictionary",
+        "dictionary": "dictionary",
+    }
+
+    new_links = []
+    nodes_to_add = []
+
+    for link in links:
+        source_node = node_lookup.get(link.get("source_id"))
+        sink_node = node_lookup.get(link.get("sink_id"))
+
+        if not source_node or not sink_node:
+            new_links.append(link)
+            continue
+
+        source_block = block_map.get(source_node.get("block_id"))
+        sink_block = block_map.get(sink_node.get("block_id"))
+
+        if not source_block or not sink_block:
+            new_links.append(link)
+            continue
+
+        source_outputs = source_block.get("outputSchema", {}).get("properties", {})
+        sink_inputs = sink_block.get("inputSchema", {}).get("properties", {})
+
+        source_type = get_property_type(source_outputs, link.get("source_name", ""))
+        sink_type = get_property_type(sink_inputs, link.get("sink_name", ""))
+
+        if (
+            source_type
+            and sink_type
+            and not are_types_compatible(source_type, sink_type)
+        ):
+            # Insert type converter
+            converter_id = str(uuid.uuid4())
+            target_type = type_mapping.get(sink_type, sink_type)
+
+            converter_node = {
+                "id": converter_id,
+                "block_id": UNIVERSAL_TYPE_CONVERTER_BLOCK_ID,
+                "input_default": {"type": target_type},
+                "metadata": {"position": {"x": 0, "y": 100}},
+            }
+            nodes_to_add.append(converter_node)
+
+            # source -> converter
+            new_links.append(
+                {
+                    "id": str(uuid.uuid4()),
+                    "source_id": link["source_id"],
+                    "source_name": link["source_name"],
+                    "sink_id": converter_id,
+                    "sink_name": "value",
+                    "is_static": False,
+                }
+            )
+
+            # converter -> sink
+            new_links.append(
+                {
+                    "id": str(uuid.uuid4()),
+                    "source_id": converter_id,
+                    "source_name": "value",
+                    "sink_id": link["sink_id"],
+                    "sink_name": link["sink_name"],
+                    "is_static": False,
+                }
+            )
+
+            logger.debug(f"Inserted type converter: {source_type} -> {target_type}")
+        else:
+            new_links.append(link)
+
+    if nodes_to_add:
+        agent["nodes"] = nodes + nodes_to_add
+        agent["links"] = new_links
+
+    return agent
+
+
+def apply_all_fixes(
+    agent: dict[str, Any], blocks_info: list[dict[str, Any]] | None = None
+) -> dict[str, Any]:
+    """Apply all fixes to an agent JSON.
+
+    Args:
+        agent: Agent JSON dict
+        blocks_info: Optional list of block info dicts for advanced fixes
+
+    Returns:
+        Fixed agent JSON
+    """
+    # Basic fixes (no block info needed)
+    agent = fix_agent_ids(agent)
+    agent = fix_double_curly_braces(agent)
+    agent = fix_storevalue_before_condition(agent)
+    agent = fix_addtolist_blocks(agent)
+    agent = fix_addtodictionary_blocks(agent)
+    agent = fix_code_execution_output(agent)
+    agent = fix_data_sampling_sample_size(agent)
+    agent = fix_node_x_coordinates(agent)
+    agent = fix_getcurrentdate_offset(agent)
+
+    # Advanced fixes (require block info)
+    if blocks_info is None:
+        blocks_info = get_blocks_info()
+
+    agent = fix_ai_model_parameter(agent, blocks_info)
+    agent = fix_link_static_properties(agent, blocks_info)
+    agent = fix_data_type_mismatch(agent, blocks_info)
+
+    return agent

autogpt_platform/backend/backend/api/features/chat/tools/agent_generator/prompts.py

@@ -0,0 +1,225 @@
+"""Prompt templates for agent generation."""
+
+DECOMPOSITION_PROMPT = """
+You are an expert AutoGPT Workflow Decomposer. Your task is to analyze a user's high-level goal and break it down into a clear, step-by-step plan using the available blocks.
+
+Each step should represent a distinct, automatable action suitable for execution by an AI automation system.
+
+---
+
+FIRST: Analyze the user's goal and determine:
+1) Design-time configuration (fixed settings that won't change per run)
+2) Runtime inputs (values the agent's end-user will provide each time it runs)
+
+For anything that can vary per run (email addresses, names, dates, search terms, etc.):
+- DO NOT ask for the actual value
+- Instead, define it as an Agent Input with a clear name, type, and description
+
+Only ask clarifying questions about design-time config that affects how you build the workflow:
+- Which external service to use (e.g., "Gmail vs Outlook", "Notion vs Google Docs")
+- Required formats or structures (e.g., "CSV, JSON, or PDF output?")
+- Business rules that must be hard-coded
+
+IMPORTANT CLARIFICATIONS POLICY:
+- Ask no more than five essential questions
+- Do not ask for concrete values that can be provided at runtime as Agent Inputs
+- Do not ask for API keys or credentials; the platform handles those directly
+- If there is enough information to infer reasonable defaults, prefer to propose defaults
+
+---
+
+GUIDELINES:
+1. List each step as a numbered item
+2. Describe the action clearly and specify inputs/outputs
+3. Ensure steps are in logical, sequential order
+4. Mention block names naturally (e.g., "Use GetWeatherByLocationBlock to...")
+5. Help the user reach their goal efficiently
+
+---
+
+RULES:
+1. OUTPUT FORMAT: Only output either clarifying questions OR step-by-step instructions, not both
+2. USE ONLY THE BLOCKS PROVIDED
+3. ALL required_input fields must be provided
+4. Data types of linked properties must match
+5. Write expert-level prompts for AI-related blocks
+
+---
+
+CRITICAL BLOCK RESTRICTIONS:
+1. AddToListBlock: Outputs updated list EVERY addition, not after all additions
+2. SendEmailBlock: Draft the email for user review; set SMTP config based on email type
+3. ConditionBlock: value2 is reference, value1 is contrast
+4. CodeExecutionBlock: DO NOT USE - use AI blocks instead
+5. ReadCsvBlock: Only use the 'rows' output, not 'row'
+
+---
+
+OUTPUT FORMAT:
+
+If more information is needed:
+```json
+{{
+  "type": "clarifying_questions",
+  "questions": [
+    {{
+      "question": "Which email provider should be used? (Gmail, Outlook, custom SMTP)",
+      "keyword": "email_provider",
+      "example": "Gmail"
+    }}
+  ]
+}}
+```
+
+If ready to proceed:
+```json
+{{
+  "type": "instructions",
+  "steps": [
+    {{
+      "step_number": 1,
+      "block_name": "AgentShortTextInputBlock",
+      "description": "Get the URL of the content to analyze.",
+      "inputs": [{{"name": "name", "value": "URL"}}],
+      "outputs": [{{"name": "result", "description": "The URL entered by user"}}]
+    }}
+  ]
+}}
+```
+
+---
+
+AVAILABLE BLOCKS:
+{block_summaries}
+"""
+
+GENERATION_PROMPT = """
+You are an expert AI workflow builder. Generate a valid agent JSON from the given instructions.
+
+---
+
+NODES:
+Each node must include:
+- `id`: Unique UUID v4 (e.g. `a8f5b1e2-c3d4-4e5f-8a9b-0c1d2e3f4a5b`)
+- `block_id`: The block identifier (must match an Allowed Block)
+- `input_default`: Dict of inputs (can be empty if no static inputs needed)
+- `metadata`: Must contain:
+  - `position`: {{"x": number, "y": number}} - adjacent nodes should differ by 800+ in X
+  - `customized_name`: Clear name describing this block's purpose in the workflow
+
+---
+
+LINKS:
+Each link connects a source node's output to a sink node's input:
+- `id`: MUST be UUID v4 (NOT "link-1", "link-2", etc.)
+- `source_id`: ID of the source node
+- `source_name`: Output field name from the source block
+- `sink_id`: ID of the sink node
+- `sink_name`: Input field name on the sink block
+- `is_static`: true only if source block has static_output: true
+
+CRITICAL: All IDs must be valid UUID v4 format!
+
+---
+
+AGENT (GRAPH):
+Wrap nodes and links in:
+- `id`: UUID of the agent
+- `name`: Short, generic name (avoid specific company names, URLs)
+- `description`: Short, generic description
+- `nodes`: List of all nodes
+- `links`: List of all links
+- `version`: 1
+- `is_active`: true
+
+---
+
+TIPS:
+- All required_input fields must be provided via input_default or a valid link
+- Ensure consistent source_id and sink_id references
+- Avoid dangling links
+- Input/output pins must match block schemas
+- Do not invent unknown block_ids
+
+---
+
+ALLOWED BLOCKS:
+{block_summaries}
+
+---
+
+Generate the complete agent JSON. Output ONLY valid JSON, no explanation.
+"""
+
+PATCH_PROMPT = """
+You are an expert at modifying AutoGPT agent workflows. Given the current agent and a modification request, generate a JSON patch to update the agent.
+
+CURRENT AGENT:
+{current_agent}
+
+AVAILABLE BLOCKS:
+{block_summaries}
+
+---
+
+PATCH FORMAT:
+Return a JSON object with the following structure:
+
+```json
+{{
+  "type": "patch",
+  "intent": "Brief description of what the patch does",
+  "patches": [
+    {{
+      "type": "modify",
+      "node_id": "uuid-of-node-to-modify",
+      "changes": {{
+        "input_default": {{"field": "new_value"}},
+        "metadata": {{"customized_name": "New Name"}}
+      }}
+    }},
+    {{
+      "type": "add",
+      "new_nodes": [
+        {{
+          "id": "new-uuid",
+          "block_id": "block-uuid",
+          "input_default": {{}},
+          "metadata": {{"position": {{"x": 0, "y": 0}}, "customized_name": "Name"}}
+        }}
+      ],
+      "new_links": [
+        {{
+          "id": "link-uuid",
+          "source_id": "source-node-id",
+          "source_name": "output_field",
+          "sink_id": "sink-node-id",
+          "sink_name": "input_field"
+        }}
+      ]
+    }},
+    {{
+      "type": "remove",
+      "node_ids": ["uuid-of-node-to-remove"],
+      "link_ids": ["uuid-of-link-to-remove"]
+    }}
+  ]
+}}
+```
+
+If you need more information, return:
+```json
+{{
+  "type": "clarifying_questions",
+  "questions": [
+    {{
+      "question": "What specific change do you want?",
+      "keyword": "change_type",
+      "example": "Add error handling"
+    }}
+  ]
+}}
+```
+
+Generate the minimal patch needed. Output ONLY valid JSON.
+"""

autogpt_platform/backend/backend/api/features/chat/tools/agent_generator/utils.py

@@ -0,0 +1,213 @@
+"""Utilities for agent generation."""
+
+import json
+import re
+from typing import Any
+
+from backend.data.block import get_blocks
+
+# UUID validation regex
+UUID_REGEX = 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}$"
+)
+
+# Block IDs for various fixes
+STORE_VALUE_BLOCK_ID = "1ff065e9-88e8-4358-9d82-8dc91f622ba9"
+CONDITION_BLOCK_ID = "715696a0-e1da-45c8-b209-c2fa9c3b0be6"
+ADDTOLIST_BLOCK_ID = "aeb08fc1-2fc1-4141-bc8e-f758f183a822"
+ADDTODICTIONARY_BLOCK_ID = "31d1064e-7446-4693-a7d4-65e5ca1180d1"
+CREATELIST_BLOCK_ID = "a912d5c7-6e00-4542-b2a9-8034136930e4"
+CREATEDICT_BLOCK_ID = "b924ddf4-de4f-4b56-9a85-358930dcbc91"
+CODE_EXECUTION_BLOCK_ID = "0b02b072-abe7-11ef-8372-fb5d162dd712"
+DATA_SAMPLING_BLOCK_ID = "4a448883-71fa-49cf-91cf-70d793bd7d87"
+UNIVERSAL_TYPE_CONVERTER_BLOCK_ID = "95d1b990-ce13-4d88-9737-ba5c2070c97b"
+GET_CURRENT_DATE_BLOCK_ID = "b29c1b50-5d0e-4d9f-8f9d-1b0e6fcbf0b1"
+
+DOUBLE_CURLY_BRACES_BLOCK_IDS = [
+    "44f6c8ad-d75c-4ae1-8209-aad1c0326928",  # FillTextTemplateBlock
+    "6ab085e2-20b3-4055-bc3e-08036e01eca6",
+    "90f8c45e-e983-4644-aa0b-b4ebe2f531bc",
+    "363ae599-353e-4804-937e-b2ee3cef3da4",  # AgentOutputBlock
+    "3b191d9f-356f-482d-8238-ba04b6d18381",
+    "db7d8f02-2f44-4c55-ab7a-eae0941f0c30",
+    "3a7c4b8d-6e2f-4a5d-b9c1-f8d23c5a9b0e",
+    "ed1ae7a0-b770-4089-b520-1f0005fad19a",
+    "a892b8d9-3e4e-4e9c-9c1e-75f8efcf1bfa",
+    "b29c1b50-5d0e-4d9f-8f9d-1b0e6fcbf0b1",
+    "716a67b3-6760-42e7-86dc-18645c6e00fc",
+    "530cf046-2ce0-4854-ae2c-659db17c7a46",
+    "ed55ac19-356e-4243-a6cb-bc599e9b716f",
+    "1f292d4a-41a4-4977-9684-7c8d560b9f91",  # LLM blocks
+    "32a87eab-381e-4dd4-bdb8-4c47151be35a",
+]
+
+
+def is_valid_uuid(value: str) -> bool:
+    """Check if a string is a valid UUID v4."""
+    return isinstance(value, str) and UUID_REGEX.match(value) is not None
+
+
+def _compact_schema(schema: dict) -> dict[str, str]:
+    """Extract compact type info from a JSON schema properties dict.
+
+    Returns a dict of {field_name: type_string} for essential info only.
+    """
+    props = schema.get("properties", {})
+    result = {}
+
+    for name, prop in props.items():
+        # Skip internal/complex fields
+        if name.startswith("_"):
+            continue
+
+        # Get type string
+        type_str = prop.get("type", "any")
+
+        # Handle anyOf/oneOf (optional types)
+        if "anyOf" in prop:
+            types = [t.get("type", "?") for t in prop["anyOf"] if t.get("type")]
+            type_str = "|".join(types) if types else "any"
+        elif "allOf" in prop:
+            type_str = "object"
+
+        # Add array item type if present
+        if type_str == "array" and "items" in prop:
+            items = prop["items"]
+            if isinstance(items, dict):
+                item_type = items.get("type", "any")
+                type_str = f"array[{item_type}]"
+
+        result[name] = type_str
+
+    return result
+
+
+def get_block_summaries(include_schemas: bool = True) -> str:
+    """Generate compact block summaries for prompts.
+
+    Args:
+        include_schemas: Whether to include input/output type info
+
+    Returns:
+        Formatted string of block summaries (compact format)
+    """
+    blocks = get_blocks()
+    summaries = []
+
+    for block_id, block_cls in blocks.items():
+        block = block_cls()
+        name = block.name
+        desc = getattr(block, "description", "") or ""
+
+        # Truncate description
+        if len(desc) > 150:
+            desc = desc[:147] + "..."
+
+        if not include_schemas:
+            summaries.append(f"- {name} (id: {block_id}): {desc}")
+        else:
+            # Compact format with type info only
+            inputs = {}
+            outputs = {}
+            required = []
+
+            if hasattr(block, "input_schema"):
+                try:
+                    schema = block.input_schema.jsonschema()
+                    inputs = _compact_schema(schema)
+                    required = schema.get("required", [])
+                except Exception:
+                    pass
+
+            if hasattr(block, "output_schema"):
+                try:
+                    schema = block.output_schema.jsonschema()
+                    outputs = _compact_schema(schema)
+                except Exception:
+                    pass
+
+            # Build compact line format
+            # Format: NAME (id): desc | in: {field:type, ...} [required] | out: {field:type}
+            in_str = ", ".join(f"{k}:{v}" for k, v in inputs.items())
+            out_str = ", ".join(f"{k}:{v}" for k, v in outputs.items())
+            req_str = f" req=[{','.join(required)}]" if required else ""
+
+            static = " [static]" if getattr(block, "static_output", False) else ""
+
+            line = f"- {name} (id: {block_id}): {desc}"
+            if in_str:
+                line += f"\n  in: {{{in_str}}}{req_str}"
+            if out_str:
+                line += f"\n  out: {{{out_str}}}{static}"
+
+            summaries.append(line)
+
+    return "\n".join(summaries)
+
+
+def get_blocks_info() -> list[dict[str, Any]]:
+    """Get block information with schemas for validation and fixing."""
+    blocks = get_blocks()
+    blocks_info = []
+    for block_id, block_cls in blocks.items():
+        block = block_cls()
+        blocks_info.append(
+            {
+                "id": block_id,
+                "name": block.name,
+                "description": getattr(block, "description", ""),
+                "categories": getattr(block, "categories", []),
+                "staticOutput": getattr(block, "static_output", False),
+                "inputSchema": (
+                    block.input_schema.jsonschema()
+                    if hasattr(block, "input_schema")
+                    else {}
+                ),
+                "outputSchema": (
+                    block.output_schema.jsonschema()
+                    if hasattr(block, "output_schema")
+                    else {}
+                ),
+            }
+        )
+    return blocks_info
+
+
+def parse_json_from_llm(text: str) -> dict[str, Any] | None:
+    """Extract JSON from LLM response (handles markdown code blocks)."""
+    if not text:
+        return None
+
+    # Try fenced code block
+    match = re.search(r"```(?:json)?\s*([\s\S]*?)```", text, re.IGNORECASE)
+    if match:
+        try:
+            return json.loads(match.group(1).strip())
+        except json.JSONDecodeError:
+            pass
+
+    # Try raw text
+    try:
+        return json.loads(text.strip())
+    except json.JSONDecodeError:
+        pass
+
+    # Try finding {...} span
+    start = text.find("{")
+    end = text.rfind("}")
+    if start != -1 and end > start:
+        try:
+            return json.loads(text[start : end + 1])
+        except json.JSONDecodeError:
+            pass
+
+    # Try finding [...] span
+    start = text.find("[")
+    end = text.rfind("]")
+    if start != -1 and end > start:
+        try:
+            return json.loads(text[start : end + 1])
+        except json.JSONDecodeError:
+            pass
+
+    return None

autogpt_platform/backend/backend/api/features/chat/tools/agent_generator/validator.py

@@ -0,0 +1,279 @@
+"""Agent validator - Validates agent structure and connections."""
+
+import logging
+import re
+from typing import Any
+
+from .utils import get_blocks_info
+
+logger = logging.getLogger(__name__)
+
+
+class AgentValidator:
+    """Validator for AutoGPT agents with detailed error reporting."""
+
+    def __init__(self):
+        self.errors: list[str] = []
+
+    def add_error(self, error: str) -> None:
+        """Add an error message."""
+        self.errors.append(error)
+
+    def validate_block_existence(
+        self, agent: dict[str, Any], blocks_info: list[dict[str, Any]]
+    ) -> bool:
+        """Validate all block IDs exist in the blocks library."""
+        valid = True
+        valid_block_ids = {b.get("id") for b in blocks_info if b.get("id")}
+
+        for node in agent.get("nodes", []):
+            block_id = node.get("block_id")
+            node_id = node.get("id")
+
+            if not block_id:
+                self.add_error(f"Node '{node_id}' is missing 'block_id' field.")
+                valid = False
+                continue
+
+            if block_id not in valid_block_ids:
+                self.add_error(
+                    f"Node '{node_id}' references block_id '{block_id}' which does not exist."
+                )
+                valid = False
+
+        return valid
+
+    def validate_link_node_references(self, agent: dict[str, Any]) -> bool:
+        """Validate all node IDs referenced in links exist."""
+        valid = True
+        valid_node_ids = {n.get("id") for n in agent.get("nodes", []) if n.get("id")}
+
+        for link in agent.get("links", []):
+            link_id = link.get("id", "Unknown")
+            source_id = link.get("source_id")
+            sink_id = link.get("sink_id")
+
+            if not source_id:
+                self.add_error(f"Link '{link_id}' is missing 'source_id'.")
+                valid = False
+            elif source_id not in valid_node_ids:
+                self.add_error(
+                    f"Link '{link_id}' references non-existent source_id '{source_id}'."
+                )
+                valid = False
+
+            if not sink_id:
+                self.add_error(f"Link '{link_id}' is missing 'sink_id'.")
+                valid = False
+            elif sink_id not in valid_node_ids:
+                self.add_error(
+                    f"Link '{link_id}' references non-existent sink_id '{sink_id}'."
+                )
+                valid = False
+
+        return valid
+
+    def validate_required_inputs(
+        self, agent: dict[str, Any], blocks_info: list[dict[str, Any]]
+    ) -> bool:
+        """Validate required inputs are provided."""
+        valid = True
+        block_map = {b.get("id"): b for b in blocks_info}
+
+        for node in agent.get("nodes", []):
+            block_id = node.get("block_id")
+            block = block_map.get(block_id)
+
+            if not block:
+                continue
+
+            required_inputs = block.get("inputSchema", {}).get("required", [])
+            input_defaults = node.get("input_default", {})
+            node_id = node.get("id")
+
+            # Get linked inputs
+            linked_inputs = {
+                link["sink_name"]
+                for link in agent.get("links", [])
+                if link.get("sink_id") == node_id
+            }
+
+            for req_input in required_inputs:
+                if (
+                    req_input not in input_defaults
+                    and req_input not in linked_inputs
+                    and req_input != "credentials"
+                ):
+                    block_name = block.get("name", "Unknown Block")
+                    self.add_error(
+                        f"Node '{node_id}' ({block_name}) is missing required input '{req_input}'."
+                    )
+                    valid = False
+
+        return valid
+
+    def validate_data_type_compatibility(
+        self, agent: dict[str, Any], blocks_info: list[dict[str, Any]]
+    ) -> bool:
+        """Validate linked data types are compatible."""
+        valid = True
+        block_map = {b.get("id"): b for b in blocks_info}
+        node_lookup = {n.get("id"): n for n in agent.get("nodes", [])}
+
+        def get_type(schema: dict, name: str) -> str | None:
+            if "_#_" in name:
+                parent, child = name.split("_#_", 1)
+                parent_schema = schema.get(parent, {})
+                if "properties" in parent_schema:
+                    return parent_schema["properties"].get(child, {}).get("type")
+                return None
+            return schema.get(name, {}).get("type")
+
+        def are_compatible(src: str, sink: str) -> bool:
+            if {src, sink} <= {"integer", "number"}:
+                return True
+            return src == sink
+
+        for link in agent.get("links", []):
+            source_node = node_lookup.get(link.get("source_id"))
+            sink_node = node_lookup.get(link.get("sink_id"))
+
+            if not source_node or not sink_node:
+                continue
+
+            source_block = block_map.get(source_node.get("block_id"))
+            sink_block = block_map.get(sink_node.get("block_id"))
+
+            if not source_block or not sink_block:
+                continue
+
+            source_outputs = source_block.get("outputSchema", {}).get("properties", {})
+            sink_inputs = sink_block.get("inputSchema", {}).get("properties", {})
+
+            source_type = get_type(source_outputs, link.get("source_name", ""))
+            sink_type = get_type(sink_inputs, link.get("sink_name", ""))
+
+            if source_type and sink_type and not are_compatible(source_type, sink_type):
+                self.add_error(
+                    f"Type mismatch: {source_block.get('name')} output '{link['source_name']}' "
+                    f"({source_type}) -> {sink_block.get('name')} input '{link['sink_name']}' ({sink_type})."
+                )
+                valid = False
+
+        return valid
+
+    def validate_nested_sink_links(
+        self, agent: dict[str, Any], blocks_info: list[dict[str, Any]]
+    ) -> bool:
+        """Validate nested sink links (with _#_ notation)."""
+        valid = True
+        block_map = {b.get("id"): b for b in blocks_info}
+        node_lookup = {n.get("id"): n for n in agent.get("nodes", [])}
+
+        for link in agent.get("links", []):
+            sink_name = link.get("sink_name", "")
+
+            if "_#_" in sink_name:
+                parent, child = sink_name.split("_#_", 1)
+
+                sink_node = node_lookup.get(link.get("sink_id"))
+                if not sink_node:
+                    continue
+
+                block = block_map.get(sink_node.get("block_id"))
+                if not block:
+                    continue
+
+                input_props = block.get("inputSchema", {}).get("properties", {})
+                parent_schema = input_props.get(parent)
+
+                if not parent_schema:
+                    self.add_error(
+                        f"Invalid nested link '{sink_name}': parent '{parent}' not found."
+                    )
+                    valid = False
+                    continue
+
+                if not parent_schema.get("additionalProperties"):
+                    if not (
+                        isinstance(parent_schema, dict)
+                        and "properties" in parent_schema
+                        and child in parent_schema.get("properties", {})
+                    ):
+                        self.add_error(
+                            f"Invalid nested link '{sink_name}': child '{child}' not found in '{parent}'."
+                        )
+                        valid = False
+
+        return valid
+
+    def validate_prompt_spaces(self, agent: dict[str, Any]) -> bool:
+        """Validate prompts don't have spaces in template variables."""
+        valid = True
+
+        for node in agent.get("nodes", []):
+            input_default = node.get("input_default", {})
+            prompt = input_default.get("prompt", "")
+
+            if not isinstance(prompt, str):
+                continue
+
+            # Find {{...}} with spaces
+            matches = re.finditer(r"\{\{([^}]+)\}\}", prompt)
+            for match in matches:
+                content = match.group(1)
+                if " " in content:
+                    self.add_error(
+                        f"Node '{node.get('id')}' has spaces in template variable: "
+                        f"'{{{{{content}}}}}' should be '{{{{{content.replace(' ', '_')}}}}}'."
+                    )
+                    valid = False
+
+        return valid
+
+    def validate(
+        self, agent: dict[str, Any], blocks_info: list[dict[str, Any]] | None = None
+    ) -> tuple[bool, str | None]:
+        """Run all validations.
+
+        Returns:
+            Tuple of (is_valid, error_message)
+        """
+        self.errors = []
+
+        if blocks_info is None:
+            blocks_info = get_blocks_info()
+
+        checks = [
+            self.validate_block_existence(agent, blocks_info),
+            self.validate_link_node_references(agent),
+            self.validate_required_inputs(agent, blocks_info),
+            self.validate_data_type_compatibility(agent, blocks_info),
+            self.validate_nested_sink_links(agent, blocks_info),
+            self.validate_prompt_spaces(agent),
+        ]
+
+        all_passed = all(checks)
+
+        if all_passed:
+            logger.info("Agent validation successful")
+            return True, None
+
+        error_message = "Agent validation failed:\n"
+        for i, error in enumerate(self.errors, 1):
+            error_message += f"{i}. {error}\n"
+
+        logger.warning(f"Agent validation failed with {len(self.errors)} errors")
+        return False, error_message
+
+
+def validate_agent(
+    agent: dict[str, Any], blocks_info: list[dict[str, Any]] | None = None
+) -> tuple[bool, str | None]:
+    """Convenience function to validate an agent.
+
+    Returns:
+        Tuple of (is_valid, error_message)
+    """
+    validator = AgentValidator()
+    return validator.validate(agent, blocks_info)

autogpt_platform/backend/backend/api/features/chat/tools/agent_output.py

@@ -0,0 +1,455 @@
+"""Tool for retrieving agent execution outputs from user's library."""
+
+import logging
+import re
+from datetime import datetime, timedelta, timezone
+from typing import Any
+
+from pydantic import BaseModel, field_validator
+
+from backend.api.features.chat.model import ChatSession
+from backend.api.features.library import db as library_db
+from backend.api.features.library.model import LibraryAgent
+from backend.data import execution as execution_db
+from backend.data.execution import ExecutionStatus, GraphExecution, GraphExecutionMeta
+
+from .base import BaseTool
+from .models import (
+    AgentOutputResponse,
+    ErrorResponse,
+    ExecutionOutputInfo,
+    NoResultsResponse,
+    ToolResponseBase,
+)
+from .utils import fetch_graph_from_store_slug
+
+logger = logging.getLogger(__name__)
+
+
+class AgentOutputInput(BaseModel):
+    """Input parameters for the agent_output tool."""
+
+    agent_name: str = ""
+    library_agent_id: str = ""
+    store_slug: str = ""
+    execution_id: str = ""
+    run_time: str = "latest"
+
+    @field_validator(
+        "agent_name",
+        "library_agent_id",
+        "store_slug",
+        "execution_id",
+        "run_time",
+        mode="before",
+    )
+    @classmethod
+    def strip_strings(cls, v: Any) -> Any:
+        """Strip whitespace from string fields."""
+        return v.strip() if isinstance(v, str) else v
+
+
+def parse_time_expression(
+    time_expr: str | None,
+) -> tuple[datetime | None, datetime | None]:
+    """
+    Parse time expression into datetime range (start, end).
+
+    Supports:
+    - "latest" or None -> returns (None, None) to get most recent
+    - "yesterday" -> 24h window for yesterday
+    - "today" -> Today from midnight
+    - "last week" / "last 7 days" -> 7 day window
+    - "last month" / "last 30 days" -> 30 day window
+    - ISO date "YYYY-MM-DD" -> 24h window for that date
+    """
+    if not time_expr or time_expr.lower() == "latest":
+        return None, None
+
+    now = datetime.now(timezone.utc)
+    expr = time_expr.lower().strip()
+
+    # Relative expressions
+    if expr == "yesterday":
+        end = now.replace(hour=0, minute=0, second=0, microsecond=0)
+        start = end - timedelta(days=1)
+        return start, end
+
+    if expr in ("last week", "last 7 days"):
+        return now - timedelta(days=7), now
+
+    if expr in ("last month", "last 30 days"):
+        return now - timedelta(days=30), now
+
+    if expr == "today":
+        start = now.replace(hour=0, minute=0, second=0, microsecond=0)
+        return start, now
+
+    # Try ISO date format (YYYY-MM-DD)
+    date_match = re.match(r"^(\d{4})-(\d{2})-(\d{2})$", expr)
+    if date_match:
+        year, month, day = map(int, date_match.groups())
+        start = datetime(year, month, day, 0, 0, 0, tzinfo=timezone.utc)
+        end = start + timedelta(days=1)
+        return start, end
+
+    # Try ISO datetime
+    try:
+        parsed = datetime.fromisoformat(expr.replace("Z", "+00:00"))
+        if parsed.tzinfo is None:
+            parsed = parsed.replace(tzinfo=timezone.utc)
+        # Return +/- 1 hour window around the specified time
+        return parsed - timedelta(hours=1), parsed + timedelta(hours=1)
+    except ValueError:
+        pass
+
+    # Fallback: treat as "latest"
+    return None, None
+
+
+class AgentOutputTool(BaseTool):
+    """Tool for retrieving execution outputs from user's library agents."""
+
+    @property
+    def name(self) -> str:
+        return "agent_output"
+
+    @property
+    def description(self) -> str:
+        return """Retrieve execution outputs from agents in the user's library.
+
+        Identify the agent using one of:
+        - agent_name: Fuzzy search in user's library
+        - library_agent_id: Exact library agent ID
+        - store_slug: Marketplace format 'username/agent-name'
+
+        Select which run to retrieve using:
+        - execution_id: Specific execution ID
+        - run_time: 'latest' (default), 'yesterday', 'last week', or ISO date 'YYYY-MM-DD'
+        """
+
+    @property
+    def parameters(self) -> dict[str, Any]:
+        return {
+            "type": "object",
+            "properties": {
+                "agent_name": {
+                    "type": "string",
+                    "description": "Agent name to search for in user's library (fuzzy match)",
+                },
+                "library_agent_id": {
+                    "type": "string",
+                    "description": "Exact library agent ID",
+                },
+                "store_slug": {
+                    "type": "string",
+                    "description": "Marketplace identifier: 'username/agent-slug'",
+                },
+                "execution_id": {
+                    "type": "string",
+                    "description": "Specific execution ID to retrieve",
+                },
+                "run_time": {
+                    "type": "string",
+                    "description": (
+                        "Time filter: 'latest', 'yesterday', 'last week', or 'YYYY-MM-DD'"
+                    ),
+                },
+            },
+            "required": [],
+        }
+
+    @property
+    def requires_auth(self) -> bool:
+        return True
+
+    async def _resolve_agent(
+        self,
+        user_id: str,
+        agent_name: str | None,
+        library_agent_id: str | None,
+        store_slug: str | None,
+    ) -> tuple[LibraryAgent | None, str | None]:
+        """
+        Resolve agent from provided identifiers.
+        Returns (library_agent, error_message).
+        """
+        # Priority 1: Exact library agent ID
+        if library_agent_id:
+            try:
+                agent = await library_db.get_library_agent(library_agent_id, user_id)
+                return agent, None
+            except Exception as e:
+                logger.warning(f"Failed to get library agent by ID: {e}")
+                return None, f"Library agent '{library_agent_id}' not found"
+
+        # Priority 2: Store slug (username/agent-name)
+        if store_slug and "/" in store_slug:
+            username, agent_slug = store_slug.split("/", 1)
+            graph, _ = await fetch_graph_from_store_slug(username, agent_slug)
+            if not graph:
+                return None, f"Agent '{store_slug}' not found in marketplace"
+
+            # Find in user's library by graph_id
+            agent = await library_db.get_library_agent_by_graph_id(user_id, graph.id)
+            if not agent:
+                return (
+                    None,
+                    f"Agent '{store_slug}' is not in your library. "
+                    "Add it first to see outputs.",
+                )
+            return agent, None
+
+        # Priority 3: Fuzzy name search in library
+        if agent_name:
+            try:
+                response = await library_db.list_library_agents(
+                    user_id=user_id,
+                    search_term=agent_name,
+                    page_size=5,
+                )
+                if not response.agents:
+                    return (
+                        None,
+                        f"No agents matching '{agent_name}' found in your library",
+                    )
+
+                # Return best match (first result from search)
+                return response.agents[0], None
+            except Exception as e:
+                logger.error(f"Error searching library agents: {e}")
+                return None, f"Error searching for agent: {e}"
+
+        return (
+            None,
+            "Please specify an agent name, library_agent_id, or store_slug",
+        )
+
+    async def _get_execution(
+        self,
+        user_id: str,
+        graph_id: str,
+        execution_id: str | None,
+        time_start: datetime | None,
+        time_end: datetime | None,
+    ) -> tuple[GraphExecution | None, list[GraphExecutionMeta], str | None]:
+        """
+        Fetch execution(s) based on filters.
+        Returns (single_execution, available_executions_meta, error_message).
+        """
+        # If specific execution_id provided, fetch it directly
+        if execution_id:
+            execution = await execution_db.get_graph_execution(
+                user_id=user_id,
+                execution_id=execution_id,
+                include_node_executions=False,
+            )
+            if not execution:
+                return None, [], f"Execution '{execution_id}' not found"
+            return execution, [], None
+
+        # Get completed executions with time filters
+        executions = await execution_db.get_graph_executions(
+            graph_id=graph_id,
+            user_id=user_id,
+            statuses=[ExecutionStatus.COMPLETED],
+            created_time_gte=time_start,
+            created_time_lte=time_end,
+            limit=10,
+        )
+
+        if not executions:
+            return None, [], None  # No error, just no executions
+
+        # If only one execution, fetch full details
+        if len(executions) == 1:
+            full_execution = await execution_db.get_graph_execution(
+                user_id=user_id,
+                execution_id=executions[0].id,
+                include_node_executions=False,
+            )
+            return full_execution, [], None
+
+        # Multiple executions - return latest with full details, plus list of available
+        full_execution = await execution_db.get_graph_execution(
+            user_id=user_id,
+            execution_id=executions[0].id,
+            include_node_executions=False,
+        )
+        return full_execution, executions, None
+
+    def _build_response(
+        self,
+        agent: LibraryAgent,
+        execution: GraphExecution | None,
+        available_executions: list[GraphExecutionMeta],
+        session_id: str | None,
+    ) -> AgentOutputResponse:
+        """Build the response based on execution data."""
+        library_agent_link = f"/library/agents/{agent.id}"
+
+        if not execution:
+            return AgentOutputResponse(
+                message=f"No completed executions found for agent '{agent.name}'",
+                session_id=session_id,
+                agent_name=agent.name,
+                agent_id=agent.graph_id,
+                library_agent_id=agent.id,
+                library_agent_link=library_agent_link,
+                total_executions=0,
+            )
+
+        execution_info = ExecutionOutputInfo(
+            execution_id=execution.id,
+            status=execution.status.value,
+            started_at=execution.started_at,
+            ended_at=execution.ended_at,
+            outputs=dict(execution.outputs),
+            inputs_summary=execution.inputs if execution.inputs else None,
+        )
+
+        available_list = None
+        if len(available_executions) > 1:
+            available_list = [
+                {
+                    "id": e.id,
+                    "status": e.status.value,
+                    "started_at": e.started_at.isoformat() if e.started_at else None,
+                }
+                for e in available_executions[:5]
+            ]
+
+        message = f"Found execution outputs for agent '{agent.name}'"
+        if len(available_executions) > 1:
+            message += (
+                f". Showing latest of {len(available_executions)} matching executions."
+            )
+
+        return AgentOutputResponse(
+            message=message,
+            session_id=session_id,
+            agent_name=agent.name,
+            agent_id=agent.graph_id,
+            library_agent_id=agent.id,
+            library_agent_link=library_agent_link,
+            execution=execution_info,
+            available_executions=available_list,
+            total_executions=len(available_executions) if available_executions else 1,
+        )
+
+    async def _execute(
+        self,
+        user_id: str | None,
+        session: ChatSession,
+        **kwargs,
+    ) -> ToolResponseBase:
+        """Execute the agent_output tool."""
+        session_id = session.session_id
+
+        # Parse and validate input
+        try:
+            input_data = AgentOutputInput(**kwargs)
+        except Exception as e:
+            logger.error(f"Invalid input: {e}")
+            return ErrorResponse(
+                message="Invalid input parameters",
+                error=str(e),
+                session_id=session_id,
+            )
+
+        # Ensure user_id is present (should be guaranteed by requires_auth)
+        if not user_id:
+            return ErrorResponse(
+                message="User authentication required",
+                session_id=session_id,
+            )
+
+        # Check if at least one identifier is provided
+        if not any(
+            [
+                input_data.agent_name,
+                input_data.library_agent_id,
+                input_data.store_slug,
+                input_data.execution_id,
+            ]
+        ):
+            return ErrorResponse(
+                message=(
+                    "Please specify at least one of: agent_name, "
+                    "library_agent_id, store_slug, or execution_id"
+                ),
+                session_id=session_id,
+            )
+
+        # If only execution_id provided, we need to find the agent differently
+        if (
+            input_data.execution_id
+            and not input_data.agent_name
+            and not input_data.library_agent_id
+            and not input_data.store_slug
+        ):
+            # Fetch execution directly to get graph_id
+            execution = await execution_db.get_graph_execution(
+                user_id=user_id,
+                execution_id=input_data.execution_id,
+                include_node_executions=False,
+            )
+            if not execution:
+                return ErrorResponse(
+                    message=f"Execution '{input_data.execution_id}' not found",
+                    session_id=session_id,
+                )
+
+            # Find library agent by graph_id
+            agent = await library_db.get_library_agent_by_graph_id(
+                user_id, execution.graph_id
+            )
+            if not agent:
+                return NoResultsResponse(
+                    message=(
+                        f"Execution found but agent not in your library. "
+                        f"Graph ID: {execution.graph_id}"
+                    ),
+                    session_id=session_id,
+                    suggestions=["Add the agent to your library to see more details"],
+                )
+
+            return self._build_response(agent, execution, [], session_id)
+
+        # Resolve agent from identifiers
+        agent, error = await self._resolve_agent(
+            user_id=user_id,
+            agent_name=input_data.agent_name or None,
+            library_agent_id=input_data.library_agent_id or None,
+            store_slug=input_data.store_slug or None,
+        )
+
+        if error or not agent:
+            return NoResultsResponse(
+                message=error or "Agent not found",
+                session_id=session_id,
+                suggestions=[
+                    "Check the agent name or ID",
+                    "Make sure the agent is in your library",
+                ],
+            )
+
+        # Parse time expression
+        time_start, time_end = parse_time_expression(input_data.run_time)
+
+        # Fetch execution(s)
+        execution, available_executions, exec_error = await self._get_execution(
+            user_id=user_id,
+            graph_id=agent.graph_id,
+            execution_id=input_data.execution_id or None,
+            time_start=time_start,
+            time_end=time_end,
+        )
+
+        if exec_error:
+            return ErrorResponse(
+                message=exec_error,
+                session_id=session_id,
+            )
+
+        return self._build_response(agent, execution, available_executions, session_id)

autogpt_platform/backend/backend/api/features/chat/tools/create_agent.py

@@ -0,0 +1,279 @@
+"""CreateAgentTool - Creates agents from natural language descriptions."""
+
+import logging
+from typing import Any
+
+from backend.api.features.chat.model import ChatSession
+
+from .agent_generator import (
+    apply_all_fixes,
+    decompose_goal,
+    generate_agent,
+    get_blocks_info,
+    save_agent_to_library,
+    validate_agent,
+)
+from .base import BaseTool
+from .models import (
+    AgentPreviewResponse,
+    AgentSavedResponse,
+    ClarificationNeededResponse,
+    ClarifyingQuestion,
+    ErrorResponse,
+    ToolResponseBase,
+)
+
+logger = logging.getLogger(__name__)
+
+# Maximum retries for agent generation with validation feedback
+MAX_GENERATION_RETRIES = 2
+
+
+class CreateAgentTool(BaseTool):
+    """Tool for creating agents from natural language descriptions."""
+
+    @property
+    def name(self) -> str:
+        return "create_agent"
+
+    @property
+    def description(self) -> str:
+        return (
+            "Create a new agent workflow from a natural language description. "
+            "First generates a preview, then saves to library if save=true."
+        )
+
+    @property
+    def requires_auth(self) -> bool:
+        return True
+
+    @property
+    def parameters(self) -> dict[str, Any]:
+        return {
+            "type": "object",
+            "properties": {
+                "description": {
+                    "type": "string",
+                    "description": (
+                        "Natural language description of what the agent should do. "
+                        "Be specific about inputs, outputs, and the workflow steps."
+                    ),
+                },
+                "context": {
+                    "type": "string",
+                    "description": (
+                        "Additional context or answers to previous clarifying questions. "
+                        "Include any preferences or constraints mentioned by the user."
+                    ),
+                },
+                "save": {
+                    "type": "boolean",
+                    "description": (
+                        "Whether to save the agent to the user's library. "
+                        "Default is true. Set to false for preview only."
+                    ),
+                    "default": True,
+                },
+            },
+            "required": ["description"],
+        }
+
+    async def _execute(
+        self,
+        user_id: str | None,
+        session: ChatSession,
+        **kwargs,
+    ) -> ToolResponseBase:
+        """Execute the create_agent tool.
+
+        Flow:
+        1. Decompose the description into steps (may return clarifying questions)
+        2. Generate agent JSON from the steps
+        3. Apply fixes to correct common LLM errors
+        4. Preview or save based on the save parameter
+        """
+        description = kwargs.get("description", "").strip()
+        context = kwargs.get("context", "")
+        save = kwargs.get("save", True)
+        session_id = session.session_id if session else None
+
+        if not description:
+            return ErrorResponse(
+                message="Please provide a description of what the agent should do.",
+                error="Missing description parameter",
+                session_id=session_id,
+            )
+
+        # Step 1: Decompose goal into steps
+        try:
+            decomposition_result = await decompose_goal(description, context)
+        except ValueError as e:
+            # Handle missing API key or configuration errors
+            return ErrorResponse(
+                message=f"Agent generation is not configured: {str(e)}",
+                error="configuration_error",
+                session_id=session_id,
+            )
+
+        if decomposition_result is None:
+            return ErrorResponse(
+                message="Failed to analyze the goal. Please try rephrasing.",
+                error="Decomposition failed",
+                session_id=session_id,
+            )
+
+        # Check if LLM returned clarifying questions
+        if decomposition_result.get("type") == "clarifying_questions":
+            questions = decomposition_result.get("questions", [])
+            return ClarificationNeededResponse(
+                message=(
+                    "I need some more information to create this agent. "
+                    "Please answer the following questions:"
+                ),
+                questions=[
+                    ClarifyingQuestion(
+                        question=q.get("question", ""),
+                        keyword=q.get("keyword", ""),
+                        example=q.get("example"),
+                    )
+                    for q in questions
+                ],
+                session_id=session_id,
+            )
+
+        # Check for unachievable/vague goals
+        if decomposition_result.get("type") == "unachievable_goal":
+            suggested = decomposition_result.get("suggested_goal", "")
+            reason = decomposition_result.get("reason", "")
+            return ErrorResponse(
+                message=(
+                    f"This goal cannot be accomplished with the available blocks. "
+                    f"{reason} "
+                    f"Suggestion: {suggested}"
+                ),
+                error="unachievable_goal",
+                details={"suggested_goal": suggested, "reason": reason},
+                session_id=session_id,
+            )
+
+        if decomposition_result.get("type") == "vague_goal":
+            suggested = decomposition_result.get("suggested_goal", "")
+            return ErrorResponse(
+                message=(
+                    f"The goal is too vague to create a specific workflow. "
+                    f"Suggestion: {suggested}"
+                ),
+                error="vague_goal",
+                details={"suggested_goal": suggested},
+                session_id=session_id,
+            )
+
+        # Step 2: Generate agent JSON with retry on validation failure
+        blocks_info = get_blocks_info()
+        agent_json = None
+        validation_errors = None
+
+        for attempt in range(MAX_GENERATION_RETRIES + 1):
+            # Generate agent (include validation errors from previous attempt)
+            if attempt == 0:
+                agent_json = await generate_agent(decomposition_result)
+            else:
+                # Retry with validation error feedback
+                logger.info(
+                    f"Retry {attempt}/{MAX_GENERATION_RETRIES} with validation feedback"
+                )
+                retry_instructions = {
+                    **decomposition_result,
+                    "previous_errors": validation_errors,
+                    "retry_instructions": (
+                        "The previous generation had validation errors. "
+                        "Please fix these issues in the new generation:\n"
+                        f"{validation_errors}"
+                    ),
+                }
+                agent_json = await generate_agent(retry_instructions)
+
+            if agent_json is None:
+                if attempt == MAX_GENERATION_RETRIES:
+                    return ErrorResponse(
+                        message="Failed to generate the agent. Please try again.",
+                        error="Generation failed",
+                        session_id=session_id,
+                    )
+                continue
+
+            # Step 3: Apply fixes to correct common errors
+            agent_json = apply_all_fixes(agent_json, blocks_info)
+
+            # Step 4: Validate the agent
+            is_valid, validation_errors = validate_agent(agent_json, blocks_info)
+
+            if is_valid:
+                logger.info(f"Agent generated successfully on attempt {attempt + 1}")
+                break
+
+            logger.warning(
+                f"Validation failed on attempt {attempt + 1}: {validation_errors}"
+            )
+
+            if attempt == MAX_GENERATION_RETRIES:
+                # Return error with validation details
+                return ErrorResponse(
+                    message=(
+                        f"Generated agent has validation errors after {MAX_GENERATION_RETRIES + 1} attempts. "
+                        f"Please try rephrasing your request or simplify the workflow."
+                    ),
+                    error="validation_failed",
+                    details={"validation_errors": validation_errors},
+                    session_id=session_id,
+                )
+
+        agent_name = agent_json.get("name", "Generated Agent")
+        agent_description = agent_json.get("description", "")
+        node_count = len(agent_json.get("nodes", []))
+        link_count = len(agent_json.get("links", []))
+
+        # Step 4: Preview or save
+        if not save:
+            return AgentPreviewResponse(
+                message=(
+                    f"I've generated an agent called '{agent_name}' with {node_count} blocks. "
+                    f"Review it and call create_agent with save=true to save it to your library."
+                ),
+                agent_json=agent_json,
+                agent_name=agent_name,
+                description=agent_description,
+                node_count=node_count,
+                link_count=link_count,
+                session_id=session_id,
+            )
+
+        # Save to library
+        if not user_id:
+            return ErrorResponse(
+                message="You must be logged in to save agents.",
+                error="auth_required",
+                session_id=session_id,
+            )
+
+        try:
+            created_graph, library_agent = await save_agent_to_library(
+                agent_json, user_id
+            )
+
+            return AgentSavedResponse(
+                message=f"Agent '{created_graph.name}' has been saved to your library!",
+                agent_id=created_graph.id,
+                agent_name=created_graph.name,
+                library_agent_id=library_agent.id,
+                library_agent_link=f"/library/{library_agent.id}",
+                agent_page_link=f"/build?flowID={created_graph.id}",
+                session_id=session_id,
+            )
+        except Exception as e:
+            return ErrorResponse(
+                message=f"Failed to save the agent: {str(e)}",
+                error="save_failed",
+                details={"exception": str(e)},
+                session_id=session_id,
+            )

autogpt_platform/backend/backend/api/features/chat/tools/edit_agent.py

@@ -0,0 +1,294 @@
+"""EditAgentTool - Edits existing agents using natural language."""
+
+import logging
+from typing import Any
+
+from backend.api.features.chat.model import ChatSession
+
+from .agent_generator import (
+    apply_agent_patch,
+    apply_all_fixes,
+    generate_agent_patch,
+    get_agent_as_json,
+    get_blocks_info,
+    save_agent_to_library,
+    validate_agent,
+)
+from .base import BaseTool
+from .models import (
+    AgentPreviewResponse,
+    AgentSavedResponse,
+    ClarificationNeededResponse,
+    ClarifyingQuestion,
+    ErrorResponse,
+    ToolResponseBase,
+)
+
+logger = logging.getLogger(__name__)
+
+# Maximum retries for patch generation with validation feedback
+MAX_GENERATION_RETRIES = 2
+
+
+class EditAgentTool(BaseTool):
+    """Tool for editing existing agents using natural language."""
+
+    @property
+    def name(self) -> str:
+        return "edit_agent"
+
+    @property
+    def description(self) -> str:
+        return (
+            "Edit an existing agent from the user's library using natural language. "
+            "Generates a patch to update the agent while preserving unchanged parts."
+        )
+
+    @property
+    def requires_auth(self) -> bool:
+        return True
+
+    @property
+    def parameters(self) -> dict[str, Any]:
+        return {
+            "type": "object",
+            "properties": {
+                "agent_id": {
+                    "type": "string",
+                    "description": (
+                        "The ID of the agent to edit. "
+                        "Can be a graph ID or library agent ID."
+                    ),
+                },
+                "changes": {
+                    "type": "string",
+                    "description": (
+                        "Natural language description of what changes to make. "
+                        "Be specific about what to add, remove, or modify."
+                    ),
+                },
+                "context": {
+                    "type": "string",
+                    "description": (
+                        "Additional context or answers to previous clarifying questions."
+                    ),
+                },
+                "save": {
+                    "type": "boolean",
+                    "description": (
+                        "Whether to save the changes. "
+                        "Default is true. Set to false for preview only."
+                    ),
+                    "default": True,
+                },
+            },
+            "required": ["agent_id", "changes"],
+        }
+
+    async def _execute(
+        self,
+        user_id: str | None,
+        session: ChatSession,
+        **kwargs,
+    ) -> ToolResponseBase:
+        """Execute the edit_agent tool.
+
+        Flow:
+        1. Fetch the current agent
+        2. Generate a patch based on the requested changes
+        3. Apply the patch to create an updated agent
+        4. Preview or save based on the save parameter
+        """
+        agent_id = kwargs.get("agent_id", "").strip()
+        changes = kwargs.get("changes", "").strip()
+        context = kwargs.get("context", "")
+        save = kwargs.get("save", True)
+        session_id = session.session_id if session else None
+
+        if not agent_id:
+            return ErrorResponse(
+                message="Please provide the agent ID to edit.",
+                error="Missing agent_id parameter",
+                session_id=session_id,
+            )
+
+        if not changes:
+            return ErrorResponse(
+                message="Please describe what changes you want to make.",
+                error="Missing changes parameter",
+                session_id=session_id,
+            )
+
+        # Step 1: Fetch current agent
+        current_agent = await get_agent_as_json(agent_id, user_id)
+
+        if current_agent is None:
+            return ErrorResponse(
+                message=f"Could not find agent with ID '{agent_id}' in your library.",
+                error="agent_not_found",
+                session_id=session_id,
+            )
+
+        # Build the update request with context
+        update_request = changes
+        if context:
+            update_request = f"{changes}\n\nAdditional context:\n{context}"
+
+        # Step 2: Generate patch with retry on validation failure
+        blocks_info = get_blocks_info()
+        updated_agent = None
+        validation_errors = None
+        intent = "Applied requested changes"
+
+        for attempt in range(MAX_GENERATION_RETRIES + 1):
+            # Generate patch (include validation errors from previous attempt)
+            try:
+                if attempt == 0:
+                    patch_result = await generate_agent_patch(
+                        update_request, current_agent
+                    )
+                else:
+                    # Retry with validation error feedback
+                    logger.info(
+                        f"Retry {attempt}/{MAX_GENERATION_RETRIES} with validation feedback"
+                    )
+                    retry_request = (
+                        f"{update_request}\n\n"
+                        f"IMPORTANT: The previous edit had validation errors. "
+                        f"Please fix these issues:\n{validation_errors}"
+                    )
+                    patch_result = await generate_agent_patch(
+                        retry_request, current_agent
+                    )
+            except ValueError as e:
+                # Handle missing API key or configuration errors
+                return ErrorResponse(
+                    message=f"Agent generation is not configured: {str(e)}",
+                    error="configuration_error",
+                    session_id=session_id,
+                )
+
+            if patch_result is None:
+                if attempt == MAX_GENERATION_RETRIES:
+                    return ErrorResponse(
+                        message="Failed to generate changes. Please try rephrasing.",
+                        error="Patch generation failed",
+                        session_id=session_id,
+                    )
+                continue
+
+            # Check if LLM returned clarifying questions
+            if patch_result.get("type") == "clarifying_questions":
+                questions = patch_result.get("questions", [])
+                return ClarificationNeededResponse(
+                    message=(
+                        "I need some more information about the changes. "
+                        "Please answer the following questions:"
+                    ),
+                    questions=[
+                        ClarifyingQuestion(
+                            question=q.get("question", ""),
+                            keyword=q.get("keyword", ""),
+                            example=q.get("example"),
+                        )
+                        for q in questions
+                    ],
+                    session_id=session_id,
+                )
+
+            # Step 3: Apply patch and fixes
+            try:
+                updated_agent = apply_agent_patch(current_agent, patch_result)
+                updated_agent = apply_all_fixes(updated_agent, blocks_info)
+            except Exception as e:
+                if attempt == MAX_GENERATION_RETRIES:
+                    return ErrorResponse(
+                        message=f"Failed to apply changes: {str(e)}",
+                        error="patch_apply_failed",
+                        details={"exception": str(e)},
+                        session_id=session_id,
+                    )
+                validation_errors = str(e)
+                continue
+
+            # Step 4: Validate the updated agent
+            is_valid, validation_errors = validate_agent(updated_agent, blocks_info)
+
+            if is_valid:
+                logger.info(f"Agent edited successfully on attempt {attempt + 1}")
+                intent = patch_result.get("intent", "Applied requested changes")
+                break
+
+            logger.warning(
+                f"Validation failed on attempt {attempt + 1}: {validation_errors}"
+            )
+
+            if attempt == MAX_GENERATION_RETRIES:
+                # Return error with validation details
+                return ErrorResponse(
+                    message=(
+                        f"Updated agent has validation errors after "
+                        f"{MAX_GENERATION_RETRIES + 1} attempts. "
+                        f"Please try rephrasing your request or simplify the changes."
+                    ),
+                    error="validation_failed",
+                    details={"validation_errors": validation_errors},
+                    session_id=session_id,
+                )
+
+        # At this point, updated_agent is guaranteed to be set (we return on all failure paths)
+        assert updated_agent is not None
+
+        agent_name = updated_agent.get("name", "Updated Agent")
+        agent_description = updated_agent.get("description", "")
+        node_count = len(updated_agent.get("nodes", []))
+        link_count = len(updated_agent.get("links", []))
+
+        # Step 5: Preview or save
+        if not save:
+            return AgentPreviewResponse(
+                message=(
+                    f"I've updated the agent. Changes: {intent}. "
+                    f"The agent now has {node_count} blocks. "
+                    f"Review it and call edit_agent with save=true to save the changes."
+                ),
+                agent_json=updated_agent,
+                agent_name=agent_name,
+                description=agent_description,
+                node_count=node_count,
+                link_count=link_count,
+                session_id=session_id,
+            )
+
+        # Save to library (creates a new version)
+        if not user_id:
+            return ErrorResponse(
+                message="You must be logged in to save agents.",
+                error="auth_required",
+                session_id=session_id,
+            )
+
+        try:
+            created_graph, library_agent = await save_agent_to_library(
+                updated_agent, user_id, is_update=True
+            )
+
+            return AgentSavedResponse(
+                message=(
+                    f"Updated agent '{created_graph.name}' has been saved to your library! "
+                    f"Changes: {intent}"
+                ),
+                agent_id=created_graph.id,
+                agent_name=created_graph.name,
+                library_agent_id=library_agent.id,
+                library_agent_link=f"/library/{library_agent.id}",
+                agent_page_link=f"/build?flowID={created_graph.id}",
+                session_id=session_id,
+            )
+        except Exception as e:
+            return ErrorResponse(
+                message=f"Failed to save the updated agent: {str(e)}",
+                error="save_failed",
+                details={"exception": str(e)},
+                session_id=session_id,
+            )

autogpt_platform/backend/backend/api/features/chat/tools/find_block.py

@@ -0,0 +1,253 @@
+"""Tool for searching available blocks using hybrid search."""
+
+import logging
+from typing import Any
+
+from backend.api.features.chat.model import ChatSession
+from backend.blocks import load_all_blocks
+
+from .base import BaseTool
+from .models import (
+    BlockInfoSummary,
+    BlockListResponse,
+    ErrorResponse,
+    NoResultsResponse,
+    ToolResponseBase,
+)
+from .search_blocks import get_block_search_index
+
+logger = logging.getLogger(__name__)
+
+
+class FindBlockTool(BaseTool):
+    """Tool for searching available blocks."""
+
+    @property
+    def name(self) -> str:
+        return "find_block"
+
+    @property
+    def description(self) -> str:
+        return (
+            "Search for available blocks by name or description. "
+            "Blocks are reusable components that perform specific tasks like "
+            "sending emails, making API calls, processing text, etc. "
+            "Use this to find blocks that can be executed directly."
+        )
+
+    @property
+    def parameters(self) -> dict[str, Any]:
+        return {
+            "type": "object",
+            "properties": {
+                "query": {
+                    "type": "string",
+                    "description": (
+                        "Search query to find blocks by name or description. "
+                        "Use keywords like 'email', 'http', 'text', 'ai', etc."
+                    ),
+                },
+            },
+            "required": ["query"],
+        }
+
+    @property
+    def requires_auth(self) -> bool:
+        return True
+
+    def _matches_query(self, block, query: str) -> tuple[int, bool]:
+        """
+        Check if a block matches the query and return a priority score.
+
+        Returns (priority, matches) where:
+        - priority 0: exact name match
+        - priority 1: name contains query
+        - priority 2: description contains query
+        - priority 3: category contains query
+        """
+        query_lower = query.lower()
+        name_lower = block.name.lower()
+        desc_lower = block.description.lower()
+
+        # Exact name match
+        if query_lower == name_lower:
+            return 0, True
+
+        # Name contains query
+        if query_lower in name_lower:
+            return 1, True
+
+        # Description contains query
+        if query_lower in desc_lower:
+            return 2, True
+
+        # Category contains query
+        for category in block.categories:
+            if query_lower in category.name.lower():
+                return 3, True
+
+        return 4, False
+
+    async def _execute(
+        self,
+        user_id: str | None,
+        session: ChatSession,
+        **kwargs,
+    ) -> ToolResponseBase:
+        """Search for blocks matching the query.
+
+        Args:
+            user_id: User ID (required)
+            session: Chat session
+            query: Search query
+
+        Returns:
+            BlockListResponse: List of matching blocks
+            NoResultsResponse: No blocks found
+            ErrorResponse: Error message
+        """
+        query = kwargs.get("query", "").strip()
+        session_id = session.session_id
+
+        if not query:
+            return ErrorResponse(
+                message="Please provide a search query",
+                session_id=session_id,
+            )
+
+        try:
+            # Try hybrid search first
+            search_results = self._hybrid_search(query)
+
+            if search_results is not None:
+                # Hybrid search succeeded
+                if not search_results:
+                    return NoResultsResponse(
+                        message=f"No blocks found matching '{query}'",
+                        session_id=session_id,
+                        suggestions=[
+                            "Try more general terms",
+                            "Search by category: ai, text, social, search, etc.",
+                            "Check block names like 'SendEmail', 'HttpRequest', etc.",
+                        ],
+                    )
+
+                # Get full block info for each result
+                all_blocks = load_all_blocks()
+                blocks = []
+                for result in search_results:
+                    block_cls = all_blocks.get(result.block_id)
+                    if block_cls:
+                        block = block_cls()
+                        blocks.append(
+                            BlockInfoSummary(
+                                id=block.id,
+                                name=block.name,
+                                description=block.description,
+                                categories=[cat.name for cat in block.categories],
+                                input_schema=block.input_schema.jsonschema(),
+                                output_schema=block.output_schema.jsonschema(),
+                            )
+                        )
+
+                return BlockListResponse(
+                    message=(
+                        f"Found {len(blocks)} block{'s' if len(blocks) != 1 else ''} "
+                        f"matching '{query}'. Use run_block to execute a block with "
+                        "the required inputs."
+                    ),
+                    blocks=blocks,
+                    count=len(blocks),
+                    query=query,
+                    session_id=session_id,
+                )
+
+            # Fallback to simple search if hybrid search failed
+            return self._simple_search(query, session_id)
+
+        except Exception as e:
+            logger.error(f"Error searching blocks: {e}", exc_info=True)
+            return ErrorResponse(
+                message="Failed to search blocks. Please try again.",
+                error=str(e),
+                session_id=session_id,
+            )
+
+    def _hybrid_search(self, query: str) -> list | None:
+        """
+        Perform hybrid search using embeddings and BM25.
+
+        Returns:
+            List of BlockSearchResult if successful, None if index not available
+        """
+        try:
+            index = get_block_search_index()
+            if not index.load():
+                logger.info(
+                    "Block search index not available, falling back to simple search"
+                )
+                return None
+
+            results = index.search(query, top_k=10)
+            logger.info(f"Hybrid search found {len(results)} blocks for: {query}")
+            return results
+
+        except Exception as e:
+            logger.warning(f"Hybrid search failed, falling back to simple: {e}")
+            return None
+
+    def _simple_search(self, query: str, session_id: str) -> ToolResponseBase:
+        """Fallback simple search using substring matching."""
+        all_blocks = load_all_blocks()
+        logger.info(f"Simple searching {len(all_blocks)} blocks for: {query}")
+
+        # Find matching blocks with priority scores
+        matches: list[tuple[int, Any]] = []
+        for block_id, block_cls in all_blocks.items():
+            block = block_cls()
+            priority, is_match = self._matches_query(block, query)
+            if is_match:
+                matches.append((priority, block))
+
+        # Sort by priority (lower is better)
+        matches.sort(key=lambda x: x[0])
+
+        # Take top 10 results
+        top_matches = [block for _, block in matches[:10]]
+
+        if not top_matches:
+            return NoResultsResponse(
+                message=f"No blocks found matching '{query}'",
+                session_id=session_id,
+                suggestions=[
+                    "Try more general terms",
+                    "Search by category: ai, text, social, search, etc.",
+                    "Check block names like 'SendEmail', 'HttpRequest', etc.",
+                ],
+            )
+
+        # Build response
+        blocks = []
+        for block in top_matches:
+            blocks.append(
+                BlockInfoSummary(
+                    id=block.id,
+                    name=block.name,
+                    description=block.description,
+                    categories=[cat.name for cat in block.categories],
+                    input_schema=block.input_schema.jsonschema(),
+                    output_schema=block.output_schema.jsonschema(),
+                )
+            )
+
+        return BlockListResponse(
+            message=(
+                f"Found {len(blocks)} block{'s' if len(blocks) != 1 else ''} "
+                f"matching '{query}'. Use run_block to execute a block with "
+                "the required inputs."
+            ),
+            blocks=blocks,
+            count=len(blocks),
+            query=query,
+            session_id=session_id,
+        )

autogpt_platform/backend/backend/api/features/chat/tools/find_library_agent.py

@@ -0,0 +1,157 @@
+"""Tool for searching agents in the user's library."""
+
+import logging
+from typing import Any
+
+from backend.api.features.chat.model import ChatSession
+from backend.api.features.library import db as library_db
+from backend.util.exceptions import DatabaseError
+
+from .base import BaseTool
+from .models import (
+    AgentCarouselResponse,
+    AgentInfo,
+    ErrorResponse,
+    NoResultsResponse,
+    ToolResponseBase,
+)
+
+logger = logging.getLogger(__name__)
+
+
+class FindLibraryAgentTool(BaseTool):
+    """Tool for searching agents in the user's library."""
+
+    @property
+    def name(self) -> str:
+        return "find_library_agent"
+
+    @property
+    def description(self) -> str:
+        return (
+            "Search for agents in the user's library. Use this to find agents "
+            "the user has already added to their library, including agents they "
+            "created or added from the marketplace."
+        )
+
+    @property
+    def parameters(self) -> dict[str, Any]:
+        return {
+            "type": "object",
+            "properties": {
+                "query": {
+                    "type": "string",
+                    "description": (
+                        "Search query to find agents by name or description. "
+                        "Use keywords for best results."
+                    ),
+                },
+            },
+            "required": ["query"],
+        }
+
+    @property
+    def requires_auth(self) -> bool:
+        return True
+
+    async def _execute(
+        self,
+        user_id: str | None,
+        session: ChatSession,
+        **kwargs,
+    ) -> ToolResponseBase:
+        """Search for agents in the user's library.
+
+        Args:
+            user_id: User ID (required)
+            session: Chat session
+            query: Search query
+
+        Returns:
+            AgentCarouselResponse: List of agents found in the library
+            NoResultsResponse: No agents found
+            ErrorResponse: Error message
+        """
+        query = kwargs.get("query", "").strip()
+        session_id = session.session_id
+
+        if not query:
+            return ErrorResponse(
+                message="Please provide a search query",
+                session_id=session_id,
+            )
+
+        if not user_id:
+            return ErrorResponse(
+                message="User authentication required to search library",
+                session_id=session_id,
+            )
+
+        agents = []
+        try:
+            logger.info(f"Searching user library for: {query}")
+            library_results = await library_db.list_library_agents(
+                user_id=user_id,
+                search_term=query,
+                page_size=10,
+            )
+
+            logger.info(
+                f"Find library agents tool found {len(library_results.agents)} agents"
+            )
+
+            for agent in library_results.agents:
+                agents.append(
+                    AgentInfo(
+                        id=agent.id,
+                        name=agent.name,
+                        description=agent.description or "",
+                        source="library",
+                        in_library=True,
+                        creator=agent.creator_name,
+                        status=agent.status.value,
+                        can_access_graph=agent.can_access_graph,
+                        has_external_trigger=agent.has_external_trigger,
+                        new_output=agent.new_output,
+                        graph_id=agent.graph_id,
+                    ),
+                )
+
+        except DatabaseError as e:
+            logger.error(f"Error searching library agents: {e}", exc_info=True)
+            return ErrorResponse(
+                message="Failed to search library. Please try again.",
+                error=str(e),
+                session_id=session_id,
+            )
+
+        if not agents:
+            return NoResultsResponse(
+                message=(
+                    f"No agents found matching '{query}' in your library. "
+                    "Try different keywords or use find_agent to search the marketplace."
+                ),
+                session_id=session_id,
+                suggestions=[
+                    "Try more general terms",
+                    "Use find_agent to search the marketplace",
+                    "Check your library at /library",
+                ],
+            )
+
+        title = (
+            f"Found {len(agents)} agent{'s' if len(agents) != 1 else ''} "
+            f"in your library for '{query}'"
+        )
+
+        return AgentCarouselResponse(
+            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."
+            ),
+            title=title,
+            agents=agents,
+            count=len(agents),
+            session_id=session_id,
+        )

autogpt_platform/backend/backend/api/features/chat/tools/index_blocks.py

@@ -0,0 +1,483 @@
+#!/usr/bin/env python3
+"""
+Block Indexer for Hybrid Search
+
+Creates a hybrid search index from blocks:
+- OpenAI embeddings (text-embedding-3-small)
+- BM25 index for lexical search
+- Name index for title matching boost
+
+Supports incremental updates by tracking content hashes.
+
+Usage:
+    python -m backend.server.v2.chat.tools.index_blocks [--force]
+"""
+
+import argparse
+import base64
+import hashlib
+import json
+import logging
+import os
+import re
+import sys
+from collections import defaultdict
+from datetime import datetime, timezone
+from pathlib import Path
+from typing import Any
+
+import numpy as np
+
+logger = logging.getLogger(__name__)
+
+# Check for OpenAI availability
+try:
+    import openai  # noqa: F401
+
+    HAS_OPENAI = True
+except ImportError:
+    HAS_OPENAI = False
+    print("Warning: openai not installed. Run: pip install openai")
+
+# Default embedding model (OpenAI)
+DEFAULT_EMBEDDING_MODEL = "text-embedding-3-small"
+DEFAULT_EMBEDDING_DIM = 1536
+
+# Output path (relative to this file)
+INDEX_PATH = Path(__file__).parent / "blocks_index.json"
+
+# Stopwords for tokenization
+STOPWORDS = {
+    "the",
+    "a",
+    "an",
+    "is",
+    "are",
+    "was",
+    "were",
+    "be",
+    "been",
+    "being",
+    "have",
+    "has",
+    "had",
+    "do",
+    "does",
+    "did",
+    "will",
+    "would",
+    "could",
+    "should",
+    "may",
+    "might",
+    "must",
+    "shall",
+    "can",
+    "need",
+    "dare",
+    "ought",
+    "used",
+    "to",
+    "of",
+    "in",
+    "for",
+    "on",
+    "with",
+    "at",
+    "by",
+    "from",
+    "as",
+    "into",
+    "through",
+    "during",
+    "before",
+    "after",
+    "above",
+    "below",
+    "between",
+    "under",
+    "again",
+    "further",
+    "then",
+    "once",
+    "and",
+    "but",
+    "or",
+    "nor",
+    "so",
+    "yet",
+    "both",
+    "either",
+    "neither",
+    "not",
+    "only",
+    "own",
+    "same",
+    "than",
+    "too",
+    "very",
+    "just",
+    "also",
+    "now",
+    "here",
+    "there",
+    "when",
+    "where",
+    "why",
+    "how",
+    "all",
+    "each",
+    "every",
+    "few",
+    "more",
+    "most",
+    "other",
+    "some",
+    "such",
+    "no",
+    "any",
+    "this",
+    "that",
+    "these",
+    "those",
+    "it",
+    "its",
+    "block",  # Too common in block context
+}
+
+
+def tokenize(text: str) -> list[str]:
+    """Simple tokenizer for BM25."""
+    text = text.lower()
+    # Remove code blocks if any
+    text = re.sub(r"```[\s\S]*?```", "", text)
+    text = re.sub(r"`[^`]+`", "", text)
+    # Extract words (including camelCase split)
+    # First, split camelCase
+    text = re.sub(r"([a-z])([A-Z])", r"\1 \2", text)
+    # Extract words
+    words = re.findall(r"\b[a-z][a-z0-9_-]*\b", text)
+    # Remove very short words and stopwords
+    return [w for w in words if len(w) > 2 and w not in STOPWORDS]
+
+
+def build_searchable_text(block: Any) -> str:
+    """Build searchable text from block attributes."""
+    parts = []
+
+    # Block name (split camelCase for better tokenization)
+    name = block.name
+    # Split camelCase: GetCurrentTimeBlock -> Get Current Time Block
+    name_split = re.sub(r"([a-z])([A-Z])", r"\1 \2", name)
+    parts.append(name_split)
+
+    # Description
+    if block.description:
+        parts.append(block.description)
+
+    # Categories
+    for category in block.categories:
+        parts.append(category.name)
+
+    # Input schema field names and descriptions
+    try:
+        input_schema = block.input_schema.jsonschema()
+        if "properties" in input_schema:
+            for field_name, field_info in input_schema["properties"].items():
+                parts.append(field_name)
+                if "description" in field_info:
+                    parts.append(field_info["description"])
+    except Exception:
+        pass
+
+    # Output schema field names
+    try:
+        output_schema = block.output_schema.jsonschema()
+        if "properties" in output_schema:
+            for field_name in output_schema["properties"]:
+                parts.append(field_name)
+    except Exception:
+        pass
+
+    return " ".join(parts)
+
+
+def compute_content_hash(text: str) -> str:
+    """Compute MD5 hash of text for change detection."""
+    return hashlib.md5(text.encode()).hexdigest()
+
+
+def load_existing_index(index_path: Path) -> dict[str, Any] | None:
+    """Load existing index if present."""
+    if not index_path.exists():
+        return None
+
+    try:
+        with open(index_path, "r", encoding="utf-8") as f:
+            return json.load(f)
+    except Exception as e:
+        logger.warning(f"Failed to load existing index: {e}")
+        return None
+
+
+def create_embeddings(
+    texts: list[str],
+    model_name: str = DEFAULT_EMBEDDING_MODEL,
+    batch_size: int = 100,
+) -> np.ndarray:
+    """Create embeddings using OpenAI API."""
+    if not HAS_OPENAI:
+        raise RuntimeError("openai not installed. Run: pip install openai")
+
+    # Import here to satisfy type checker
+    from openai import OpenAI
+
+    # Check for API key
+    api_key = os.getenv("OPENAI_API_KEY")
+    if not api_key:
+        raise RuntimeError("OPENAI_API_KEY environment variable not set")
+
+    client = OpenAI(api_key=api_key)
+    embeddings = []
+
+    print(f"Creating embeddings for {len(texts)} texts using {model_name}...")
+
+    for i in range(0, len(texts), batch_size):
+        batch = texts[i : i + batch_size]
+        # Truncate texts to max token limit (8191 tokens for text-embedding-3-small)
+        # Roughly 4 chars per token, so ~32000 chars max
+        batch = [text[:32000] for text in batch]
+
+        response = client.embeddings.create(
+            model=model_name,
+            input=batch,
+        )
+
+        for embedding_data in response.data:
+            embeddings.append(embedding_data.embedding)
+
+        print(f"  Processed {min(i + batch_size, len(texts))}/{len(texts)} texts")
+
+    return np.array(embeddings, dtype=np.float32)
+
+
+def build_bm25_data(
+    blocks_data: list[dict[str, Any]],
+) -> dict[str, Any]:
+    """Build BM25 metadata from block data."""
+    # Tokenize all searchable texts
+    tokenized_docs = []
+    for block in blocks_data:
+        tokens = tokenize(block["searchable_text"])
+        tokenized_docs.append(tokens)
+
+    # Calculate document frequencies
+    doc_freq: dict[str, int] = {}
+    for tokens in tokenized_docs:
+        seen = set()
+        for token in tokens:
+            if token not in seen:
+                doc_freq[token] = doc_freq.get(token, 0) + 1
+                seen.add(token)
+
+    n_docs = len(tokenized_docs)
+    doc_lens = [len(d) for d in tokenized_docs]
+    avgdl = sum(doc_lens) / max(n_docs, 1)
+
+    return {
+        "n_docs": n_docs,
+        "avgdl": avgdl,
+        "df": doc_freq,
+        "doc_lens": doc_lens,
+    }
+
+
+def build_name_index(
+    blocks_data: list[dict[str, Any]],
+) -> dict[str, list[list[int | float]]]:
+    """Build inverted index for name search boost."""
+    index: dict[str, list[list[int | float]]] = defaultdict(list)
+
+    for idx, block in enumerate(blocks_data):
+        # Tokenize block name
+        name_tokens = tokenize(block["name"])
+        seen = set()
+
+        for i, token in enumerate(name_tokens):
+            if token in seen:
+                continue
+            seen.add(token)
+
+            # Score: first token gets higher weight
+            score = 1.5 if i == 0 else 1.0
+            index[token].append([idx, score])
+
+    return dict(index)
+
+
+def build_block_index(
+    force_rebuild: bool = False,
+    output_path: Path = INDEX_PATH,
+) -> dict[str, Any]:
+    """
+    Build the block search index.
+
+    Args:
+        force_rebuild: If True, rebuild all embeddings even if unchanged
+        output_path: Path to save the index
+
+    Returns:
+        The generated index dictionary
+    """
+    # Import here to avoid circular imports
+    from backend.blocks import load_all_blocks
+
+    print("Loading all blocks...")
+    all_blocks = load_all_blocks()
+    print(f"Found {len(all_blocks)} blocks")
+
+    # Load existing index for incremental updates
+    existing_index = None if force_rebuild else load_existing_index(output_path)
+    existing_blocks: dict[str, dict[str, Any]] = {}
+
+    if existing_index:
+        print(
+            f"Loaded existing index with {len(existing_index.get('blocks', []))} blocks"
+        )
+        for block in existing_index.get("blocks", []):
+            existing_blocks[block["id"]] = block
+
+    # Process each block
+    blocks_data: list[dict[str, Any]] = []
+    blocks_needing_embedding: list[tuple[int, str]] = []  # (index, searchable_text)
+
+    for block_id, block_cls in all_blocks.items():
+        try:
+            block = block_cls()
+
+            # Skip disabled blocks
+            if block.disabled:
+                continue
+
+            searchable_text = build_searchable_text(block)
+            content_hash = compute_content_hash(searchable_text)
+
+            block_data = {
+                "id": block.id,
+                "name": block.name,
+                "description": block.description,
+                "categories": [cat.name for cat in block.categories],
+                "searchable_text": searchable_text,
+                "content_hash": content_hash,
+                "emb": None,  # Will be filled later
+            }
+
+            # Check if we can reuse existing embedding
+            if (
+                block.id in existing_blocks
+                and existing_blocks[block.id].get("content_hash") == content_hash
+                and existing_blocks[block.id].get("emb")
+            ):
+                # Reuse existing embedding
+                block_data["emb"] = existing_blocks[block.id]["emb"]
+            else:
+                # Need new embedding
+                blocks_needing_embedding.append((len(blocks_data), searchable_text))
+
+            blocks_data.append(block_data)
+
+        except Exception as e:
+            logger.warning(f"Failed to process block {block_id}: {e}")
+            continue
+
+    print(f"Processed {len(blocks_data)} blocks")
+    print(f"Blocks needing new embeddings: {len(blocks_needing_embedding)}")
+
+    # Create embeddings for new/changed blocks
+    if blocks_needing_embedding and HAS_OPENAI:
+        texts_to_embed = [text for _, text in blocks_needing_embedding]
+        try:
+            embeddings = create_embeddings(texts_to_embed)
+
+            # Assign embeddings to blocks
+            for i, (block_idx, _) in enumerate(blocks_needing_embedding):
+                emb = embeddings[i].astype(np.float32)
+                # Encode as base64
+                blocks_data[block_idx]["emb"] = base64.b64encode(emb.tobytes()).decode(
+                    "ascii"
+                )
+        except Exception as e:
+            print(f"Warning: Failed to create embeddings: {e}")
+    elif blocks_needing_embedding:
+        print(
+            "Warning: Cannot create embeddings (openai not installed or OPENAI_API_KEY not set)"
+        )
+
+    # Build BM25 data
+    print("Building BM25 index...")
+    bm25_data = build_bm25_data(blocks_data)
+
+    # Build name index
+    print("Building name index...")
+    name_index = build_name_index(blocks_data)
+
+    # Build final index
+    index = {
+        "version": "1.0.0",
+        "embedding_model": DEFAULT_EMBEDDING_MODEL,
+        "embedding_dim": DEFAULT_EMBEDDING_DIM,
+        "generated_at": datetime.now(timezone.utc).isoformat(),
+        "blocks": blocks_data,
+        "bm25": bm25_data,
+        "name_index": name_index,
+    }
+
+    # Save index
+    print(f"Saving index to {output_path}...")
+    with open(output_path, "w", encoding="utf-8") as f:
+        json.dump(index, f, separators=(",", ":"))
+
+    size_kb = output_path.stat().st_size / 1024
+    print(f"Index saved ({size_kb:.1f} KB)")
+
+    # Print statistics
+    print("\nIndex Statistics:")
+    print(f"  Blocks indexed: {len(blocks_data)}")
+    print(f"  BM25 vocabulary size: {len(bm25_data['df'])}")
+    print(f"  Name index terms: {len(name_index)}")
+    print(f"  Embeddings: {'Yes' if any(b.get('emb') for b in blocks_data) else 'No'}")
+
+    return index
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Build hybrid search index for blocks")
+    parser.add_argument(
+        "--force",
+        action="store_true",
+        help="Force rebuild all embeddings even if unchanged",
+    )
+    parser.add_argument(
+        "--output",
+        type=Path,
+        default=INDEX_PATH,
+        help=f"Output index file path (default: {INDEX_PATH})",
+    )
+
+    args = parser.parse_args()
+
+    try:
+        build_block_index(
+            force_rebuild=args.force,
+            output_path=args.output,
+        )
+    except Exception as e:
+        print(f"Error building index: {e}")
+        import traceback
+
+        traceback.print_exc()
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    main()

autogpt_platform/backend/backend/api/features/chat/tools/models.py

@@ -1,5 +1,6 @@
 """Pydantic models for tool responses."""
 
+from datetime import datetime
 from enum import Enum
 from typing import Any
 
@@ -19,6 +20,15 @@ class ResponseType(str, Enum):
     ERROR = "error"
     NO_RESULTS = "no_results"
     SUCCESS = "success"
+    DOC_SEARCH_RESULTS = "doc_search_results"
+    AGENT_OUTPUT = "agent_output"
+    BLOCK_LIST = "block_list"
+    BLOCK_OUTPUT = "block_output"
+    UNDERSTANDING_UPDATED = "understanding_updated"
+    # Agent generation responses
+    AGENT_PREVIEW = "agent_preview"
+    AGENT_SAVED = "agent_saved"
+    CLARIFICATION_NEEDED = "clarification_needed"
 
 
 # Base response model
@@ -173,3 +183,128 @@ class ErrorResponse(ToolResponseBase):
     type: ResponseType = ResponseType.ERROR
     error: str | None = None
     details: dict[str, Any] | None = None
+
+
+# Documentation search models
+class DocSearchResult(BaseModel):
+    """A single documentation search result."""
+
+    title: str
+    path: str
+    section: str
+    snippet: str  # Short excerpt for UI display
+    content: str  # Full text content for LLM to read and understand
+    score: float
+    doc_url: str | None = None
+
+
+class DocSearchResultsResponse(ToolResponseBase):
+    """Response for search_docs tool."""
+
+    type: ResponseType = ResponseType.DOC_SEARCH_RESULTS
+    results: list[DocSearchResult]
+    count: int
+    query: str
+
+
+# Agent output models
+class ExecutionOutputInfo(BaseModel):
+    """Summary of a single execution's outputs."""
+
+    execution_id: str
+    status: str
+    started_at: datetime | None = None
+    ended_at: datetime | None = None
+    outputs: dict[str, list[Any]]
+    inputs_summary: dict[str, Any] | None = None
+
+
+class AgentOutputResponse(ToolResponseBase):
+    """Response for agent_output tool."""
+
+    type: ResponseType = ResponseType.AGENT_OUTPUT
+    agent_name: str
+    agent_id: str
+    library_agent_id: str | None = None
+    library_agent_link: str | None = None
+    execution: ExecutionOutputInfo | None = None
+    available_executions: list[dict[str, Any]] | None = None
+    total_executions: int = 0
+
+
+# Block models
+class BlockInfoSummary(BaseModel):
+    """Summary of a block for search results."""
+
+    id: str
+    name: str
+    description: str
+    categories: list[str]
+    input_schema: dict[str, Any]
+    output_schema: dict[str, Any]
+
+
+class BlockListResponse(ToolResponseBase):
+    """Response for find_block tool."""
+
+    type: ResponseType = ResponseType.BLOCK_LIST
+    blocks: list[BlockInfoSummary]
+    count: int
+    query: str
+
+
+class BlockOutputResponse(ToolResponseBase):
+    """Response for run_block tool."""
+
+    type: ResponseType = ResponseType.BLOCK_OUTPUT
+    block_id: str
+    block_name: str
+    outputs: dict[str, list[Any]]
+    success: bool = True
+
+
+# Business understanding models
+class UnderstandingUpdatedResponse(ToolResponseBase):
+    """Response for add_understanding tool."""
+
+    type: ResponseType = ResponseType.UNDERSTANDING_UPDATED
+    updated_fields: list[str] = Field(default_factory=list)
+    current_understanding: dict[str, Any] = Field(default_factory=dict)
+
+
+# Agent generation models
+class ClarifyingQuestion(BaseModel):
+    """A question that needs user clarification."""
+
+    question: str
+    keyword: str
+    example: str | None = None
+
+
+class AgentPreviewResponse(ToolResponseBase):
+    """Response for previewing a generated agent before saving."""
+
+    type: ResponseType = ResponseType.AGENT_PREVIEW
+    agent_json: dict[str, Any]
+    agent_name: str
+    description: str
+    node_count: int
+    link_count: int = 0
+
+
+class AgentSavedResponse(ToolResponseBase):
+    """Response when an agent is saved to the library."""
+
+    type: ResponseType = ResponseType.AGENT_SAVED
+    agent_id: str
+    agent_name: str
+    library_agent_id: str
+    library_agent_link: str
+    agent_page_link: str  # Link to the agent builder/editor page
+
+
+class ClarificationNeededResponse(ToolResponseBase):
+    """Response when the LLM needs more information from the user."""
+
+    type: ResponseType = ResponseType.CLARIFICATION_NEEDED
+    questions: list[ClarifyingQuestion] = Field(default_factory=list)

autogpt_platform/backend/backend/api/features/chat/tools/run_agent.py

@@ -7,6 +7,7 @@ from pydantic import BaseModel, Field, field_validator
 
 from backend.api.features.chat.config import ChatConfig
 from backend.api.features.chat.model import ChatSession
+from backend.api.features.library import db as library_db
 from backend.data.graph import GraphModel
 from backend.data.model import CredentialsMetaInput
 from backend.data.user import get_user_by_id
@@ -57,6 +58,7 @@ class RunAgentInput(BaseModel):
     """Input parameters for the run_agent tool."""
 
     username_agent_slug: str = ""
+    library_agent_id: str = ""
     inputs: dict[str, Any] = Field(default_factory=dict)
     use_defaults: bool = False
     schedule_name: str = ""
@@ -64,7 +66,12 @@ class RunAgentInput(BaseModel):
     timezone: str = "UTC"
 
     @field_validator(
-        "username_agent_slug", "schedule_name", "cron", "timezone", mode="before"
+        "username_agent_slug",
+        "library_agent_id",
+        "schedule_name",
+        "cron",
+        "timezone",
+        mode="before",
     )
     @classmethod
     def strip_strings(cls, v: Any) -> Any:
@@ -90,7 +97,7 @@ class RunAgentTool(BaseTool):
 
     @property
     def description(self) -> str:
-        return """Run or schedule an agent from the marketplace.
+        return """Run or schedule an agent from the marketplace or user's library.
 
         The tool automatically handles the setup flow:
         - Returns missing inputs if required fields are not provided
@@ -98,6 +105,10 @@ class RunAgentTool(BaseTool):
         - Executes immediately if all requirements are met
         - Schedules execution if cron expression is provided
 
+        Identify the agent using either:
+        - username_agent_slug: Marketplace format 'username/agent-name'
+        - library_agent_id: ID of an agent in the user's library
+
         For scheduled execution, provide: schedule_name, cron, and optionally timezone."""
 
     @property
@@ -109,6 +120,10 @@ class RunAgentTool(BaseTool):
                     "type": "string",
                     "description": "Agent identifier in format 'username/agent-name'",
                 },
+                "library_agent_id": {
+                    "type": "string",
+                    "description": "Library agent ID from user's library",
+                },
                 "inputs": {
                     "type": "object",
                     "description": "Input values for the agent",
@@ -131,7 +146,7 @@ class RunAgentTool(BaseTool):
                     "description": "IANA timezone for schedule (default: UTC)",
                 },
             },
-            "required": ["username_agent_slug"],
+            "required": [],
         }
 
     @property
@@ -149,10 +164,16 @@ class RunAgentTool(BaseTool):
         params = RunAgentInput(**kwargs)
         session_id = session.session_id
 
-        # Validate agent slug format
-        if not params.username_agent_slug or "/" not in params.username_agent_slug:
+        # Validate at least one identifier is provided
+        has_slug = params.username_agent_slug and "/" in params.username_agent_slug
+        has_library_id = bool(params.library_agent_id)
+
+        if not has_slug and not has_library_id:
             return ErrorResponse(
-                message="Please provide an agent slug in format 'username/agent-name'",
+                message=(
+                    "Please provide either a username_agent_slug "
+                    "(format 'username/agent-name') or a library_agent_id"
+                ),
                 session_id=session_id,
             )
 
@@ -167,13 +188,41 @@ class RunAgentTool(BaseTool):
         is_schedule = bool(params.schedule_name or params.cron)
 
         try:
-            # Step 1: Fetch agent details (always happens first)
-            username, agent_name = params.username_agent_slug.split("/", 1)
-            graph, store_agent = await fetch_graph_from_store_slug(username, agent_name)
+            # Step 1: Fetch agent details
+            graph: GraphModel | None = None
+            library_agent = None
+
+            # Priority: library_agent_id if provided
+            if has_library_id:
+                library_agent = await library_db.get_library_agent(
+                    params.library_agent_id, user_id
+                )
+                if not library_agent:
+                    return ErrorResponse(
+                        message=f"Library agent '{params.library_agent_id}' not found",
+                        session_id=session_id,
+                    )
+                # Get the graph from the library agent
+                from backend.data.graph import get_graph
+
+                graph = await get_graph(
+                    library_agent.graph_id,
+                    library_agent.graph_version,
+                    user_id=user_id,
+                )
+            else:
+                # Fetch from marketplace slug
+                username, agent_name = params.username_agent_slug.split("/", 1)
+                graph, _ = await fetch_graph_from_store_slug(username, agent_name)
 
             if not graph:
+                identifier = (
+                    params.library_agent_id
+                    if has_library_id
+                    else params.username_agent_slug
+                )
                 return ErrorResponse(
-                    message=f"Agent '{params.username_agent_slug}' not found in marketplace",
+                    message=f"Agent '{identifier}' not found",
                     session_id=session_id,
                 )
 

autogpt_platform/backend/backend/api/features/chat/tools/run_block.py

@@ -0,0 +1,287 @@
+"""Tool for executing blocks directly."""
+
+import logging
+from collections import defaultdict
+from typing import Any
+
+from backend.api.features.chat.model import ChatSession
+from backend.data.block import get_block
+from backend.data.model import CredentialsMetaInput
+from backend.integrations.creds_manager import IntegrationCredentialsManager
+from backend.util.exceptions import BlockError
+
+from .base import BaseTool
+from .models import (
+    BlockOutputResponse,
+    ErrorResponse,
+    SetupInfo,
+    SetupRequirementsResponse,
+    ToolResponseBase,
+    UserReadiness,
+)
+
+logger = logging.getLogger(__name__)
+
+
+class RunBlockTool(BaseTool):
+    """Tool for executing a block and returning its outputs."""
+
+    @property
+    def name(self) -> str:
+        return "run_block"
+
+    @property
+    def description(self) -> str:
+        return (
+            "Execute a specific block with the provided input data. "
+            "Use find_block to discover available blocks and their input schemas. "
+            "The block will run and return its outputs once complete."
+        )
+
+    @property
+    def parameters(self) -> dict[str, Any]:
+        return {
+            "type": "object",
+            "properties": {
+                "block_id": {
+                    "type": "string",
+                    "description": "The UUID of the block to execute",
+                },
+                "input_data": {
+                    "type": "object",
+                    "description": (
+                        "Input values for the block. Must match the block's input schema. "
+                        "Check the block's input_schema from find_block for required fields."
+                    ),
+                },
+            },
+            "required": ["block_id", "input_data"],
+        }
+
+    @property
+    def requires_auth(self) -> bool:
+        return True
+
+    async def _check_block_credentials(
+        self,
+        user_id: str,
+        block: Any,
+    ) -> tuple[dict[str, CredentialsMetaInput], list[CredentialsMetaInput]]:
+        """
+        Check if user has required credentials for a block.
+
+        Returns:
+            tuple[matched_credentials, missing_credentials]
+        """
+        matched_credentials: dict[str, CredentialsMetaInput] = {}
+        missing_credentials: list[CredentialsMetaInput] = []
+
+        # Get credential field info from block's input schema
+        credentials_fields_info = block.input_schema.get_credentials_fields_info()
+
+        if not credentials_fields_info:
+            return matched_credentials, missing_credentials
+
+        # Get user's available credentials
+        creds_manager = IntegrationCredentialsManager()
+        available_creds = await creds_manager.store.get_all_creds(user_id)
+
+        for field_name, field_info in credentials_fields_info.items():
+            # field_info.provider is a frozenset of acceptable providers
+            # field_info.supported_types is a frozenset of acceptable types
+            matching_cred = next(
+                (
+                    cred
+                    for cred in available_creds
+                    if cred.provider in field_info.provider
+                    and cred.type in field_info.supported_types
+                ),
+                None,
+            )
+
+            if matching_cred:
+                matched_credentials[field_name] = CredentialsMetaInput(
+                    id=matching_cred.id,
+                    provider=matching_cred.provider,  # type: ignore
+                    type=matching_cred.type,
+                    title=matching_cred.title,
+                )
+            else:
+                # Create a placeholder for the missing credential
+                provider = next(iter(field_info.provider), "unknown")
+                cred_type = next(iter(field_info.supported_types), "api_key")
+                missing_credentials.append(
+                    CredentialsMetaInput(
+                        id=field_name,
+                        provider=provider,  # type: ignore
+                        type=cred_type,  # type: ignore
+                        title=field_name.replace("_", " ").title(),
+                    )
+                )
+
+        return matched_credentials, missing_credentials
+
+    async def _execute(
+        self,
+        user_id: str | None,
+        session: ChatSession,
+        **kwargs,
+    ) -> ToolResponseBase:
+        """Execute a block with the given input data.
+
+        Args:
+            user_id: User ID (required)
+            session: Chat session
+            block_id: Block UUID to execute
+            input_data: Input values for the block
+
+        Returns:
+            BlockOutputResponse: Block execution outputs
+            SetupRequirementsResponse: Missing credentials
+            ErrorResponse: Error message
+        """
+        block_id = kwargs.get("block_id", "").strip()
+        input_data = kwargs.get("input_data", {})
+        session_id = session.session_id
+
+        if not block_id:
+            return ErrorResponse(
+                message="Please provide a block_id",
+                session_id=session_id,
+            )
+
+        if not isinstance(input_data, dict):
+            return ErrorResponse(
+                message="input_data must be an object",
+                session_id=session_id,
+            )
+
+        if not user_id:
+            return ErrorResponse(
+                message="Authentication required",
+                session_id=session_id,
+            )
+
+        # Get the block
+        block = get_block(block_id)
+        if not block:
+            return ErrorResponse(
+                message=f"Block '{block_id}' not found",
+                session_id=session_id,
+            )
+
+        logger.info(f"Executing block {block.name} ({block_id}) for user {user_id}")
+
+        # Check credentials
+        creds_manager = IntegrationCredentialsManager()
+        matched_credentials, missing_credentials = await self._check_block_credentials(
+            user_id, block
+        )
+
+        if missing_credentials:
+            # Return setup requirements response with missing credentials
+            missing_creds_dict = {c.id: c.model_dump() for c in missing_credentials}
+
+            return SetupRequirementsResponse(
+                message=(
+                    f"Block '{block.name}' requires credentials that are not configured. "
+                    "Please set up the required credentials before running this block."
+                ),
+                session_id=session_id,
+                setup_info=SetupInfo(
+                    agent_id=block_id,
+                    agent_name=block.name,
+                    user_readiness=UserReadiness(
+                        has_all_credentials=False,
+                        missing_credentials=missing_creds_dict,
+                        ready_to_run=False,
+                    ),
+                    requirements={
+                        "credentials": [c.model_dump() for c in missing_credentials],
+                        "inputs": self._get_inputs_list(block),
+                        "execution_modes": ["immediate"],
+                    },
+                ),
+                graph_id=None,
+                graph_version=None,
+            )
+
+        try:
+            # Fetch actual credentials and prepare kwargs for block execution
+            exec_kwargs: dict[str, Any] = {"user_id": user_id}
+
+            for field_name, cred_meta in matched_credentials.items():
+                # Inject metadata into input_data (for validation)
+                if field_name not in input_data:
+                    input_data[field_name] = cred_meta.model_dump()
+
+                # Fetch actual credentials and pass as kwargs (for execution)
+                actual_credentials = await creds_manager.get(
+                    user_id, cred_meta.id, lock=False
+                )
+                if actual_credentials:
+                    exec_kwargs[field_name] = actual_credentials
+                else:
+                    return ErrorResponse(
+                        message=f"Failed to retrieve credentials for {field_name}",
+                        session_id=session_id,
+                    )
+
+            # Execute the block and collect outputs
+            outputs: dict[str, list[Any]] = defaultdict(list)
+            async for output_name, output_data in block.execute(
+                input_data,
+                **exec_kwargs,
+            ):
+                outputs[output_name].append(output_data)
+
+            return BlockOutputResponse(
+                message=f"Block '{block.name}' executed successfully",
+                block_id=block_id,
+                block_name=block.name,
+                outputs=dict(outputs),
+                success=True,
+                session_id=session_id,
+            )
+
+        except BlockError as e:
+            logger.warning(f"Block execution failed: {e}")
+            return ErrorResponse(
+                message=f"Block execution failed: {e}",
+                error=str(e),
+                session_id=session_id,
+            )
+        except Exception as e:
+            logger.error(f"Unexpected error executing block: {e}", exc_info=True)
+            return ErrorResponse(
+                message=f"Failed to execute block: {str(e)}",
+                error=str(e),
+                session_id=session_id,
+            )
+
+    def _get_inputs_list(self, block: Any) -> list[dict[str, Any]]:
+        """Extract non-credential inputs from block schema."""
+        inputs_list = []
+        schema = block.input_schema.jsonschema()
+        properties = schema.get("properties", {})
+        required_fields = set(schema.get("required", []))
+
+        # Get credential field names to exclude
+        credentials_fields = set(block.input_schema.get_credentials_fields().keys())
+
+        for field_name, field_schema in properties.items():
+            # Skip credential fields
+            if field_name in credentials_fields:
+                continue
+
+            inputs_list.append(
+                {
+                    "name": field_name,
+                    "title": field_schema.get("title", field_name),
+                    "type": field_schema.get("type", "string"),
+                    "description": field_schema.get("description", ""),
+                    "required": field_name in required_fields,
+                }
+            )
+
+        return inputs_list

autogpt_platform/backend/backend/api/features/chat/tools/search_blocks.py

@@ -0,0 +1,460 @@
+"""
+Block Hybrid Search
+
+Combines multiple ranking signals for block search:
+- Semantic search (OpenAI embeddings + cosine similarity)
+- Lexical search (BM25)
+- Name matching (boost for block name matches)
+- Category matching (boost for category matches)
+
+Based on the docs search implementation.
+"""
+
+import base64
+import json
+import logging
+import math
+import os
+import re
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Any, Optional
+
+import numpy as np
+
+logger = logging.getLogger(__name__)
+
+# OpenAI embedding model
+EMBEDDING_MODEL = "text-embedding-3-small"
+
+# Path to the JSON index file
+INDEX_PATH = Path(__file__).parent / "blocks_index.json"
+
+# Stopwords for tokenization (same as index_blocks.py)
+STOPWORDS = {
+    "the",
+    "a",
+    "an",
+    "is",
+    "are",
+    "was",
+    "were",
+    "be",
+    "been",
+    "being",
+    "have",
+    "has",
+    "had",
+    "do",
+    "does",
+    "did",
+    "will",
+    "would",
+    "could",
+    "should",
+    "may",
+    "might",
+    "must",
+    "shall",
+    "can",
+    "need",
+    "dare",
+    "ought",
+    "used",
+    "to",
+    "of",
+    "in",
+    "for",
+    "on",
+    "with",
+    "at",
+    "by",
+    "from",
+    "as",
+    "into",
+    "through",
+    "during",
+    "before",
+    "after",
+    "above",
+    "below",
+    "between",
+    "under",
+    "again",
+    "further",
+    "then",
+    "once",
+    "and",
+    "but",
+    "or",
+    "nor",
+    "so",
+    "yet",
+    "both",
+    "either",
+    "neither",
+    "not",
+    "only",
+    "own",
+    "same",
+    "than",
+    "too",
+    "very",
+    "just",
+    "also",
+    "now",
+    "here",
+    "there",
+    "when",
+    "where",
+    "why",
+    "how",
+    "all",
+    "each",
+    "every",
+    "few",
+    "more",
+    "most",
+    "other",
+    "some",
+    "such",
+    "no",
+    "any",
+    "this",
+    "that",
+    "these",
+    "those",
+    "it",
+    "its",
+    "block",
+}
+
+
+def tokenize(text: str) -> list[str]:
+    """Simple tokenizer for search."""
+    text = text.lower()
+    # Remove code blocks if any
+    text = re.sub(r"```[\s\S]*?```", "", text)
+    text = re.sub(r"`[^`]+`", "", text)
+    # Split camelCase
+    text = re.sub(r"([a-z])([A-Z])", r"\1 \2", text)
+    # Extract words
+    words = re.findall(r"\b[a-z][a-z0-9_-]*\b", text)
+    # Remove very short words and stopwords
+    return [w for w in words if len(w) > 2 and w not in STOPWORDS]
+
+
+@dataclass
+class SearchWeights:
+    """Configuration for hybrid search signal weights."""
+
+    semantic: float = 0.40  # Embedding similarity
+    bm25: float = 0.25  # Lexical matching
+    name_match: float = 0.25  # Block name matches
+    category_match: float = 0.10  # Category matches
+
+
+@dataclass
+class BlockSearchResult:
+    """A single block search result."""
+
+    block_id: str
+    name: str
+    description: str
+    categories: list[str]
+    score: float
+
+    # Individual signal scores (for debugging)
+    semantic_score: float = 0.0
+    bm25_score: float = 0.0
+    name_score: float = 0.0
+    category_score: float = 0.0
+
+
+class BlockSearchIndex:
+    """Hybrid search index for blocks combining BM25 + embeddings."""
+
+    def __init__(self, index_path: Path = INDEX_PATH):
+        self.blocks: list[dict[str, Any]] = []
+        self.bm25_data: dict[str, Any] = {}
+        self.name_index: dict[str, list[list[int | float]]] = {}
+        self.embeddings: Optional[np.ndarray] = None
+        self.normalized_embeddings: Optional[np.ndarray] = None
+        self._loaded = False
+        self._index_path = index_path
+        self._embedding_model: Any = None
+
+    def load(self) -> bool:
+        """Load the index from JSON file."""
+        if self._loaded:
+            return True
+
+        if not self._index_path.exists():
+            logger.warning(f"Block index not found at {self._index_path}")
+            return False
+
+        try:
+            with open(self._index_path, "r", encoding="utf-8") as f:
+                data = json.load(f)
+
+            self.blocks = data.get("blocks", [])
+            self.bm25_data = data.get("bm25", {})
+            self.name_index = data.get("name_index", {})
+
+            # Decode embeddings from base64
+            embeddings_list = []
+            for block in self.blocks:
+                if block.get("emb"):
+                    emb_bytes = base64.b64decode(block["emb"])
+                    emb = np.frombuffer(emb_bytes, dtype=np.float32)
+                    embeddings_list.append(emb)
+                else:
+                    # No embedding, use zeros
+                    dim = data.get("embedding_dim", 384)
+                    embeddings_list.append(np.zeros(dim, dtype=np.float32))
+
+            if embeddings_list:
+                self.embeddings = np.stack(embeddings_list)
+                # Precompute normalized embeddings for cosine similarity
+                norms = np.linalg.norm(self.embeddings, axis=1, keepdims=True)
+                self.normalized_embeddings = self.embeddings / (norms + 1e-10)
+
+            self._loaded = True
+            logger.info(f"Loaded block index with {len(self.blocks)} blocks")
+            return True
+
+        except Exception as e:
+            logger.error(f"Failed to load block index: {e}")
+            return False
+
+    def _get_openai_client(self) -> Any:
+        """Get OpenAI client for query embedding."""
+        if self._embedding_model is None:
+            try:
+                from openai import OpenAI
+
+                api_key = os.getenv("OPENAI_API_KEY")
+                if not api_key:
+                    logger.warning("OPENAI_API_KEY not set")
+                    return None
+                self._embedding_model = OpenAI(api_key=api_key)
+            except ImportError:
+                logger.warning("openai not installed")
+                return None
+        return self._embedding_model
+
+    def _embed_query(self, query: str) -> Optional[np.ndarray]:
+        """Embed the search query using OpenAI."""
+        client = self._get_openai_client()
+        if client is None:
+            return None
+
+        try:
+            response = client.embeddings.create(
+                model=EMBEDDING_MODEL,
+                input=query,
+            )
+            embedding = response.data[0].embedding
+            return np.array(embedding, dtype=np.float32)
+        except Exception as e:
+            logger.warning(f"Failed to embed query: {e}")
+            return None
+
+    def _compute_semantic_scores(self, query_embedding: np.ndarray) -> np.ndarray:
+        """Compute cosine similarity between query and all blocks."""
+        if self.normalized_embeddings is None:
+            return np.zeros(len(self.blocks))
+
+        # Normalize query embedding
+        query_norm = query_embedding / (np.linalg.norm(query_embedding) + 1e-10)
+
+        # Cosine similarity via dot product
+        similarities = self.normalized_embeddings @ query_norm
+
+        # Scale to [0, 1] (cosine ranges from -1 to 1)
+        return (similarities + 1) / 2
+
+    def _compute_bm25_scores(self, query_tokens: list[str]) -> np.ndarray:
+        """Compute BM25 scores for all blocks."""
+        scores = np.zeros(len(self.blocks))
+
+        if not self.bm25_data or not query_tokens:
+            return scores
+
+        # BM25 parameters
+        k1 = 1.5
+        b = 0.75
+        n_docs = self.bm25_data.get("n_docs", len(self.blocks))
+        avgdl = self.bm25_data.get("avgdl", 100)
+        df = self.bm25_data.get("df", {})
+        doc_lens = self.bm25_data.get("doc_lens", [100] * len(self.blocks))
+
+        for i, block in enumerate(self.blocks):
+            # Tokenize block's searchable text
+            block_tokens = tokenize(block.get("searchable_text", ""))
+            doc_len = doc_lens[i] if i < len(doc_lens) else len(block_tokens)
+
+            # Calculate BM25 score
+            score = 0.0
+            for token in query_tokens:
+                if token not in df:
+                    continue
+
+                # Term frequency in this document
+                tf = block_tokens.count(token)
+                if tf == 0:
+                    continue
+
+                # IDF
+                doc_freq = df.get(token, 0)
+                idf = math.log((n_docs - doc_freq + 0.5) / (doc_freq + 0.5) + 1)
+
+                # BM25 score component
+                numerator = tf * (k1 + 1)
+                denominator = tf + k1 * (1 - b + b * doc_len / avgdl)
+                score += idf * numerator / denominator
+
+            scores[i] = score
+
+        # Normalize to [0, 1]
+        max_score = scores.max()
+        if max_score > 0:
+            scores = scores / max_score
+
+        return scores
+
+    def _compute_name_scores(self, query_tokens: list[str]) -> np.ndarray:
+        """Compute name match scores using the name index."""
+        scores = np.zeros(len(self.blocks))
+
+        if not self.name_index or not query_tokens:
+            return scores
+
+        for token in query_tokens:
+            if token in self.name_index:
+                for block_idx, weight in self.name_index[token]:
+                    if block_idx < len(scores):
+                        scores[int(block_idx)] += weight
+
+        # Also check for partial matches in block names
+        for i, block in enumerate(self.blocks):
+            name_lower = block.get("name", "").lower()
+            for token in query_tokens:
+                if token in name_lower:
+                    scores[i] += 0.5
+
+        # Normalize to [0, 1]
+        max_score = scores.max()
+        if max_score > 0:
+            scores = scores / max_score
+
+        return scores
+
+    def _compute_category_scores(self, query_tokens: list[str]) -> np.ndarray:
+        """Compute category match scores."""
+        scores = np.zeros(len(self.blocks))
+
+        if not query_tokens:
+            return scores
+
+        for i, block in enumerate(self.blocks):
+            categories = block.get("categories", [])
+            category_text = " ".join(categories).lower()
+
+            for token in query_tokens:
+                if token in category_text:
+                    scores[i] += 1.0
+
+        # Normalize to [0, 1]
+        max_score = scores.max()
+        if max_score > 0:
+            scores = scores / max_score
+
+        return scores
+
+    def search(
+        self,
+        query: str,
+        top_k: int = 10,
+        weights: Optional[SearchWeights] = None,
+    ) -> list[BlockSearchResult]:
+        """
+        Perform hybrid search combining multiple signals.
+
+        Args:
+            query: Search query string
+            top_k: Number of results to return
+            weights: Optional custom weights for signals
+
+        Returns:
+            List of BlockSearchResult sorted by score
+        """
+        if not self._loaded and not self.load():
+            return []
+
+        if weights is None:
+            weights = SearchWeights()
+
+        # Tokenize query
+        query_tokens = tokenize(query)
+        if not query_tokens:
+            # Fallback: try raw query words
+            query_tokens = query.lower().split()
+
+        # Compute semantic scores
+        semantic_scores = np.zeros(len(self.blocks))
+        if self.normalized_embeddings is not None:
+            query_embedding = self._embed_query(query)
+            if query_embedding is not None:
+                semantic_scores = self._compute_semantic_scores(query_embedding)
+
+        # Compute other scores
+        bm25_scores = self._compute_bm25_scores(query_tokens)
+        name_scores = self._compute_name_scores(query_tokens)
+        category_scores = self._compute_category_scores(query_tokens)
+
+        # Combine scores using weights
+        combined_scores = (
+            weights.semantic * semantic_scores
+            + weights.bm25 * bm25_scores
+            + weights.name_match * name_scores
+            + weights.category_match * category_scores
+        )
+
+        # Get top-k indices
+        top_indices = np.argsort(combined_scores)[::-1][:top_k]
+
+        # Build results
+        results = []
+        for idx in top_indices:
+            if combined_scores[idx] <= 0:
+                continue
+
+            block = self.blocks[idx]
+            results.append(
+                BlockSearchResult(
+                    block_id=block["id"],
+                    name=block["name"],
+                    description=block["description"],
+                    categories=block.get("categories", []),
+                    score=float(combined_scores[idx]),
+                    semantic_score=float(semantic_scores[idx]),
+                    bm25_score=float(bm25_scores[idx]),
+                    name_score=float(name_scores[idx]),
+                    category_score=float(category_scores[idx]),
+                )
+            )
+
+        return results
+
+
+# Global index instance (lazy loaded)
+_block_search_index: Optional[BlockSearchIndex] = None
+
+
+def get_block_search_index() -> BlockSearchIndex:
+    """Get or create the block search index singleton."""
+    global _block_search_index
+    if _block_search_index is None:
+        _block_search_index = BlockSearchIndex(INDEX_PATH)
+    return _block_search_index

autogpt_platform/backend/backend/api/features/chat/tools/search_docs.py

@@ -0,0 +1,386 @@
+"""Tool for searching platform documentation."""
+
+import json
+import logging
+import math
+import re
+from pathlib import Path
+from typing import Any
+
+from backend.api.features.chat.model import ChatSession
+
+from .base import BaseTool
+from .models import (
+    DocSearchResult,
+    DocSearchResultsResponse,
+    ErrorResponse,
+    NoResultsResponse,
+    ToolResponseBase,
+)
+
+logger = logging.getLogger(__name__)
+
+# Documentation base URL
+DOCS_BASE_URL = "https://docs.agpt.co/platform"
+
+# Path to the JSON index file (relative to this file)
+INDEX_PATH = Path(__file__).parent / "docs_index.json"
+
+
+def tokenize(text: str) -> list[str]:
+    """Simple tokenizer for BM25."""
+    text = text.lower()
+    # Remove code blocks
+    text = re.sub(r"```[\s\S]*?```", "", text)
+    text = re.sub(r"`[^`]+`", "", text)
+    # Extract words
+    words = re.findall(r"\b[a-z][a-z0-9_-]*\b", text)
+    # Remove very short words and stopwords
+    stopwords = {
+        "the",
+        "a",
+        "an",
+        "is",
+        "are",
+        "was",
+        "were",
+        "be",
+        "been",
+        "being",
+        "have",
+        "has",
+        "had",
+        "do",
+        "does",
+        "did",
+        "will",
+        "would",
+        "could",
+        "should",
+        "may",
+        "might",
+        "must",
+        "shall",
+        "can",
+        "need",
+        "dare",
+        "ought",
+        "used",
+        "to",
+        "of",
+        "in",
+        "for",
+        "on",
+        "with",
+        "at",
+        "by",
+        "from",
+        "as",
+        "into",
+        "through",
+        "during",
+        "before",
+        "after",
+        "above",
+        "below",
+        "between",
+        "under",
+        "again",
+        "further",
+        "then",
+        "once",
+        "and",
+        "but",
+        "or",
+        "nor",
+        "so",
+        "yet",
+        "both",
+        "either",
+        "neither",
+        "not",
+        "only",
+        "own",
+        "same",
+        "than",
+        "too",
+        "very",
+        "just",
+        "also",
+        "now",
+        "here",
+        "there",
+        "when",
+        "where",
+        "why",
+        "how",
+        "all",
+        "each",
+        "every",
+        "both",
+        "few",
+        "more",
+        "most",
+        "other",
+        "some",
+        "such",
+        "no",
+        "any",
+        "this",
+        "that",
+        "these",
+        "those",
+        "it",
+        "its",
+    }
+    return [w for w in words if len(w) > 2 and w not in stopwords]
+
+
+class DocSearchIndex:
+    """Lightweight documentation search index using BM25."""
+
+    def __init__(self, index_path: Path):
+        self.chunks: list[dict] = []
+        self.bm25_data: dict = {}
+        self._loaded = False
+        self._index_path = index_path
+
+    def load(self) -> bool:
+        """Load the index from JSON file."""
+        if self._loaded:
+            return True
+
+        if not self._index_path.exists():
+            logger.warning(f"Documentation index not found at {self._index_path}")
+            return False
+
+        try:
+            with open(self._index_path, "r", encoding="utf-8") as f:
+                data = json.load(f)
+
+            self.chunks = data.get("chunks", [])
+            self.bm25_data = data.get("bm25", {})
+            self._loaded = True
+            logger.info(f"Loaded documentation index with {len(self.chunks)} chunks")
+            return True
+        except Exception as e:
+            logger.error(f"Failed to load documentation index: {e}")
+            return False
+
+    def search(self, query: str, top_k: int = 5) -> list[dict]:
+        """Search the index using BM25."""
+        if not self._loaded and not self.load():
+            return []
+
+        query_tokens = tokenize(query)
+        if not query_tokens:
+            return []
+
+        # BM25 parameters
+        k1 = 1.5
+        b = 0.75
+        n_docs = self.bm25_data.get("n_docs", len(self.chunks))
+        avgdl = self.bm25_data.get("avgdl", 100)
+        df = self.bm25_data.get("df", {})
+        doc_lens = self.bm25_data.get("doc_lens", [100] * len(self.chunks))
+
+        scores = []
+        for i, chunk in enumerate(self.chunks):
+            # Tokenize chunk text
+            chunk_tokens = tokenize(chunk.get("text", ""))
+            doc_len = doc_lens[i] if i < len(doc_lens) else len(chunk_tokens)
+
+            # Calculate BM25 score
+            score = 0.0
+            for token in query_tokens:
+                if token not in df:
+                    continue
+
+                # Term frequency in this document
+                tf = chunk_tokens.count(token)
+                if tf == 0:
+                    continue
+
+                # IDF
+                doc_freq = df.get(token, 0)
+                idf = math.log((n_docs - doc_freq + 0.5) / (doc_freq + 0.5) + 1)
+
+                # BM25 score component
+                numerator = tf * (k1 + 1)
+                denominator = tf + k1 * (1 - b + b * doc_len / avgdl)
+                score += idf * numerator / denominator
+
+            # Boost for title/heading matches
+            title = chunk.get("title", "").lower()
+            heading = chunk.get("heading", "").lower()
+            for token in query_tokens:
+                if token in title:
+                    score *= 1.5
+                if token in heading:
+                    score *= 1.2
+
+            scores.append((i, score))
+
+        # Sort by score and return top_k
+        scores.sort(key=lambda x: x[1], reverse=True)
+
+        results = []
+        seen_sections = set()
+        for idx, score in scores:
+            if score <= 0:
+                continue
+
+            chunk = self.chunks[idx]
+            section_key = (chunk.get("doc", ""), chunk.get("heading", ""))
+
+            # Deduplicate by section
+            if section_key in seen_sections:
+                continue
+            seen_sections.add(section_key)
+
+            results.append(
+                {
+                    "title": chunk.get("title", ""),
+                    "path": chunk.get("doc", ""),
+                    "heading": chunk.get("heading", ""),
+                    "text": chunk.get("text", ""),  # Full text for LLM comprehension
+                    "score": score,
+                }
+            )
+
+            if len(results) >= top_k:
+                break
+
+        return results
+
+
+# Global index instance (lazy loaded)
+_search_index: DocSearchIndex | None = None
+
+
+def get_search_index() -> DocSearchIndex:
+    """Get or create the search index singleton."""
+    global _search_index
+    if _search_index is None:
+        _search_index = DocSearchIndex(INDEX_PATH)
+    return _search_index
+
+
+class SearchDocsTool(BaseTool):
+    """Tool for searching AutoGPT platform documentation."""
+
+    @property
+    def name(self) -> str:
+        return "search_platform_docs"
+
+    @property
+    def description(self) -> str:
+        return (
+            "Search the AutoGPT platform documentation and support Q&A for information about "
+            "how to use the platform, create agents, configure blocks, "
+            "set up integrations, troubleshoot issues, and more. Use this when users ask "
+            "support questions or want to learn how to do something with AutoGPT."
+        )
+
+    @property
+    def parameters(self) -> dict[str, Any]:
+        return {
+            "type": "object",
+            "properties": {
+                "query": {
+                    "type": "string",
+                    "description": (
+                        "Search query describing what the user wants to learn about. "
+                        "Use keywords like 'blocks', 'agents', 'credentials', 'API', etc."
+                    ),
+                },
+            },
+            "required": ["query"],
+        }
+
+    async def _execute(
+        self,
+        user_id: str | None,
+        session: ChatSession,
+        **kwargs,
+    ) -> ToolResponseBase:
+        """Search documentation for the query.
+
+        Args:
+            user_id: User ID (may be anonymous)
+            session: Chat session
+            query: Search query
+
+        Returns:
+            DocSearchResultsResponse: List of matching documentation sections
+            NoResultsResponse: No results found
+            ErrorResponse: Error message
+        """
+        query = kwargs.get("query", "").strip()
+        session_id = session.session_id
+
+        if not query:
+            return ErrorResponse(
+                message="Please provide a search query",
+                session_id=session_id,
+            )
+
+        try:
+            index = get_search_index()
+            results = index.search(query, top_k=5)
+
+            if not results:
+                return NoResultsResponse(
+                    message=f"No documentation found for '{query}'. Try different keywords.",
+                    session_id=session_id,
+                    suggestions=[
+                        "Try more general terms like 'blocks', 'agents', 'setup'",
+                        "Check the documentation at docs.agpt.co",
+                    ],
+                )
+
+            # Convert to response format
+            doc_results = []
+            for r in results:
+                # Build documentation URL
+                path = r["path"]
+                if path.endswith(".md"):
+                    path = path[:-3]  # Remove .md extension
+                doc_url = f"{DOCS_BASE_URL}/{path}"
+
+                full_text = r["text"]
+                doc_results.append(
+                    DocSearchResult(
+                        title=r["title"],
+                        path=r["path"],
+                        section=r["heading"],
+                        snippet=(
+                            full_text[:300] + "..."
+                            if len(full_text) > 300
+                            else full_text
+                        ),
+                        content=full_text,  # Full text for LLM to read and understand
+                        score=round(r["score"], 3),
+                        doc_url=doc_url,
+                    )
+                )
+
+            return DocSearchResultsResponse(
+                message=(
+                    f"Found {len(doc_results)} relevant documentation sections. "
+                    "Use these to help answer the user's question. "
+                    "Include links to the documentation when helpful."
+                ),
+                results=doc_results,
+                count=len(doc_results),
+                query=query,
+                session_id=session_id,
+            )
+
+        except Exception as e:
+            logger.error(f"Error searching documentation: {e}", exc_info=True)
+            return ErrorResponse(
+                message="Failed to search documentation. Please try again.",
+                error=str(e),
+                session_id=session_id,
+            )

autogpt_platform/backend/backend/api/features/store/backfill_embeddings.py

@@ -0,0 +1,72 @@
+#!/usr/bin/env python3
+"""
+CLI script to backfill embeddings for store agents.
+
+Usage:
+    poetry run python -m backend.server.v2.store.backfill_embeddings [--batch-size N]
+"""
+
+import argparse
+import asyncio
+import sys
+
+import prisma
+
+
+async def main(batch_size: int = 100) -> int:
+    """Run the backfill process."""
+    # Initialize Prisma client
+    client = prisma.Prisma()
+    await client.connect()
+    prisma.register(client)
+
+    try:
+        from backend.api.features.store.embeddings import (
+            backfill_missing_embeddings,
+            get_embedding_stats,
+        )
+
+        # Get current stats
+        print("Current embedding stats:")
+        stats = await get_embedding_stats()
+        print(f"  Total approved: {stats['total_approved']}")
+        print(f"  With embeddings: {stats['with_embeddings']}")
+        print(f"  Without embeddings: {stats['without_embeddings']}")
+        print(f"  Coverage: {stats['coverage_percent']}%")
+
+        if stats["without_embeddings"] == 0:
+            print("\nAll agents already have embeddings. Nothing to do.")
+            return 0
+
+        # Run backfill
+        print(f"\nBackfilling up to {batch_size} embeddings...")
+        result = await backfill_missing_embeddings(batch_size=batch_size)
+        print(f"  Processed: {result['processed']}")
+        print(f"  Success: {result['success']}")
+        print(f"  Failed: {result['failed']}")
+
+        # Get final stats
+        print("\nFinal embedding stats:")
+        stats = await get_embedding_stats()
+        print(f"  Total approved: {stats['total_approved']}")
+        print(f"  With embeddings: {stats['with_embeddings']}")
+        print(f"  Without embeddings: {stats['without_embeddings']}")
+        print(f"  Coverage: {stats['coverage_percent']}%")
+
+        return 0 if result["failed"] == 0 else 1
+
+    finally:
+        await client.disconnect()
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Backfill embeddings for store agents")
+    parser.add_argument(
+        "--batch-size",
+        type=int,
+        default=100,
+        help="Number of embeddings to generate (default: 100)",
+    )
+    args = parser.parse_args()
+
+    sys.exit(asyncio.run(main(batch_size=args.batch_size)))

autogpt_platform/backend/backend/api/features/store/db.py

@@ -1,6 +1,5 @@
 import asyncio
 import logging
-import typing
 from datetime import datetime, timezone
 from typing import Literal
 
@@ -10,7 +9,7 @@ import prisma.errors
 import prisma.models
 import prisma.types
 
-from backend.data.db import query_raw_with_schema, transaction
+from backend.data.db import transaction
 from backend.data.graph import (
     GraphMeta,
     GraphModel,
@@ -57,95 +56,21 @@ async def get_store_agents(
     )
 
     try:
-        # If search_query is provided, use full-text search
+        # If search_query is provided, use hybrid search (embeddings + tsvector)
         if search_query:
-            offset = (page - 1) * page_size
+            from backend.api.features.store.hybrid_search import hybrid_search
 
-            # Whitelist allowed order_by columns
-            ALLOWED_ORDER_BY = {
-                "rating": "rating DESC, rank DESC",
-                "runs": "runs DESC, rank DESC",
-                "name": "agent_name ASC, rank ASC",
-                "updated_at": "updated_at DESC, rank DESC",
-            }
+            # Use hybrid search combining semantic and lexical signals
+            agents, total = await hybrid_search(
+                query=search_query,
+                featured=featured,
+                creators=creators,
+                category=category,
+                sorted_by="relevance",  # Use hybrid scoring for relevance
+                page=page,
+                page_size=page_size,
+            )
 
-            # Validate and get order clause
-            if sorted_by and sorted_by in ALLOWED_ORDER_BY:
-                order_by_clause = ALLOWED_ORDER_BY[sorted_by]
-            else:
-                order_by_clause = "updated_at DESC, rank DESC"
-
-            # Build WHERE conditions and parameters list
-            where_parts: list[str] = []
-            params: list[typing.Any] = [search_query]  # $1 - search term
-            param_index = 2  # Start at $2 for next parameter
-
-            # Always filter for available agents
-            where_parts.append("is_available = true")
-
-            if featured:
-                where_parts.append("featured = true")
-
-            if creators and creators:
-                # Use ANY with array parameter
-                where_parts.append(f"creator_username = ANY(${param_index})")
-                params.append(creators)
-                param_index += 1
-
-            if category and category:
-                where_parts.append(f"${param_index} = ANY(categories)")
-                params.append(category)
-                param_index += 1
-
-            sql_where_clause: str = " AND ".join(where_parts) if where_parts else "1=1"
-
-            # Add pagination params
-            params.extend([page_size, offset])
-            limit_param = f"${param_index}"
-            offset_param = f"${param_index + 1}"
-
-            # Execute full-text search query with parameterized values
-            sql_query = f"""
-                SELECT
-                    slug,
-                    agent_name,
-                    agent_image,
-                    creator_username,
-                    creator_avatar,
-                    sub_heading,
-                    description,
-                    runs,
-                    rating,
-                    categories,
-                    featured,
-                    is_available,
-                    updated_at,
-                    ts_rank_cd(search, query) AS rank
-                FROM {{schema_prefix}}"StoreAgent",
-                    plainto_tsquery('english', $1) AS query
-                WHERE {sql_where_clause}
-                    AND search @@ query
-                ORDER BY {order_by_clause}
-                LIMIT {limit_param} OFFSET {offset_param}
-            """
-
-            # Count query for pagination - only uses search term parameter
-            count_query = f"""
-                SELECT COUNT(*) as count
-                FROM {{schema_prefix}}"StoreAgent",
-                    plainto_tsquery('english', $1) AS query
-                WHERE {sql_where_clause}
-                    AND search @@ query
-            """
-
-            # Execute both queries with parameters
-            agents = await query_raw_with_schema(sql_query, *params)
-
-            # For count, use params without pagination (last 2 params)
-            count_params = params[:-2]
-            count_result = await query_raw_with_schema(count_query, *count_params)
-
-            total = count_result[0]["count"] if count_result else 0
             total_pages = (total + page_size - 1) // page_size
 
             # Convert raw results to StoreAgent models
@@ -1564,6 +1489,24 @@ async def review_store_submission(
                     },
                 )
 
+            # Generate embedding for approved listing (non-blocking)
+            try:
+                from backend.api.features.store.embeddings import ensure_embedding
+
+                await ensure_embedding(
+                    version_id=store_listing_version_id,
+                    name=store_listing_version.name,
+                    description=store_listing_version.description,
+                    sub_heading=store_listing_version.subHeading,
+                    categories=store_listing_version.categories or [],
+                )
+            except Exception as e:
+                # Don't fail approval if embedding generation fails
+                logger.warning(
+                    f"Failed to generate embedding for approved listing "
+                    f"{store_listing_version_id}: {e}"
+                )
+
         # If rejecting an approved agent, update the StoreListing accordingly
         if is_rejecting_approved:
             # Check if there are other approved versions

autogpt_platform/backend/backend/api/features/store/embeddings.py

@@ -0,0 +1,408 @@
+"""
+Store Listing Embeddings Service
+
+Handles generation and storage of OpenAI embeddings for store listings
+to enable semantic/hybrid search.
+"""
+
+import hashlib
+import logging
+import os
+from typing import Any
+
+import prisma
+
+logger = logging.getLogger(__name__)
+
+# OpenAI embedding model configuration
+EMBEDDING_MODEL = "text-embedding-3-small"
+EMBEDDING_DIM = 1536
+
+
+def build_searchable_text(
+    name: str,
+    description: str,
+    sub_heading: str,
+    categories: list[str],
+) -> str:
+    """
+    Build searchable text from listing version fields.
+
+    Combines relevant fields into a single string for embedding.
+    """
+    parts = []
+
+    # Name is important - include it
+    if name:
+        parts.append(name)
+
+    # Sub-heading provides context
+    if sub_heading:
+        parts.append(sub_heading)
+
+    # Description is the main content
+    if description:
+        parts.append(description)
+
+    # Categories help with semantic matching
+    if categories:
+        parts.append(" ".join(categories))
+
+    return " ".join(parts)
+
+
+def compute_content_hash(text: str) -> str:
+    """Compute MD5 hash of text for change detection."""
+    return hashlib.md5(text.encode()).hexdigest()
+
+
+async def generate_embedding(text: str) -> list[float] | None:
+    """
+    Generate embedding for text using OpenAI API.
+
+    Returns None if embedding generation fails.
+    """
+    try:
+        from openai import OpenAI
+
+        api_key = os.getenv("OPENAI_API_KEY")
+        if not api_key:
+            logger.warning("OPENAI_API_KEY not set, cannot generate embedding")
+            return None
+
+        client = OpenAI(api_key=api_key)
+
+        # Truncate text to avoid token limits (~32k chars for safety)
+        truncated_text = text[:32000]
+
+        response = client.embeddings.create(
+            model=EMBEDDING_MODEL,
+            input=truncated_text,
+        )
+
+        embedding = response.data[0].embedding
+        logger.debug(f"Generated embedding with {len(embedding)} dimensions")
+        return embedding
+
+    except Exception as e:
+        logger.error(f"Failed to generate embedding: {e}")
+        return None
+
+
+async def store_embedding(
+    version_id: str,
+    embedding: list[float],
+    searchable_text: str,
+    content_hash: str,
+    tx: prisma.Prisma | None = None,
+) -> bool:
+    """
+    Store embedding in the database.
+
+    Uses raw SQL since Prisma doesn't natively support pgvector.
+    """
+    try:
+        client = tx if tx else prisma.get_client()
+
+        # Convert embedding to PostgreSQL vector format
+        embedding_str = "[" + ",".join(str(x) for x in embedding) + "]"
+
+        # Upsert the embedding
+        # Set search_path to include public for vector type visibility
+        await client.execute_raw(
+            """
+            SET LOCAL search_path TO platform, public;
+            INSERT INTO platform."StoreListingEmbedding" (
+                "id", "storeListingVersionId", "embedding",
+                "searchableText", "contentHash", "createdAt", "updatedAt"
+            )
+            VALUES (
+                gen_random_uuid(), $1, $2::vector,
+                $3, $4, NOW(), NOW()
+            )
+            ON CONFLICT ("storeListingVersionId")
+            DO UPDATE SET
+                "embedding" = $2::vector,
+                "searchableText" = $3,
+                "contentHash" = $4,
+                "updatedAt" = NOW()
+            """,
+            version_id,
+            embedding_str,
+            searchable_text,
+            content_hash,
+        )
+
+        logger.info(f"Stored embedding for version {version_id}")
+        return True
+
+    except Exception as e:
+        logger.error(f"Failed to store embedding for version {version_id}: {e}")
+        return False
+
+
+async def get_embedding(version_id: str) -> dict[str, Any] | None:
+    """
+    Retrieve embedding record for a listing version.
+
+    Returns dict with embedding, searchableText, contentHash or None if not found.
+    """
+    try:
+        client = prisma.get_client()
+
+        result = await client.query_raw(
+            """
+            SELECT
+                "id",
+                "storeListingVersionId",
+                "embedding"::text as "embedding",
+                "searchableText",
+                "contentHash",
+                "createdAt",
+                "updatedAt"
+            FROM platform."StoreListingEmbedding"
+            WHERE "storeListingVersionId" = $1
+            """,
+            version_id,
+        )
+
+        if result and len(result) > 0:
+            return result[0]
+        return None
+
+    except Exception as e:
+        logger.error(f"Failed to get embedding for version {version_id}: {e}")
+        return None
+
+
+async def ensure_embedding(
+    version_id: str,
+    name: str,
+    description: str,
+    sub_heading: str,
+    categories: list[str],
+    force: bool = False,
+    tx: prisma.Prisma | None = None,
+) -> bool:
+    """
+    Ensure an embedding exists for the listing version.
+
+    Creates embedding if missing or if content has changed.
+    Skips if content hash matches existing embedding.
+
+    Args:
+        version_id: The StoreListingVersion ID
+        name: Agent name
+        description: Agent description
+        sub_heading: Agent sub-heading
+        categories: Agent categories
+        force: Force regeneration even if hash matches
+        tx: Optional transaction client
+
+    Returns:
+        True if embedding exists/was created, False on failure
+    """
+    try:
+        # Build searchable text and compute hash
+        searchable_text = build_searchable_text(
+            name, description, sub_heading, categories
+        )
+        content_hash = compute_content_hash(searchable_text)
+
+        # Check if embedding already exists with same hash
+        if not force:
+            existing = await get_embedding(version_id)
+            if existing and existing.get("contentHash") == content_hash:
+                logger.debug(
+                    f"Embedding for version {version_id} is up to date (hash match)"
+                )
+                return True
+
+        # Generate new embedding
+        embedding = await generate_embedding(searchable_text)
+        if embedding is None:
+            logger.warning(f"Could not generate embedding for version {version_id}")
+            return False
+
+        # Store the embedding
+        return await store_embedding(
+            version_id=version_id,
+            embedding=embedding,
+            searchable_text=searchable_text,
+            content_hash=content_hash,
+            tx=tx,
+        )
+
+    except Exception as e:
+        logger.error(f"Failed to ensure embedding for version {version_id}: {e}")
+        return False
+
+
+async def delete_embedding(version_id: str) -> bool:
+    """
+    Delete embedding for a listing version.
+
+    Note: This is usually handled automatically by CASCADE delete,
+    but provided for manual cleanup if needed.
+    """
+    try:
+        client = prisma.get_client()
+
+        await client.execute_raw(
+            """
+            DELETE FROM platform."StoreListingEmbedding"
+            WHERE "storeListingVersionId" = $1
+            """,
+            version_id,
+        )
+
+        logger.info(f"Deleted embedding for version {version_id}")
+        return True
+
+    except Exception as e:
+        logger.error(f"Failed to delete embedding for version {version_id}: {e}")
+        return False
+
+
+async def get_embedding_stats() -> dict[str, Any]:
+    """
+    Get statistics about embedding coverage.
+
+    Returns counts of:
+    - Total approved listing versions
+    - Versions with embeddings
+    - Versions without embeddings
+    """
+    try:
+        client = prisma.get_client()
+
+        # Count approved versions
+        approved_result = await client.query_raw(
+            """
+            SELECT COUNT(*) as count
+            FROM platform."StoreListingVersion"
+            WHERE "submissionStatus" = 'APPROVED'
+            AND "isDeleted" = false
+            """
+        )
+        total_approved = approved_result[0]["count"] if approved_result else 0
+
+        # Count versions with embeddings
+        embedded_result = await client.query_raw(
+            """
+            SELECT COUNT(*) as count
+            FROM platform."StoreListingVersion" slv
+            JOIN platform."StoreListingEmbedding" sle ON slv.id = sle."storeListingVersionId"
+            WHERE slv."submissionStatus" = 'APPROVED'
+            AND slv."isDeleted" = false
+            """
+        )
+        with_embeddings = embedded_result[0]["count"] if embedded_result else 0
+
+        return {
+            "total_approved": total_approved,
+            "with_embeddings": with_embeddings,
+            "without_embeddings": total_approved - with_embeddings,
+            "coverage_percent": (
+                round(with_embeddings / total_approved * 100, 1)
+                if total_approved > 0
+                else 0
+            ),
+        }
+
+    except Exception as e:
+        logger.error(f"Failed to get embedding stats: {e}")
+        return {
+            "total_approved": 0,
+            "with_embeddings": 0,
+            "without_embeddings": 0,
+            "coverage_percent": 0,
+            "error": str(e),
+        }
+
+
+async def backfill_missing_embeddings(batch_size: int = 10) -> dict[str, Any]:
+    """
+    Generate embeddings for approved listings that don't have them.
+
+    Args:
+        batch_size: Number of embeddings to generate in one call
+
+    Returns:
+        Dict with success/failure counts
+    """
+    try:
+        client = prisma.get_client()
+
+        # Find approved versions without embeddings
+        missing = await client.query_raw(
+            """
+            SELECT
+                slv.id,
+                slv.name,
+                slv.description,
+                slv."subHeading",
+                slv.categories
+            FROM platform."StoreListingVersion" slv
+            LEFT JOIN platform."StoreListingEmbedding" sle ON slv.id = sle."storeListingVersionId"
+            WHERE slv."submissionStatus" = 'APPROVED'
+            AND slv."isDeleted" = false
+            AND sle.id IS NULL
+            LIMIT $1
+            """,
+            batch_size,
+        )
+
+        if not missing:
+            return {
+                "processed": 0,
+                "success": 0,
+                "failed": 0,
+                "message": "No missing embeddings",
+            }
+
+        success = 0
+        failed = 0
+
+        for row in missing:
+            result = await ensure_embedding(
+                version_id=row["id"],
+                name=row["name"],
+                description=row["description"],
+                sub_heading=row["subHeading"],
+                categories=row["categories"] or [],
+            )
+            if result:
+                success += 1
+            else:
+                failed += 1
+
+        return {
+            "processed": len(missing),
+            "success": success,
+            "failed": failed,
+            "message": f"Backfilled {success} embeddings, {failed} failed",
+        }
+
+    except Exception as e:
+        logger.error(f"Failed to backfill embeddings: {e}")
+        return {
+            "processed": 0,
+            "success": 0,
+            "failed": 0,
+            "error": str(e),
+        }
+
+
+async def embed_query(query: str) -> list[float] | None:
+    """
+    Generate embedding for a search query.
+
+    Same as generate_embedding but with clearer intent.
+    """
+    return await generate_embedding(query)
+
+
+def embedding_to_vector_string(embedding: list[float]) -> str:
+    """Convert embedding list to PostgreSQL vector string format."""
+    return "[" + ",".join(str(x) for x in embedding) + "]"

autogpt_platform/backend/backend/api/features/store/hybrid_search.py

@@ -0,0 +1,440 @@
+"""
+Hybrid Search for Store Agents
+
+Combines semantic (embedding) search with lexical (tsvector) search
+for improved relevance in marketplace agent discovery.
+"""
+
+import logging
+from dataclasses import dataclass
+from datetime import datetime
+from typing import Any, Literal
+
+import prisma
+
+from backend.api.features.store.embeddings import (
+    embed_query,
+    embedding_to_vector_string,
+)
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class HybridSearchWeights:
+    """Weights for combining search signals."""
+
+    semantic: float = 0.35  # Embedding cosine similarity
+    lexical: float = 0.35  # tsvector ts_rank_cd score
+    category: float = 0.20  # Category match boost
+    recency: float = 0.10  # Newer agents ranked higher
+
+
+DEFAULT_WEIGHTS = HybridSearchWeights()
+
+# Minimum relevance score threshold - agents below this are filtered out
+# With weights (0.35 semantic + 0.35 lexical + 0.20 category + 0.10 recency):
+# - 0.20 means at least ~50% semantic match OR strong lexical match required
+# - Ensures only genuinely relevant results are returned
+# - Recency alone (0.10 max) won't pass the threshold
+DEFAULT_MIN_SCORE = 0.20
+
+
+@dataclass
+class HybridSearchResult:
+    """A single search result with score breakdown."""
+
+    slug: str
+    agent_name: str
+    agent_image: str
+    creator_username: str
+    creator_avatar: str
+    sub_heading: str
+    description: str
+    runs: int
+    rating: float
+    categories: list[str]
+    featured: bool
+    is_available: bool
+    updated_at: datetime
+
+    # Score breakdown (for debugging/tuning)
+    combined_score: float
+    semantic_score: float = 0.0
+    lexical_score: float = 0.0
+    category_score: float = 0.0
+    recency_score: float = 0.0
+
+
+async def hybrid_search(
+    query: str,
+    featured: bool = False,
+    creators: list[str] | None = None,
+    category: str | None = None,
+    sorted_by: (
+        Literal["relevance", "rating", "runs", "name", "updated_at"] | None
+    ) = None,
+    page: int = 1,
+    page_size: int = 20,
+    weights: HybridSearchWeights | None = None,
+    min_score: float | None = None,
+) -> tuple[list[dict[str, Any]], int]:
+    """
+    Perform hybrid search combining semantic and lexical signals.
+
+    Args:
+        query: Search query string
+        featured: Filter for featured agents only
+        creators: Filter by creator usernames
+        category: Filter by category
+        sorted_by: Sort order (relevance uses hybrid scoring)
+        page: Page number (1-indexed)
+        page_size: Results per page
+        weights: Custom weights for search signals
+        min_score: Minimum relevance score threshold (0-1). Results below
+                   this score are filtered out. Defaults to DEFAULT_MIN_SCORE.
+
+    Returns:
+        Tuple of (results list, total count). Returns empty list if no
+        results meet the minimum relevance threshold.
+    """
+    if weights is None:
+        weights = DEFAULT_WEIGHTS
+    if min_score is None:
+        min_score = DEFAULT_MIN_SCORE
+
+    offset = (page - 1) * page_size
+    client = prisma.get_client()
+
+    # Generate query embedding
+    query_embedding = await embed_query(query)
+
+    # Build WHERE clause conditions
+    where_parts: list[str] = ["sa.is_available = true"]
+    params: list[Any] = []
+    param_index = 1
+
+    # Add search query for lexical matching
+    params.append(query)
+    query_param = f"${param_index}"
+    param_index += 1
+
+    if featured:
+        where_parts.append("sa.featured = true")
+
+    if creators:
+        where_parts.append(f"sa.creator_username = ANY(${param_index})")
+        params.append(creators)
+        param_index += 1
+
+    if category:
+        where_parts.append(f"${param_index} = ANY(sa.categories)")
+        params.append(category)
+        param_index += 1
+
+    where_clause = " AND ".join(where_parts)
+
+    # Determine if we can use hybrid search (have query embedding)
+    use_hybrid = query_embedding is not None
+
+    if use_hybrid:
+        # Add embedding parameter
+        embedding_str = embedding_to_vector_string(query_embedding)
+        params.append(embedding_str)
+        embedding_param = f"${param_index}"
+        param_index += 1
+
+        # Build hybrid search query with weighted scoring
+        # The semantic score is (1 - cosine_distance), normalized to [0,1]
+        # The lexical score is ts_rank_cd, normalized by max value
+        # Set search_path to include public for vector type visibility
+        sql_query = f"""
+            SET LOCAL search_path TO platform, public;
+            WITH search_scores AS (
+                SELECT
+                    sa.*,
+                    -- Semantic score: cosine similarity (1 - distance)
+                    COALESCE(1 - (sle.embedding <=> {embedding_param}::vector), 0) as semantic_score,
+                    -- Lexical score: ts_rank_cd normalized
+                    COALESCE(ts_rank_cd(sa.search, plainto_tsquery('english', {query_param})), 0) as lexical_raw,
+                    -- Category match: 1 if query term appears in categories, else 0
+                    CASE
+                        WHEN EXISTS (
+                            SELECT 1 FROM unnest(sa.categories) cat
+                            WHERE LOWER(cat) LIKE '%' || LOWER({query_param}) || '%'
+                        ) THEN 1.0
+                        ELSE 0.0
+                    END as category_score,
+                    -- Recency score: exponential decay over 90 days
+                    EXP(-EXTRACT(EPOCH FROM (NOW() - sa.updated_at)) / (90 * 24 * 3600)) as recency_score
+                FROM platform."StoreAgent" sa
+                LEFT JOIN platform."StoreListing" sl ON sa.slug = sl.slug
+                LEFT JOIN platform."StoreListingVersion" slv ON sl."activeVersionId" = slv.id
+                LEFT JOIN platform."StoreListingEmbedding" sle ON slv.id = sle."storeListingVersionId"
+                WHERE {where_clause}
+                AND (
+                    sa.search @@ plainto_tsquery('english', {query_param})
+                    OR sle.embedding IS NOT NULL
+                )
+            ),
+            normalized AS (
+                SELECT
+                    *,
+                    -- Normalize lexical score by max in result set
+                    CASE
+                        WHEN MAX(lexical_raw) OVER () > 0
+                        THEN lexical_raw / MAX(lexical_raw) OVER ()
+                        ELSE 0
+                    END as lexical_score
+                FROM search_scores
+            ),
+            scored AS (
+                SELECT
+                    slug,
+                    agent_name,
+                    agent_image,
+                    creator_username,
+                    creator_avatar,
+                    sub_heading,
+                    description,
+                    runs,
+                    rating,
+                    categories,
+                    featured,
+                    is_available,
+                    updated_at,
+                    semantic_score,
+                    lexical_score,
+                    category_score,
+                    recency_score,
+                    (
+                        {weights.semantic} * semantic_score +
+                        {weights.lexical} * lexical_score +
+                        {weights.category} * category_score +
+                        {weights.recency} * recency_score
+                    ) as combined_score
+                FROM normalized
+            )
+            SELECT * FROM scored
+            WHERE combined_score >= {min_score}
+            ORDER BY combined_score DESC
+            LIMIT ${param_index} OFFSET ${param_index + 1}
+        """
+
+        # Add pagination params
+        params.extend([page_size, offset])
+
+        # Count query - must also filter by min_score
+        count_query = f"""
+            SET LOCAL search_path TO platform, public;
+            WITH search_scores AS (
+                SELECT
+                    sa.slug,
+                    COALESCE(1 - (sle.embedding <=> {embedding_param}::vector), 0) as semantic_score,
+                    COALESCE(ts_rank_cd(sa.search, plainto_tsquery('english', {query_param})), 0) as lexical_raw,
+                    CASE
+                        WHEN EXISTS (
+                            SELECT 1 FROM unnest(sa.categories) cat
+                            WHERE LOWER(cat) LIKE '%' || LOWER({query_param}) || '%'
+                        ) THEN 1.0
+                        ELSE 0.0
+                    END as category_score,
+                    EXP(-EXTRACT(EPOCH FROM (NOW() - sa.updated_at)) / (90 * 24 * 3600)) as recency_score
+                FROM platform."StoreAgent" sa
+                LEFT JOIN platform."StoreListing" sl ON sa.slug = sl.slug
+                LEFT JOIN platform."StoreListingVersion" slv ON sl."activeVersionId" = slv.id
+                LEFT JOIN platform."StoreListingEmbedding" sle ON slv.id = sle."storeListingVersionId"
+                WHERE {where_clause}
+                AND (
+                    sa.search @@ plainto_tsquery('english', {query_param})
+                    OR sle.embedding IS NOT NULL
+                )
+            ),
+            normalized AS (
+                SELECT
+                    slug,
+                    semantic_score,
+                    category_score,
+                    recency_score,
+                    CASE
+                        WHEN MAX(lexical_raw) OVER () > 0
+                        THEN lexical_raw / MAX(lexical_raw) OVER ()
+                        ELSE 0
+                    END as lexical_score
+                FROM search_scores
+            ),
+            scored AS (
+                SELECT
+                    slug,
+                    (
+                        {weights.semantic} * semantic_score +
+                        {weights.lexical} * lexical_score +
+                        {weights.category} * category_score +
+                        {weights.recency} * recency_score
+                    ) as combined_score
+                FROM normalized
+            )
+            SELECT COUNT(*) as count FROM scored
+            WHERE combined_score >= {min_score}
+        """
+
+    else:
+        # Fallback to lexical-only search (existing behavior)
+        # Note: For lexical-only, we still require tsvector match but don't
+        # apply min_score since ts_rank_cd isn't normalized to [0,1]
+        logger.warning("Falling back to lexical-only search (no query embedding)")
+
+        sql_query = f"""
+            WITH lexical_scores AS (
+                SELECT
+                    slug,
+                    agent_name,
+                    agent_image,
+                    creator_username,
+                    creator_avatar,
+                    sub_heading,
+                    description,
+                    runs,
+                    rating,
+                    categories,
+                    featured,
+                    is_available,
+                    updated_at,
+                    0.0 as semantic_score,
+                    ts_rank_cd(search, plainto_tsquery('english', {query_param})) as lexical_raw,
+                    CASE
+                        WHEN EXISTS (
+                            SELECT 1 FROM unnest(categories) cat
+                            WHERE LOWER(cat) LIKE '%' || LOWER({query_param}) || '%'
+                        ) THEN 1.0
+                        ELSE 0.0
+                    END as category_score,
+                    EXP(-EXTRACT(EPOCH FROM (NOW() - updated_at)) / (90 * 24 * 3600)) as recency_score
+                FROM platform."StoreAgent" sa
+                WHERE {where_clause}
+                AND search @@ plainto_tsquery('english', {query_param})
+            ),
+            normalized AS (
+                SELECT
+                    *,
+                    CASE
+                        WHEN MAX(lexical_raw) OVER () > 0
+                        THEN lexical_raw / MAX(lexical_raw) OVER ()
+                        ELSE 0
+                    END as lexical_score
+                FROM lexical_scores
+            ),
+            scored AS (
+                SELECT
+                    slug,
+                    agent_name,
+                    agent_image,
+                    creator_username,
+                    creator_avatar,
+                    sub_heading,
+                    description,
+                    runs,
+                    rating,
+                    categories,
+                    featured,
+                    is_available,
+                    updated_at,
+                    semantic_score,
+                    lexical_score,
+                    category_score,
+                    recency_score,
+                    (
+                        {weights.lexical} * lexical_score +
+                        {weights.category} * category_score +
+                        {weights.recency} * recency_score
+                    ) as combined_score
+                FROM normalized
+            )
+            SELECT * FROM scored
+            WHERE combined_score >= {min_score}
+            ORDER BY combined_score DESC
+            LIMIT ${param_index} OFFSET ${param_index + 1}
+        """
+
+        params.extend([page_size, offset])
+
+        count_query = f"""
+            WITH lexical_scores AS (
+                SELECT
+                    slug,
+                    ts_rank_cd(search, plainto_tsquery('english', {query_param})) as lexical_raw,
+                    CASE
+                        WHEN EXISTS (
+                            SELECT 1 FROM unnest(categories) cat
+                            WHERE LOWER(cat) LIKE '%' || LOWER({query_param}) || '%'
+                        ) THEN 1.0
+                        ELSE 0.0
+                    END as category_score,
+                    EXP(-EXTRACT(EPOCH FROM (NOW() - updated_at)) / (90 * 24 * 3600)) as recency_score
+                FROM platform."StoreAgent" sa
+                WHERE {where_clause}
+                AND search @@ plainto_tsquery('english', {query_param})
+            ),
+            normalized AS (
+                SELECT
+                    slug,
+                    category_score,
+                    recency_score,
+                    CASE
+                        WHEN MAX(lexical_raw) OVER () > 0
+                        THEN lexical_raw / MAX(lexical_raw) OVER ()
+                        ELSE 0
+                    END as lexical_score
+                FROM lexical_scores
+            ),
+            scored AS (
+                SELECT
+                    slug,
+                    (
+                        {weights.lexical} * lexical_score +
+                        {weights.category} * category_score +
+                        {weights.recency} * recency_score
+                    ) as combined_score
+                FROM normalized
+            )
+            SELECT COUNT(*) as count FROM scored
+            WHERE combined_score >= {min_score}
+        """
+
+    try:
+        # Execute search query
+        # Dynamic SQL is safe here - all user inputs are parameterized ($1, $2, etc.)
+        results = await client.query_raw(sql_query, *params)  # type: ignore[arg-type]
+
+        # Execute count query (without pagination params)
+        count_params = params[:-2]  # Remove LIMIT and OFFSET params
+        count_result = await client.query_raw(count_query, *count_params)  # type: ignore[arg-type]
+        total = count_result[0]["count"] if count_result else 0
+
+        logger.info(
+            f"Hybrid search for '{query}': {len(results)} results, {total} total "
+            f"(hybrid={use_hybrid})"
+        )
+
+        return results, total
+
+    except Exception as e:
+        logger.error(f"Hybrid search failed: {e}")
+        raise
+
+
+async def hybrid_search_simple(
+    query: str,
+    page: int = 1,
+    page_size: int = 20,
+) -> tuple[list[dict[str, Any]], int]:
+    """
+    Simplified hybrid search for common use cases.
+
+    Uses default weights and no filters.
+    """
+    return await hybrid_search(
+        query=query,
+        page=page,
+        page_size=page_size,
+    )

autogpt_platform/backend/backend/api/rest_api.py

@@ -39,7 +39,7 @@ import backend.data.user
 import backend.integrations.webhooks.utils
 import backend.util.service
 import backend.util.settings
-from backend.blocks.llm import LlmModel
+from backend.blocks.llm import DEFAULT_LLM_MODEL
 from backend.data.model import Credentials
 from backend.integrations.providers import ProviderName
 from backend.monitoring.instrumentation import instrument_fastapi
@@ -113,7 +113,7 @@ async def lifespan_context(app: fastapi.FastAPI):
 
     await backend.data.user.migrate_and_encrypt_user_integrations()
     await backend.data.graph.fix_llm_provider_credentials()
-    await backend.data.graph.migrate_llm_models(LlmModel.GPT4O)
+    await backend.data.graph.migrate_llm_models(DEFAULT_LLM_MODEL)
     await backend.integrations.webhooks.utils.migrate_legacy_triggered_graphs()
 
     with launch_darkly_context():

autogpt_platform/backend/backend/blocks/ai_condition.py

@@ -1,6 +1,7 @@
 from typing import Any
 
 from backend.blocks.llm import (
+    DEFAULT_LLM_MODEL,
     TEST_CREDENTIALS,
     TEST_CREDENTIALS_INPUT,
     AIBlockBase,
@@ -49,7 +50,7 @@ class AIConditionBlock(AIBlockBase):
         )
         model: LlmModel = SchemaField(
             title="LLM Model",
-            default=LlmModel.GPT4O,
+            default=DEFAULT_LLM_MODEL,
             description="The language model to use for evaluating the condition.",
             advanced=False,
         )
@@ -81,7 +82,7 @@ class AIConditionBlock(AIBlockBase):
                 "condition": "the input is an email address",
                 "yes_value": "Valid email",
                 "no_value": "Not an email",
-                "model": LlmModel.GPT4O,
+                "model": DEFAULT_LLM_MODEL,
                 "credentials": TEST_CREDENTIALS_INPUT,
             },
             test_credentials=TEST_CREDENTIALS,

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

@@ -182,13 +182,10 @@ class DataForSeoRelatedKeywordsBlock(Block):
             if results and len(results) > 0:
                 # results is a list, get the first element
                 first_result = results[0] if isinstance(results, list) else results
-                items = (
-                    first_result.get("items", [])
-                    if isinstance(first_result, dict)
-                    else []
-                )
-                # Ensure items is never None
-                if items is None:
+                # Handle missing key, null value, or valid list value
+                if isinstance(first_result, dict):
+                    items = first_result.get("items") or []
+                else:
                     items = []
                 for item in items:
                     # Extract keyword_data from the item

autogpt_platform/backend/backend/blocks/llm.py

@@ -92,8 +92,9 @@ class LlmModel(str, Enum, metaclass=LlmModelMeta):
     O1 = "o1"
     O1_MINI = "o1-mini"
     # GPT-5 models
-    GPT5 = "gpt-5-2025-08-07"
+    GPT5_2 = "gpt-5.2-2025-12-11"
     GPT5_1 = "gpt-5.1-2025-11-13"
+    GPT5 = "gpt-5-2025-08-07"
     GPT5_MINI = "gpt-5-mini-2025-08-07"
     GPT5_NANO = "gpt-5-nano-2025-08-07"
     GPT5_CHAT = "gpt-5-chat-latest"
@@ -194,8 +195,9 @@ MODEL_METADATA = {
     LlmModel.O1: ModelMetadata("openai", 200000, 100000),  # o1-2024-12-17
     LlmModel.O1_MINI: ModelMetadata("openai", 128000, 65536),  # o1-mini-2024-09-12
     # GPT-5 models
-    LlmModel.GPT5: ModelMetadata("openai", 400000, 128000),
+    LlmModel.GPT5_2: ModelMetadata("openai", 400000, 128000),
     LlmModel.GPT5_1: ModelMetadata("openai", 400000, 128000),
+    LlmModel.GPT5: ModelMetadata("openai", 400000, 128000),
     LlmModel.GPT5_MINI: ModelMetadata("openai", 400000, 128000),
     LlmModel.GPT5_NANO: ModelMetadata("openai", 400000, 128000),
     LlmModel.GPT5_CHAT: ModelMetadata("openai", 400000, 16384),
@@ -303,6 +305,8 @@ MODEL_METADATA = {
     LlmModel.V0_1_0_MD: ModelMetadata("v0", 128000, 64000),
 }
 
+DEFAULT_LLM_MODEL = LlmModel.GPT5_2
+
 for model in LlmModel:
     if model not in MODEL_METADATA:
         raise ValueError(f"Missing MODEL_METADATA metadata for model: {model}")
@@ -790,7 +794,7 @@ class AIStructuredResponseGeneratorBlock(AIBlockBase):
         )
         model: LlmModel = SchemaField(
             title="LLM Model",
-            default=LlmModel.GPT4O,
+            default=DEFAULT_LLM_MODEL,
             description="The language model to use for answering the prompt.",
             advanced=False,
         )
@@ -855,7 +859,7 @@ class AIStructuredResponseGeneratorBlock(AIBlockBase):
             input_schema=AIStructuredResponseGeneratorBlock.Input,
             output_schema=AIStructuredResponseGeneratorBlock.Output,
             test_input={
-                "model": LlmModel.GPT4O,
+                "model": DEFAULT_LLM_MODEL,
                 "credentials": TEST_CREDENTIALS_INPUT,
                 "expected_format": {
                     "key1": "value1",
@@ -1221,7 +1225,7 @@ class AITextGeneratorBlock(AIBlockBase):
         )
         model: LlmModel = SchemaField(
             title="LLM Model",
-            default=LlmModel.GPT4O,
+            default=DEFAULT_LLM_MODEL,
             description="The language model to use for answering the prompt.",
             advanced=False,
         )
@@ -1317,7 +1321,7 @@ class AITextSummarizerBlock(AIBlockBase):
         )
         model: LlmModel = SchemaField(
             title="LLM Model",
-            default=LlmModel.GPT4O,
+            default=DEFAULT_LLM_MODEL,
             description="The language model to use for summarizing the text.",
         )
         focus: str = SchemaField(
@@ -1534,7 +1538,7 @@ class AIConversationBlock(AIBlockBase):
         )
         model: LlmModel = SchemaField(
             title="LLM Model",
-            default=LlmModel.GPT4O,
+            default=DEFAULT_LLM_MODEL,
             description="The language model to use for the conversation.",
         )
         credentials: AICredentials = AICredentialsField()
@@ -1572,7 +1576,7 @@ class AIConversationBlock(AIBlockBase):
                     },
                     {"role": "user", "content": "Where was it played?"},
                 ],
-                "model": LlmModel.GPT4O,
+                "model": DEFAULT_LLM_MODEL,
                 "credentials": TEST_CREDENTIALS_INPUT,
             },
             test_credentials=TEST_CREDENTIALS,
@@ -1635,7 +1639,7 @@ class AIListGeneratorBlock(AIBlockBase):
         )
         model: LlmModel = SchemaField(
             title="LLM Model",
-            default=LlmModel.GPT4O,
+            default=DEFAULT_LLM_MODEL,
             description="The language model to use for generating the list.",
             advanced=True,
         )
@@ -1692,7 +1696,7 @@ class AIListGeneratorBlock(AIBlockBase):
                     "drawing explorers to uncover its mysteries. Each planet showcases the limitless possibilities of "
                     "fictional worlds."
                 ),
-                "model": LlmModel.GPT4O,
+                "model": DEFAULT_LLM_MODEL,
                 "credentials": TEST_CREDENTIALS_INPUT,
                 "max_retries": 3,
                 "force_json_output": False,

autogpt_platform/backend/backend/blocks/smart_decision_maker.py

@@ -226,7 +226,7 @@ class SmartDecisionMakerBlock(Block):
         )
         model: llm.LlmModel = SchemaField(
             title="LLM Model",
-            default=llm.LlmModel.GPT4O,
+            default=llm.DEFAULT_LLM_MODEL,
             description="The language model to use for answering the prompt.",
             advanced=False,
         )

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

@@ -196,6 +196,15 @@ class TestXMLParserBlockSecurity:
             async for _ in block.run(XMLParserBlock.Input(input_xml=large_xml)):
                 pass
 
+    async def test_rejects_text_outside_root(self):
+        """Ensure parser surfaces readable errors for invalid root text."""
+        block = XMLParserBlock()
+        invalid_xml = "<root><child>value</child></root> trailing"
+
+        with pytest.raises(ValueError, match="text outside the root element"):
+            async for _ in block.run(XMLParserBlock.Input(input_xml=invalid_xml)):
+                pass
+
 
 class TestStoreMediaFileSecurity:
     """Test file storage security limits."""

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

@@ -28,7 +28,7 @@ class TestLLMStatsTracking:
 
             response = await llm.llm_call(
                 credentials=llm.TEST_CREDENTIALS,
-                llm_model=llm.LlmModel.GPT4O,
+                llm_model=llm.DEFAULT_LLM_MODEL,
                 prompt=[{"role": "user", "content": "Hello"}],
                 max_tokens=100,
             )
@@ -65,7 +65,7 @@ class TestLLMStatsTracking:
         input_data = llm.AIStructuredResponseGeneratorBlock.Input(
             prompt="Test prompt",
             expected_format={"key1": "desc1", "key2": "desc2"},
-            model=llm.LlmModel.GPT4O,
+            model=llm.DEFAULT_LLM_MODEL,
             credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore  # type: ignore
         )
 
@@ -109,7 +109,7 @@ class TestLLMStatsTracking:
         # Run the block
         input_data = llm.AITextGeneratorBlock.Input(
             prompt="Generate text",
-            model=llm.LlmModel.GPT4O,
+            model=llm.DEFAULT_LLM_MODEL,
             credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore
         )
 
@@ -170,7 +170,7 @@ class TestLLMStatsTracking:
         input_data = llm.AIStructuredResponseGeneratorBlock.Input(
             prompt="Test prompt",
             expected_format={"key1": "desc1", "key2": "desc2"},
-            model=llm.LlmModel.GPT4O,
+            model=llm.DEFAULT_LLM_MODEL,
             credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore
             retry=2,
         )
@@ -228,7 +228,7 @@ class TestLLMStatsTracking:
 
         input_data = llm.AITextSummarizerBlock.Input(
             text=long_text,
-            model=llm.LlmModel.GPT4O,
+            model=llm.DEFAULT_LLM_MODEL,
             credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore
             max_tokens=100,  # Small chunks
             chunk_overlap=10,
@@ -299,7 +299,7 @@ class TestLLMStatsTracking:
             # Test with very short text (should only need 1 chunk + 1 final summary)
             input_data = llm.AITextSummarizerBlock.Input(
                 text="This is a short text.",
-                model=llm.LlmModel.GPT4O,
+                model=llm.DEFAULT_LLM_MODEL,
                 credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore
                 max_tokens=1000,  # Large enough to avoid chunking
             )
@@ -346,7 +346,7 @@ class TestLLMStatsTracking:
                 {"role": "assistant", "content": "Hi there!"},
                 {"role": "user", "content": "How are you?"},
             ],
-            model=llm.LlmModel.GPT4O,
+            model=llm.DEFAULT_LLM_MODEL,
             credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore
         )
 
@@ -387,7 +387,7 @@ class TestLLMStatsTracking:
         # Run the block
         input_data = llm.AIListGeneratorBlock.Input(
             focus="test items",
-            model=llm.LlmModel.GPT4O,
+            model=llm.DEFAULT_LLM_MODEL,
             credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore
             max_retries=3,
         )
@@ -469,7 +469,7 @@ class TestLLMStatsTracking:
         input_data = llm.AIStructuredResponseGeneratorBlock.Input(
             prompt="Test",
             expected_format={"result": "desc"},
-            model=llm.LlmModel.GPT4O,
+            model=llm.DEFAULT_LLM_MODEL,
             credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore
         )
 
@@ -513,7 +513,7 @@ class TestAITextSummarizerValidation:
         # Create input data
         input_data = llm.AITextSummarizerBlock.Input(
             text="Some text to summarize",
-            model=llm.LlmModel.GPT4O,
+            model=llm.DEFAULT_LLM_MODEL,
             credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore
             style=llm.SummaryStyle.BULLET_POINTS,
         )
@@ -558,7 +558,7 @@ class TestAITextSummarizerValidation:
         # Create input data
         input_data = llm.AITextSummarizerBlock.Input(
             text="Some text to summarize",
-            model=llm.LlmModel.GPT4O,
+            model=llm.DEFAULT_LLM_MODEL,
             credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore
             style=llm.SummaryStyle.BULLET_POINTS,
             max_tokens=1000,
@@ -593,7 +593,7 @@ class TestAITextSummarizerValidation:
         # Create input data
         input_data = llm.AITextSummarizerBlock.Input(
             text="Some text to summarize",
-            model=llm.LlmModel.GPT4O,
+            model=llm.DEFAULT_LLM_MODEL,
             credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore
         )
 
@@ -623,7 +623,7 @@ class TestAITextSummarizerValidation:
         # Create input data
         input_data = llm.AITextSummarizerBlock.Input(
             text="Some text to summarize",
-            model=llm.LlmModel.GPT4O,
+            model=llm.DEFAULT_LLM_MODEL,
             credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore
             max_tokens=1000,
         )
@@ -654,7 +654,7 @@ class TestAITextSummarizerValidation:
         # Create input data
         input_data = llm.AITextSummarizerBlock.Input(
             text="Some text to summarize",
-            model=llm.LlmModel.GPT4O,
+            model=llm.DEFAULT_LLM_MODEL,
             credentials=llm.TEST_CREDENTIALS_INPUT,  # type: ignore
         )
 

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

@@ -233,7 +233,7 @@ async def test_smart_decision_maker_tracks_llm_stats():
         # Create test input
         input_data = SmartDecisionMakerBlock.Input(
             prompt="Should I continue with this task?",
-            model=llm_module.LlmModel.GPT4O,
+            model=llm_module.DEFAULT_LLM_MODEL,
             credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
             agent_mode_max_iterations=0,
         )
@@ -335,7 +335,7 @@ async def test_smart_decision_maker_parameter_validation():
 
         input_data = SmartDecisionMakerBlock.Input(
             prompt="Search for keywords",
-            model=llm_module.LlmModel.GPT4O,
+            model=llm_module.DEFAULT_LLM_MODEL,
             credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
             retry=2,  # Set retry to 2 for testing
             agent_mode_max_iterations=0,
@@ -402,7 +402,7 @@ async def test_smart_decision_maker_parameter_validation():
 
         input_data = SmartDecisionMakerBlock.Input(
             prompt="Search for keywords",
-            model=llm_module.LlmModel.GPT4O,
+            model=llm_module.DEFAULT_LLM_MODEL,
             credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
             agent_mode_max_iterations=0,
         )
@@ -462,7 +462,7 @@ async def test_smart_decision_maker_parameter_validation():
 
         input_data = SmartDecisionMakerBlock.Input(
             prompt="Search for keywords",
-            model=llm_module.LlmModel.GPT4O,
+            model=llm_module.DEFAULT_LLM_MODEL,
             credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
             agent_mode_max_iterations=0,
         )
@@ -526,7 +526,7 @@ async def test_smart_decision_maker_parameter_validation():
 
         input_data = SmartDecisionMakerBlock.Input(
             prompt="Search for keywords",
-            model=llm_module.LlmModel.GPT4O,
+            model=llm_module.DEFAULT_LLM_MODEL,
             credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
             agent_mode_max_iterations=0,
         )
@@ -648,7 +648,7 @@ async def test_smart_decision_maker_raw_response_conversion():
 
         input_data = SmartDecisionMakerBlock.Input(
             prompt="Test prompt",
-            model=llm_module.LlmModel.GPT4O,
+            model=llm_module.DEFAULT_LLM_MODEL,
             credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
             retry=2,
             agent_mode_max_iterations=0,
@@ -722,7 +722,7 @@ async def test_smart_decision_maker_raw_response_conversion():
     ):
         input_data = SmartDecisionMakerBlock.Input(
             prompt="Simple prompt",
-            model=llm_module.LlmModel.GPT4O,
+            model=llm_module.DEFAULT_LLM_MODEL,
             credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
             agent_mode_max_iterations=0,
         )
@@ -778,7 +778,7 @@ async def test_smart_decision_maker_raw_response_conversion():
     ):
         input_data = SmartDecisionMakerBlock.Input(
             prompt="Another test",
-            model=llm_module.LlmModel.GPT4O,
+            model=llm_module.DEFAULT_LLM_MODEL,
             credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
             agent_mode_max_iterations=0,
         )
@@ -931,7 +931,7 @@ async def test_smart_decision_maker_agent_mode():
         # Test agent mode with max_iterations = 3
         input_data = SmartDecisionMakerBlock.Input(
             prompt="Complete this task using tools",
-            model=llm_module.LlmModel.GPT4O,
+            model=llm_module.DEFAULT_LLM_MODEL,
             credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
             agent_mode_max_iterations=3,  # Enable agent mode with 3 max iterations
         )
@@ -1020,7 +1020,7 @@ async def test_smart_decision_maker_traditional_mode_default():
         # Test default behavior (traditional mode)
         input_data = SmartDecisionMakerBlock.Input(
             prompt="Test prompt",
-            model=llm_module.LlmModel.GPT4O,
+            model=llm_module.DEFAULT_LLM_MODEL,
             credentials=llm_module.TEST_CREDENTIALS_INPUT,  # type: ignore
             agent_mode_max_iterations=0,  # Traditional mode
         )

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

@@ -373,7 +373,7 @@ async def test_output_yielding_with_dynamic_fields():
             input_data = block.input_schema(
                 prompt="Create a user dictionary",
                 credentials=llm.TEST_CREDENTIALS_INPUT,
-                model=llm.LlmModel.GPT4O,
+                model=llm.DEFAULT_LLM_MODEL,
                 agent_mode_max_iterations=0,  # Use traditional mode to test output yielding
             )
 
@@ -594,7 +594,7 @@ async def test_validation_errors_dont_pollute_conversation():
                 input_data = block.input_schema(
                     prompt="Test prompt",
                     credentials=llm.TEST_CREDENTIALS_INPUT,
-                    model=llm.LlmModel.GPT4O,
+                    model=llm.DEFAULT_LLM_MODEL,
                     retry=3,  # Allow retries
                     agent_mode_max_iterations=1,
                 )

autogpt_platform/backend/backend/blocks/xml_parser.py

@@ -1,5 +1,5 @@
 from gravitasml.parser import Parser
-from gravitasml.token import tokenize
+from gravitasml.token import Token, tokenize
 
 from backend.data.block import Block, BlockOutput, BlockSchemaInput, BlockSchemaOutput
 from backend.data.model import SchemaField
@@ -25,6 +25,38 @@ class XMLParserBlock(Block):
             ],
         )
 
+    @staticmethod
+    def _validate_tokens(tokens: list[Token]) -> None:
+        """Ensure the XML has a single root element and no stray text."""
+        if not tokens:
+            raise ValueError("XML input is empty.")
+
+        depth = 0
+        root_seen = False
+
+        for token in tokens:
+            if token.type == "TAG_OPEN":
+                if depth == 0 and root_seen:
+                    raise ValueError("XML must have a single root element.")
+                depth += 1
+                if depth == 1:
+                    root_seen = True
+            elif token.type == "TAG_CLOSE":
+                depth -= 1
+                if depth < 0:
+                    raise SyntaxError("Unexpected closing tag in XML input.")
+            elif token.type in {"TEXT", "ESCAPE"}:
+                if depth == 0 and token.value:
+                    raise ValueError(
+                        "XML contains text outside the root element; "
+                        "wrap content in a single root tag."
+                    )
+
+        if depth != 0:
+            raise SyntaxError("Unclosed tag detected in XML input.")
+        if not root_seen:
+            raise ValueError("XML must include a root element.")
+
     async def run(self, input_data: Input, **kwargs) -> BlockOutput:
         # Security fix: Add size limits to prevent XML bomb attacks
         MAX_XML_SIZE = 10 * 1024 * 1024  # 10MB limit for XML input
@@ -35,7 +67,9 @@ class XMLParserBlock(Block):
             )
 
         try:
-            tokens = tokenize(input_data.input_xml)
+            tokens = list(tokenize(input_data.input_xml))
+            self._validate_tokens(tokens)
+
             parser = Parser(tokens)
             parsed_result = parser.parse()
             yield "parsed_xml", parsed_result

autogpt_platform/backend/backend/blocks/youtube.py

@@ -111,6 +111,8 @@ class TranscribeYoutubeVideoBlock(Block):
                 return parsed_url.path.split("/")[2]
             if parsed_url.path[:3] == "/v/":
                 return parsed_url.path.split("/")[2]
+            if parsed_url.path.startswith("/shorts/"):
+                return parsed_url.path.split("/")[2]
         raise ValueError(f"Invalid YouTube URL: {url}")
 
     def get_transcript(

autogpt_platform/backend/backend/data/block_cost_config.py

@@ -59,12 +59,13 @@ from backend.integrations.credentials_store import (
 
 MODEL_COST: dict[LlmModel, int] = {
     LlmModel.O3: 4,
-    LlmModel.O3_MINI: 2,  # $1.10 / $4.40
-    LlmModel.O1: 16,  # $15 / $60
+    LlmModel.O3_MINI: 2,
+    LlmModel.O1: 16,
     LlmModel.O1_MINI: 4,
     # GPT-5 models
-    LlmModel.GPT5: 2,
+    LlmModel.GPT5_2: 6,
     LlmModel.GPT5_1: 5,
+    LlmModel.GPT5: 2,
     LlmModel.GPT5_MINI: 1,
     LlmModel.GPT5_NANO: 1,
     LlmModel.GPT5_CHAT: 5,
@@ -87,7 +88,7 @@ MODEL_COST: dict[LlmModel, int] = {
     LlmModel.AIML_API_LLAMA3_3_70B: 1,
     LlmModel.AIML_API_META_LLAMA_3_1_70B: 1,
     LlmModel.AIML_API_LLAMA_3_2_3B: 1,
-    LlmModel.LLAMA3_3_70B: 1,  # $0.59 / $0.79
+    LlmModel.LLAMA3_3_70B: 1,
     LlmModel.LLAMA3_1_8B: 1,
     LlmModel.OLLAMA_LLAMA3_3: 1,
     LlmModel.OLLAMA_LLAMA3_2: 1,

autogpt_platform/backend/backend/data/credit.py

@@ -341,6 +341,19 @@ class UserCreditBase(ABC):
 
         if result:
             # UserBalance is already updated by the CTE
+
+            # Clear insufficient funds notification flags when credits are added
+            # so user can receive alerts again if they run out in the future.
+            if transaction.amount > 0 and transaction.type in [
+                CreditTransactionType.GRANT,
+                CreditTransactionType.TOP_UP,
+            ]:
+                from backend.executor.manager import (
+                    clear_insufficient_funds_notifications,
+                )
+
+                await clear_insufficient_funds_notifications(user_id)
+
             return result[0]["balance"]
 
     async def _add_transaction(
@@ -530,6 +543,22 @@ class UserCreditBase(ABC):
         if result:
             new_balance, tx_key = result[0]["balance"], result[0]["transactionKey"]
             # UserBalance is already updated by the CTE
+
+            # Clear insufficient funds notification flags when credits are added
+            # so user can receive alerts again if they run out in the future.
+            if (
+                amount > 0
+                and is_active
+                and transaction_type
+                in [CreditTransactionType.GRANT, CreditTransactionType.TOP_UP]
+            ):
+                # Lazy import to avoid circular dependency with executor.manager
+                from backend.executor.manager import (
+                    clear_insufficient_funds_notifications,
+                )
+
+                await clear_insufficient_funds_notifications(user_id)
+
             return new_balance, tx_key
 
         # If no result, either user doesn't exist or insufficient balance

autogpt_platform/backend/backend/data/understanding.py

@@ -0,0 +1,429 @@
+"""Data models and access layer for user business understanding."""
+
+import logging
+from datetime import datetime
+from typing import Any, Optional, cast
+
+import pydantic
+from prisma.models import UserBusinessUnderstanding
+from prisma.types import (
+    UserBusinessUnderstandingCreateInput,
+    UserBusinessUnderstandingUpdateInput,
+)
+
+from backend.data.redis_client import get_redis_async
+from backend.util.json import SafeJson
+
+logger = logging.getLogger(__name__)
+
+# Cache configuration
+CACHE_KEY_PREFIX = "understanding"
+CACHE_TTL_SECONDS = 48 * 60 * 60  # 48 hours
+
+
+def _cache_key(user_id: str) -> str:
+    """Generate cache key for user business understanding."""
+    return f"{CACHE_KEY_PREFIX}:{user_id}"
+
+
+def _json_to_list(value: Any) -> list[str]:
+    """Convert Json field to list[str], handling None."""
+    if value is None:
+        return []
+    if isinstance(value, list):
+        return cast(list[str], value)
+    return []
+
+
+class BusinessUnderstandingInput(pydantic.BaseModel):
+    """Input model for updating business understanding - all fields optional for incremental updates."""
+
+    # User info
+    user_name: Optional[str] = pydantic.Field(None, description="The user's name")
+    job_title: Optional[str] = pydantic.Field(None, description="The user's job title")
+
+    # Business basics
+    business_name: Optional[str] = pydantic.Field(
+        None, description="Name of the user's business"
+    )
+    industry: Optional[str] = pydantic.Field(None, description="Industry or sector")
+    business_size: Optional[str] = pydantic.Field(
+        None, description="Company size (e.g., '1-10', '11-50')"
+    )
+    user_role: Optional[str] = pydantic.Field(
+        None,
+        description="User's role in the organization (e.g., 'decision maker', 'implementer')",
+    )
+
+    # Processes & activities
+    key_workflows: Optional[list[str]] = pydantic.Field(
+        None, description="Key business workflows"
+    )
+    daily_activities: Optional[list[str]] = pydantic.Field(
+        None, description="Daily activities performed"
+    )
+
+    # Pain points & goals
+    pain_points: Optional[list[str]] = pydantic.Field(
+        None, description="Current pain points"
+    )
+    bottlenecks: Optional[list[str]] = pydantic.Field(
+        None, description="Process bottlenecks"
+    )
+    manual_tasks: Optional[list[str]] = pydantic.Field(
+        None, description="Manual/repetitive tasks"
+    )
+    automation_goals: Optional[list[str]] = pydantic.Field(
+        None, description="Desired automation goals"
+    )
+
+    # Current tools
+    current_software: Optional[list[str]] = pydantic.Field(
+        None, description="Software/tools currently used"
+    )
+    existing_automation: Optional[list[str]] = pydantic.Field(
+        None, description="Existing automations"
+    )
+
+    # Additional context
+    additional_notes: Optional[str] = pydantic.Field(
+        None, description="Any additional context"
+    )
+
+
+class BusinessUnderstanding(pydantic.BaseModel):
+    """Full business understanding model returned from database."""
+
+    id: str
+    user_id: str
+    created_at: datetime
+    updated_at: datetime
+
+    # User info
+    user_name: Optional[str] = None
+    job_title: Optional[str] = None
+
+    # Business basics
+    business_name: Optional[str] = None
+    industry: Optional[str] = None
+    business_size: Optional[str] = None
+    user_role: Optional[str] = None
+
+    # Processes & activities
+    key_workflows: list[str] = pydantic.Field(default_factory=list)
+    daily_activities: list[str] = pydantic.Field(default_factory=list)
+
+    # Pain points & goals
+    pain_points: list[str] = pydantic.Field(default_factory=list)
+    bottlenecks: list[str] = pydantic.Field(default_factory=list)
+    manual_tasks: list[str] = pydantic.Field(default_factory=list)
+    automation_goals: list[str] = pydantic.Field(default_factory=list)
+
+    # Current tools
+    current_software: list[str] = pydantic.Field(default_factory=list)
+    existing_automation: list[str] = pydantic.Field(default_factory=list)
+
+    # Additional context
+    additional_notes: Optional[str] = None
+
+    @classmethod
+    def from_db(cls, db_record: UserBusinessUnderstanding) -> "BusinessUnderstanding":
+        """Convert database record to Pydantic model."""
+        return cls(
+            id=db_record.id,
+            user_id=db_record.userId,
+            created_at=db_record.createdAt,
+            updated_at=db_record.updatedAt,
+            user_name=db_record.userName,
+            job_title=db_record.jobTitle,
+            business_name=db_record.businessName,
+            industry=db_record.industry,
+            business_size=db_record.businessSize,
+            user_role=db_record.userRole,
+            key_workflows=_json_to_list(db_record.keyWorkflows),
+            daily_activities=_json_to_list(db_record.dailyActivities),
+            pain_points=_json_to_list(db_record.painPoints),
+            bottlenecks=_json_to_list(db_record.bottlenecks),
+            manual_tasks=_json_to_list(db_record.manualTasks),
+            automation_goals=_json_to_list(db_record.automationGoals),
+            current_software=_json_to_list(db_record.currentSoftware),
+            existing_automation=_json_to_list(db_record.existingAutomation),
+            additional_notes=db_record.additionalNotes,
+        )
+
+
+def _merge_lists(existing: list | None, new: list | None) -> list | None:
+    """Merge two lists, removing duplicates while preserving order."""
+    if new is None:
+        return existing
+    if existing is None:
+        return new
+    # Preserve order, add new items that don't exist
+    merged = list(existing)
+    for item in new:
+        if item not in merged:
+            merged.append(item)
+    return merged
+
+
+async def _get_from_cache(user_id: str) -> Optional[BusinessUnderstanding]:
+    """Get business understanding from Redis cache."""
+    try:
+        redis = await get_redis_async()
+        cached_data = await redis.get(_cache_key(user_id))
+        if cached_data:
+            return BusinessUnderstanding.model_validate_json(cached_data)
+    except Exception as e:
+        logger.warning(f"Failed to get understanding from cache: {e}")
+    return None
+
+
+async def _set_cache(user_id: str, understanding: BusinessUnderstanding) -> None:
+    """Set business understanding in Redis cache with TTL."""
+    try:
+        redis = await get_redis_async()
+        await redis.setex(
+            _cache_key(user_id),
+            CACHE_TTL_SECONDS,
+            understanding.model_dump_json(),
+        )
+    except Exception as e:
+        logger.warning(f"Failed to set understanding in cache: {e}")
+
+
+async def _delete_cache(user_id: str) -> None:
+    """Delete business understanding from Redis cache."""
+    try:
+        redis = await get_redis_async()
+        await redis.delete(_cache_key(user_id))
+    except Exception as e:
+        logger.warning(f"Failed to delete understanding from cache: {e}")
+
+
+async def get_business_understanding(
+    user_id: str,
+) -> Optional[BusinessUnderstanding]:
+    """Get the business understanding for a user.
+
+    Checks cache first, falls back to database if not cached.
+    Results are cached for 48 hours.
+    """
+    # Try cache first
+    cached = await _get_from_cache(user_id)
+    if cached:
+        logger.debug(f"Business understanding cache hit for user {user_id}")
+        return cached
+
+    # Cache miss - load from database
+    logger.debug(f"Business understanding cache miss for user {user_id}")
+    record = await UserBusinessUnderstanding.prisma().find_unique(
+        where={"userId": user_id}
+    )
+    if record is None:
+        return None
+
+    understanding = BusinessUnderstanding.from_db(record)
+
+    # Store in cache for next time
+    await _set_cache(user_id, understanding)
+
+    return understanding
+
+
+async def upsert_business_understanding(
+    user_id: str,
+    data: BusinessUnderstandingInput,
+) -> BusinessUnderstanding:
+    """
+    Create or update business understanding with incremental merge strategy.
+
+    - String fields: new value overwrites if provided (not None)
+    - List fields: new items are appended to existing (deduplicated)
+    """
+    # Get existing record for merge
+    existing = await UserBusinessUnderstanding.prisma().find_unique(
+        where={"userId": user_id}
+    )
+
+    # Build update data with merge strategy
+    update_data: UserBusinessUnderstandingUpdateInput = {}
+    create_data: dict[str, Any] = {"userId": user_id}
+
+    # String fields - overwrite if provided
+    if data.user_name is not None:
+        update_data["userName"] = data.user_name
+        create_data["userName"] = data.user_name
+    if data.job_title is not None:
+        update_data["jobTitle"] = data.job_title
+        create_data["jobTitle"] = data.job_title
+    if data.business_name is not None:
+        update_data["businessName"] = data.business_name
+        create_data["businessName"] = data.business_name
+    if data.industry is not None:
+        update_data["industry"] = data.industry
+        create_data["industry"] = data.industry
+    if data.business_size is not None:
+        update_data["businessSize"] = data.business_size
+        create_data["businessSize"] = data.business_size
+    if data.user_role is not None:
+        update_data["userRole"] = data.user_role
+        create_data["userRole"] = data.user_role
+    if data.additional_notes is not None:
+        update_data["additionalNotes"] = data.additional_notes
+        create_data["additionalNotes"] = data.additional_notes
+
+    # List fields - merge with existing
+    if data.key_workflows is not None:
+        existing_list = _json_to_list(existing.keyWorkflows) if existing else None
+        merged = _merge_lists(existing_list, data.key_workflows)
+        update_data["keyWorkflows"] = SafeJson(merged)
+        create_data["keyWorkflows"] = SafeJson(merged)
+
+    if data.daily_activities is not None:
+        existing_list = _json_to_list(existing.dailyActivities) if existing else None
+        merged = _merge_lists(existing_list, data.daily_activities)
+        update_data["dailyActivities"] = SafeJson(merged)
+        create_data["dailyActivities"] = SafeJson(merged)
+
+    if data.pain_points is not None:
+        existing_list = _json_to_list(existing.painPoints) if existing else None
+        merged = _merge_lists(existing_list, data.pain_points)
+        update_data["painPoints"] = SafeJson(merged)
+        create_data["painPoints"] = SafeJson(merged)
+
+    if data.bottlenecks is not None:
+        existing_list = _json_to_list(existing.bottlenecks) if existing else None
+        merged = _merge_lists(existing_list, data.bottlenecks)
+        update_data["bottlenecks"] = SafeJson(merged)
+        create_data["bottlenecks"] = SafeJson(merged)
+
+    if data.manual_tasks is not None:
+        existing_list = _json_to_list(existing.manualTasks) if existing else None
+        merged = _merge_lists(existing_list, data.manual_tasks)
+        update_data["manualTasks"] = SafeJson(merged)
+        create_data["manualTasks"] = SafeJson(merged)
+
+    if data.automation_goals is not None:
+        existing_list = _json_to_list(existing.automationGoals) if existing else None
+        merged = _merge_lists(existing_list, data.automation_goals)
+        update_data["automationGoals"] = SafeJson(merged)
+        create_data["automationGoals"] = SafeJson(merged)
+
+    if data.current_software is not None:
+        existing_list = _json_to_list(existing.currentSoftware) if existing else None
+        merged = _merge_lists(existing_list, data.current_software)
+        update_data["currentSoftware"] = SafeJson(merged)
+        create_data["currentSoftware"] = SafeJson(merged)
+
+    if data.existing_automation is not None:
+        existing_list = _json_to_list(existing.existingAutomation) if existing else None
+        merged = _merge_lists(existing_list, data.existing_automation)
+        update_data["existingAutomation"] = SafeJson(merged)
+        create_data["existingAutomation"] = SafeJson(merged)
+
+    # Upsert
+    record = await UserBusinessUnderstanding.prisma().upsert(
+        where={"userId": user_id},
+        data={
+            "create": UserBusinessUnderstandingCreateInput(**create_data),
+            "update": update_data,
+        },
+    )
+
+    understanding = BusinessUnderstanding.from_db(record)
+
+    # Update cache with new understanding
+    await _set_cache(user_id, understanding)
+
+    return understanding
+
+
+async def clear_business_understanding(user_id: str) -> bool:
+    """Clear/delete business understanding for a user from both DB and cache."""
+    # Delete from cache first
+    await _delete_cache(user_id)
+
+    try:
+        await UserBusinessUnderstanding.prisma().delete(where={"userId": user_id})
+        return True
+    except Exception:
+        # Record might not exist
+        return False
+
+
+def format_understanding_for_prompt(understanding: BusinessUnderstanding) -> str:
+    """Format business understanding as text for system prompt injection."""
+    sections = []
+
+    # User info section
+    user_info = []
+    if understanding.user_name:
+        user_info.append(f"Name: {understanding.user_name}")
+    if understanding.job_title:
+        user_info.append(f"Job Title: {understanding.job_title}")
+    if user_info:
+        sections.append("## User\n" + "\n".join(user_info))
+
+    # Business section
+    business_info = []
+    if understanding.business_name:
+        business_info.append(f"Company: {understanding.business_name}")
+    if understanding.industry:
+        business_info.append(f"Industry: {understanding.industry}")
+    if understanding.business_size:
+        business_info.append(f"Size: {understanding.business_size}")
+    if understanding.user_role:
+        business_info.append(f"Role Context: {understanding.user_role}")
+    if business_info:
+        sections.append("## Business\n" + "\n".join(business_info))
+
+    # Processes section
+    processes = []
+    if understanding.key_workflows:
+        processes.append(f"Key Workflows: {', '.join(understanding.key_workflows)}")
+    if understanding.daily_activities:
+        processes.append(
+            f"Daily Activities: {', '.join(understanding.daily_activities)}"
+        )
+    if processes:
+        sections.append("## Processes\n" + "\n".join(processes))
+
+    # Pain points section
+    pain_points = []
+    if understanding.pain_points:
+        pain_points.append(f"Pain Points: {', '.join(understanding.pain_points)}")
+    if understanding.bottlenecks:
+        pain_points.append(f"Bottlenecks: {', '.join(understanding.bottlenecks)}")
+    if understanding.manual_tasks:
+        pain_points.append(f"Manual Tasks: {', '.join(understanding.manual_tasks)}")
+    if pain_points:
+        sections.append("## Pain Points\n" + "\n".join(pain_points))
+
+    # Goals section
+    if understanding.automation_goals:
+        sections.append(
+            "## Automation Goals\n"
+            + "\n".join(f"- {goal}" for goal in understanding.automation_goals)
+        )
+
+    # Current tools section
+    tools_info = []
+    if understanding.current_software:
+        tools_info.append(
+            f"Current Software: {', '.join(understanding.current_software)}"
+        )
+    if understanding.existing_automation:
+        tools_info.append(
+            f"Existing Automation: {', '.join(understanding.existing_automation)}"
+        )
+    if tools_info:
+        sections.append("## Current Tools\n" + "\n".join(tools_info))
+
+    # Additional notes
+    if understanding.additional_notes:
+        sections.append(f"## Additional Context\n{understanding.additional_notes}")
+
+    if not sections:
+        return ""
+
+    return "# User Business Context\n\n" + "\n\n".join(sections)

autogpt_platform/backend/backend/executor/manager.py

@@ -114,6 +114,40 @@ utilization_gauge = Gauge(
     "Ratio of active graph runs to max graph workers",
 )
 
+# Redis key prefix for tracking insufficient funds Discord notifications.
+# We only send one notification per user per agent until they top up credits.
+INSUFFICIENT_FUNDS_NOTIFIED_PREFIX = "insufficient_funds_discord_notified"
+# TTL for the notification flag (30 days) - acts as a fallback cleanup
+INSUFFICIENT_FUNDS_NOTIFIED_TTL_SECONDS = 30 * 24 * 60 * 60
+
+
+async def clear_insufficient_funds_notifications(user_id: str) -> int:
+    """
+    Clear all insufficient funds notification flags for a user.
+
+    This should be called when a user tops up their credits, allowing
+    Discord notifications to be sent again if they run out of funds.
+
+    Args:
+        user_id: The user ID to clear notifications for.
+
+    Returns:
+        The number of keys that were deleted.
+    """
+    try:
+        redis_client = await redis.get_redis_async()
+        pattern = f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:*"
+        keys = [key async for key in redis_client.scan_iter(match=pattern)]
+        if keys:
+            return await redis_client.delete(*keys)
+        return 0
+    except Exception as e:
+        logger.warning(
+            f"Failed to clear insufficient funds notification flags for user "
+            f"{user_id}: {e}"
+        )
+        return 0
+
 
 # Thread-local storage for ExecutionProcessor instances
 _tls = threading.local()
@@ -1261,12 +1295,40 @@ class ExecutionProcessor:
         graph_id: str,
         e: InsufficientBalanceError,
     ):
+        # Check if we've already sent a notification for this user+agent combo.
+        # We only send one notification per user per agent until they top up credits.
+        redis_key = f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:{graph_id}"
+        try:
+            redis_client = redis.get_redis()
+            # SET NX returns True only if the key was newly set (didn't exist)
+            is_new_notification = redis_client.set(
+                redis_key,
+                "1",
+                nx=True,
+                ex=INSUFFICIENT_FUNDS_NOTIFIED_TTL_SECONDS,
+            )
+            if not is_new_notification:
+                # Already notified for this user+agent, skip all notifications
+                logger.debug(
+                    f"Skipping duplicate insufficient funds notification for "
+                    f"user={user_id}, graph={graph_id}"
+                )
+                return
+        except Exception as redis_error:
+            # If Redis fails, log and continue to send the notification
+            # (better to occasionally duplicate than to never notify)
+            logger.warning(
+                f"Failed to check/set insufficient funds notification flag in Redis: "
+                f"{redis_error}"
+            )
+
         shortfall = abs(e.amount) - e.balance
         metadata = db_client.get_graph_metadata(graph_id)
         base_url = (
             settings.config.frontend_base_url or settings.config.platform_base_url
         )
 
+        # Queue user email notification
         queue_notification(
             NotificationEventModel(
                 user_id=user_id,
@@ -1280,6 +1342,7 @@ class ExecutionProcessor:
             )
         )
 
+        # Send Discord system alert
         try:
             user_email = db_client.get_user_email_by_id(user_id)
 

autogpt_platform/backend/backend/executor/manager_insufficient_funds_test.py

@@ -0,0 +1,560 @@
+from unittest.mock import AsyncMock, MagicMock, patch
+
+import pytest
+from prisma.enums import NotificationType
+
+from backend.data.notifications import ZeroBalanceData
+from backend.executor.manager import (
+    INSUFFICIENT_FUNDS_NOTIFIED_PREFIX,
+    ExecutionProcessor,
+    clear_insufficient_funds_notifications,
+)
+from backend.util.exceptions import InsufficientBalanceError
+from backend.util.test import SpinTestServer
+
+
+async def async_iter(items):
+    """Helper to create an async iterator from a list."""
+    for item in items:
+        yield item
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_handle_insufficient_funds_sends_discord_alert_first_time(
+    server: SpinTestServer,
+):
+    """Test that the first insufficient funds notification sends a Discord alert."""
+
+    execution_processor = ExecutionProcessor()
+    user_id = "test-user-123"
+    graph_id = "test-graph-456"
+    error = InsufficientBalanceError(
+        message="Insufficient balance",
+        user_id=user_id,
+        balance=72,  # $0.72
+        amount=-714,  # Attempting to spend $7.14
+    )
+
+    with patch(
+        "backend.executor.manager.queue_notification"
+    ) as mock_queue_notif, patch(
+        "backend.executor.manager.get_notification_manager_client"
+    ) as mock_get_client, patch(
+        "backend.executor.manager.settings"
+    ) as mock_settings, patch(
+        "backend.executor.manager.redis"
+    ) as mock_redis_module:
+
+        # Setup mocks
+        mock_client = MagicMock()
+        mock_get_client.return_value = mock_client
+        mock_settings.config.frontend_base_url = "https://test.com"
+
+        # Mock Redis to simulate first-time notification (set returns True)
+        mock_redis_client = MagicMock()
+        mock_redis_module.get_redis.return_value = mock_redis_client
+        mock_redis_client.set.return_value = True  # Key was newly set
+
+        # Create mock database client
+        mock_db_client = MagicMock()
+        mock_graph_metadata = MagicMock()
+        mock_graph_metadata.name = "Test Agent"
+        mock_db_client.get_graph_metadata.return_value = mock_graph_metadata
+        mock_db_client.get_user_email_by_id.return_value = "test@example.com"
+
+        # Test the insufficient funds handler
+        execution_processor._handle_insufficient_funds_notif(
+            db_client=mock_db_client,
+            user_id=user_id,
+            graph_id=graph_id,
+            e=error,
+        )
+
+        # Verify notification was queued
+        mock_queue_notif.assert_called_once()
+        notification_call = mock_queue_notif.call_args[0][0]
+        assert notification_call.type == NotificationType.ZERO_BALANCE
+        assert notification_call.user_id == user_id
+        assert isinstance(notification_call.data, ZeroBalanceData)
+        assert notification_call.data.current_balance == 72
+
+        # Verify Redis was checked with correct key pattern
+        expected_key = f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:{graph_id}"
+        mock_redis_client.set.assert_called_once()
+        call_args = mock_redis_client.set.call_args
+        assert call_args[0][0] == expected_key
+        assert call_args[1]["nx"] is True
+
+        # Verify Discord alert was sent
+        mock_client.discord_system_alert.assert_called_once()
+        discord_message = mock_client.discord_system_alert.call_args[0][0]
+        assert "Insufficient Funds Alert" in discord_message
+        assert "test@example.com" in discord_message
+        assert "Test Agent" in discord_message
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_handle_insufficient_funds_skips_duplicate_notifications(
+    server: SpinTestServer,
+):
+    """Test that duplicate insufficient funds notifications skip both email and Discord."""
+
+    execution_processor = ExecutionProcessor()
+    user_id = "test-user-123"
+    graph_id = "test-graph-456"
+    error = InsufficientBalanceError(
+        message="Insufficient balance",
+        user_id=user_id,
+        balance=72,
+        amount=-714,
+    )
+
+    with patch(
+        "backend.executor.manager.queue_notification"
+    ) as mock_queue_notif, patch(
+        "backend.executor.manager.get_notification_manager_client"
+    ) as mock_get_client, patch(
+        "backend.executor.manager.settings"
+    ) as mock_settings, patch(
+        "backend.executor.manager.redis"
+    ) as mock_redis_module:
+
+        # Setup mocks
+        mock_client = MagicMock()
+        mock_get_client.return_value = mock_client
+        mock_settings.config.frontend_base_url = "https://test.com"
+
+        # Mock Redis to simulate duplicate notification (set returns False/None)
+        mock_redis_client = MagicMock()
+        mock_redis_module.get_redis.return_value = mock_redis_client
+        mock_redis_client.set.return_value = None  # Key already existed
+
+        # Create mock database client
+        mock_db_client = MagicMock()
+        mock_db_client.get_graph_metadata.return_value = MagicMock(name="Test Agent")
+
+        # Test the insufficient funds handler
+        execution_processor._handle_insufficient_funds_notif(
+            db_client=mock_db_client,
+            user_id=user_id,
+            graph_id=graph_id,
+            e=error,
+        )
+
+        # Verify email notification was NOT queued (deduplication worked)
+        mock_queue_notif.assert_not_called()
+
+        # Verify Discord alert was NOT sent (deduplication worked)
+        mock_client.discord_system_alert.assert_not_called()
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_handle_insufficient_funds_different_agents_get_separate_alerts(
+    server: SpinTestServer,
+):
+    """Test that different agents for the same user get separate Discord alerts."""
+
+    execution_processor = ExecutionProcessor()
+    user_id = "test-user-123"
+    graph_id_1 = "test-graph-111"
+    graph_id_2 = "test-graph-222"
+
+    error = InsufficientBalanceError(
+        message="Insufficient balance",
+        user_id=user_id,
+        balance=72,
+        amount=-714,
+    )
+
+    with patch("backend.executor.manager.queue_notification"), patch(
+        "backend.executor.manager.get_notification_manager_client"
+    ) as mock_get_client, patch(
+        "backend.executor.manager.settings"
+    ) as mock_settings, patch(
+        "backend.executor.manager.redis"
+    ) as mock_redis_module:
+
+        mock_client = MagicMock()
+        mock_get_client.return_value = mock_client
+        mock_settings.config.frontend_base_url = "https://test.com"
+
+        mock_redis_client = MagicMock()
+        mock_redis_module.get_redis.return_value = mock_redis_client
+        # Both calls return True (first time for each agent)
+        mock_redis_client.set.return_value = True
+
+        mock_db_client = MagicMock()
+        mock_graph_metadata = MagicMock()
+        mock_graph_metadata.name = "Test Agent"
+        mock_db_client.get_graph_metadata.return_value = mock_graph_metadata
+        mock_db_client.get_user_email_by_id.return_value = "test@example.com"
+
+        # First agent notification
+        execution_processor._handle_insufficient_funds_notif(
+            db_client=mock_db_client,
+            user_id=user_id,
+            graph_id=graph_id_1,
+            e=error,
+        )
+
+        # Second agent notification
+        execution_processor._handle_insufficient_funds_notif(
+            db_client=mock_db_client,
+            user_id=user_id,
+            graph_id=graph_id_2,
+            e=error,
+        )
+
+        # Verify Discord alerts were sent for both agents
+        assert mock_client.discord_system_alert.call_count == 2
+
+        # Verify Redis was called with different keys
+        assert mock_redis_client.set.call_count == 2
+        calls = mock_redis_client.set.call_args_list
+        assert (
+            calls[0][0][0]
+            == f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:{graph_id_1}"
+        )
+        assert (
+            calls[1][0][0]
+            == f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:{graph_id_2}"
+        )
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_clear_insufficient_funds_notifications(server: SpinTestServer):
+    """Test that clearing notifications removes all keys for a user."""
+
+    user_id = "test-user-123"
+
+    with patch("backend.executor.manager.redis") as mock_redis_module:
+
+        mock_redis_client = MagicMock()
+        # get_redis_async is an async function, so we need AsyncMock for it
+        mock_redis_module.get_redis_async = AsyncMock(return_value=mock_redis_client)
+
+        # Mock scan_iter to return some keys as an async iterator
+        mock_keys = [
+            f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:graph-1",
+            f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:graph-2",
+            f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:graph-3",
+        ]
+        mock_redis_client.scan_iter.return_value = async_iter(mock_keys)
+        # delete is awaited, so use AsyncMock
+        mock_redis_client.delete = AsyncMock(return_value=3)
+
+        # Clear notifications
+        result = await clear_insufficient_funds_notifications(user_id)
+
+        # Verify correct pattern was used
+        expected_pattern = f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:*"
+        mock_redis_client.scan_iter.assert_called_once_with(match=expected_pattern)
+
+        # Verify delete was called with all keys
+        mock_redis_client.delete.assert_called_once_with(*mock_keys)
+
+        # Verify return value
+        assert result == 3
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_clear_insufficient_funds_notifications_no_keys(server: SpinTestServer):
+    """Test clearing notifications when there are no keys to clear."""
+
+    user_id = "test-user-no-notifications"
+
+    with patch("backend.executor.manager.redis") as mock_redis_module:
+
+        mock_redis_client = MagicMock()
+        # get_redis_async is an async function, so we need AsyncMock for it
+        mock_redis_module.get_redis_async = AsyncMock(return_value=mock_redis_client)
+
+        # Mock scan_iter to return no keys as an async iterator
+        mock_redis_client.scan_iter.return_value = async_iter([])
+
+        # Clear notifications
+        result = await clear_insufficient_funds_notifications(user_id)
+
+        # Verify delete was not called
+        mock_redis_client.delete.assert_not_called()
+
+        # Verify return value
+        assert result == 0
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_clear_insufficient_funds_notifications_handles_redis_error(
+    server: SpinTestServer,
+):
+    """Test that clearing notifications handles Redis errors gracefully."""
+
+    user_id = "test-user-redis-error"
+
+    with patch("backend.executor.manager.redis") as mock_redis_module:
+
+        # Mock get_redis_async to raise an error
+        mock_redis_module.get_redis_async = AsyncMock(
+            side_effect=Exception("Redis connection failed")
+        )
+
+        # Clear notifications should not raise, just return 0
+        result = await clear_insufficient_funds_notifications(user_id)
+
+        # Verify it returned 0 (graceful failure)
+        assert result == 0
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_handle_insufficient_funds_continues_on_redis_error(
+    server: SpinTestServer,
+):
+    """Test that both email and Discord notifications are still sent when Redis fails."""
+
+    execution_processor = ExecutionProcessor()
+    user_id = "test-user-123"
+    graph_id = "test-graph-456"
+    error = InsufficientBalanceError(
+        message="Insufficient balance",
+        user_id=user_id,
+        balance=72,
+        amount=-714,
+    )
+
+    with patch(
+        "backend.executor.manager.queue_notification"
+    ) as mock_queue_notif, patch(
+        "backend.executor.manager.get_notification_manager_client"
+    ) as mock_get_client, patch(
+        "backend.executor.manager.settings"
+    ) as mock_settings, patch(
+        "backend.executor.manager.redis"
+    ) as mock_redis_module:
+
+        mock_client = MagicMock()
+        mock_get_client.return_value = mock_client
+        mock_settings.config.frontend_base_url = "https://test.com"
+
+        # Mock Redis to raise an error
+        mock_redis_client = MagicMock()
+        mock_redis_module.get_redis.return_value = mock_redis_client
+        mock_redis_client.set.side_effect = Exception("Redis connection error")
+
+        mock_db_client = MagicMock()
+        mock_graph_metadata = MagicMock()
+        mock_graph_metadata.name = "Test Agent"
+        mock_db_client.get_graph_metadata.return_value = mock_graph_metadata
+        mock_db_client.get_user_email_by_id.return_value = "test@example.com"
+
+        # Test the insufficient funds handler
+        execution_processor._handle_insufficient_funds_notif(
+            db_client=mock_db_client,
+            user_id=user_id,
+            graph_id=graph_id,
+            e=error,
+        )
+
+        # Verify email notification was still queued despite Redis error
+        mock_queue_notif.assert_called_once()
+
+        # Verify Discord alert was still sent despite Redis error
+        mock_client.discord_system_alert.assert_called_once()
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_add_transaction_clears_notifications_on_grant(server: SpinTestServer):
+    """Test that _add_transaction clears notification flags when adding GRANT credits."""
+    from prisma.enums import CreditTransactionType
+
+    from backend.data.credit import UserCredit
+
+    user_id = "test-user-grant-clear"
+
+    with patch("backend.data.credit.query_raw_with_schema") as mock_query, patch(
+        "backend.executor.manager.redis"
+    ) as mock_redis_module:
+
+        # Mock the query to return a successful transaction
+        mock_query.return_value = [{"balance": 1000, "transactionKey": "test-tx-key"}]
+
+        # Mock async Redis for notification clearing
+        mock_redis_client = MagicMock()
+        mock_redis_module.get_redis_async = AsyncMock(return_value=mock_redis_client)
+        mock_redis_client.scan_iter.return_value = async_iter(
+            [f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:graph-1"]
+        )
+        mock_redis_client.delete = AsyncMock(return_value=1)
+
+        # Create a concrete instance
+        credit_model = UserCredit()
+
+        # Call _add_transaction with GRANT type (should clear notifications)
+        await credit_model._add_transaction(
+            user_id=user_id,
+            amount=500,  # Positive amount
+            transaction_type=CreditTransactionType.GRANT,
+            is_active=True,  # Active transaction
+        )
+
+        # Verify notification clearing was called
+        mock_redis_module.get_redis_async.assert_called_once()
+        mock_redis_client.scan_iter.assert_called_once_with(
+            match=f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:*"
+        )
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_add_transaction_clears_notifications_on_top_up(server: SpinTestServer):
+    """Test that _add_transaction clears notification flags when adding TOP_UP credits."""
+    from prisma.enums import CreditTransactionType
+
+    from backend.data.credit import UserCredit
+
+    user_id = "test-user-topup-clear"
+
+    with patch("backend.data.credit.query_raw_with_schema") as mock_query, patch(
+        "backend.executor.manager.redis"
+    ) as mock_redis_module:
+
+        # Mock the query to return a successful transaction
+        mock_query.return_value = [{"balance": 2000, "transactionKey": "test-tx-key-2"}]
+
+        # Mock async Redis for notification clearing
+        mock_redis_client = MagicMock()
+        mock_redis_module.get_redis_async = AsyncMock(return_value=mock_redis_client)
+        mock_redis_client.scan_iter.return_value = async_iter([])
+        mock_redis_client.delete = AsyncMock(return_value=0)
+
+        credit_model = UserCredit()
+
+        # Call _add_transaction with TOP_UP type (should clear notifications)
+        await credit_model._add_transaction(
+            user_id=user_id,
+            amount=1000,  # Positive amount
+            transaction_type=CreditTransactionType.TOP_UP,
+            is_active=True,
+        )
+
+        # Verify notification clearing was attempted
+        mock_redis_module.get_redis_async.assert_called_once()
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_add_transaction_skips_clearing_for_inactive_transaction(
+    server: SpinTestServer,
+):
+    """Test that _add_transaction does NOT clear notifications for inactive transactions."""
+    from prisma.enums import CreditTransactionType
+
+    from backend.data.credit import UserCredit
+
+    user_id = "test-user-inactive"
+
+    with patch("backend.data.credit.query_raw_with_schema") as mock_query, patch(
+        "backend.executor.manager.redis"
+    ) as mock_redis_module:
+
+        # Mock the query to return a successful transaction
+        mock_query.return_value = [{"balance": 500, "transactionKey": "test-tx-key-3"}]
+
+        # Mock async Redis
+        mock_redis_client = MagicMock()
+        mock_redis_module.get_redis_async = AsyncMock(return_value=mock_redis_client)
+
+        credit_model = UserCredit()
+
+        # Call _add_transaction with is_active=False (should NOT clear notifications)
+        await credit_model._add_transaction(
+            user_id=user_id,
+            amount=500,
+            transaction_type=CreditTransactionType.TOP_UP,
+            is_active=False,  # Inactive - pending Stripe payment
+        )
+
+        # Verify notification clearing was NOT called
+        mock_redis_module.get_redis_async.assert_not_called()
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_add_transaction_skips_clearing_for_usage_transaction(
+    server: SpinTestServer,
+):
+    """Test that _add_transaction does NOT clear notifications for USAGE transactions."""
+    from prisma.enums import CreditTransactionType
+
+    from backend.data.credit import UserCredit
+
+    user_id = "test-user-usage"
+
+    with patch("backend.data.credit.query_raw_with_schema") as mock_query, patch(
+        "backend.executor.manager.redis"
+    ) as mock_redis_module:
+
+        # Mock the query to return a successful transaction
+        mock_query.return_value = [{"balance": 400, "transactionKey": "test-tx-key-4"}]
+
+        # Mock async Redis
+        mock_redis_client = MagicMock()
+        mock_redis_module.get_redis_async = AsyncMock(return_value=mock_redis_client)
+
+        credit_model = UserCredit()
+
+        # Call _add_transaction with USAGE type (spending, should NOT clear)
+        await credit_model._add_transaction(
+            user_id=user_id,
+            amount=-100,  # Negative - spending credits
+            transaction_type=CreditTransactionType.USAGE,
+            is_active=True,
+        )
+
+        # Verify notification clearing was NOT called
+        mock_redis_module.get_redis_async.assert_not_called()
+
+
+@pytest.mark.asyncio(loop_scope="session")
+async def test_enable_transaction_clears_notifications(server: SpinTestServer):
+    """Test that _enable_transaction clears notification flags when enabling a TOP_UP."""
+    from prisma.enums import CreditTransactionType
+
+    from backend.data.credit import UserCredit
+
+    user_id = "test-user-enable"
+
+    with patch("backend.data.credit.CreditTransaction") as mock_credit_tx, patch(
+        "backend.data.credit.query_raw_with_schema"
+    ) as mock_query, patch("backend.executor.manager.redis") as mock_redis_module:
+
+        # Mock finding the pending transaction
+        mock_transaction = MagicMock()
+        mock_transaction.amount = 1000
+        mock_transaction.type = CreditTransactionType.TOP_UP
+        mock_credit_tx.prisma.return_value.find_first = AsyncMock(
+            return_value=mock_transaction
+        )
+
+        # Mock the query to return updated balance
+        mock_query.return_value = [{"balance": 1500}]
+
+        # Mock async Redis for notification clearing
+        mock_redis_client = MagicMock()
+        mock_redis_module.get_redis_async = AsyncMock(return_value=mock_redis_client)
+        mock_redis_client.scan_iter.return_value = async_iter(
+            [f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:graph-1"]
+        )
+        mock_redis_client.delete = AsyncMock(return_value=1)
+
+        credit_model = UserCredit()
+
+        # Call _enable_transaction (simulates Stripe checkout completion)
+        from backend.util.json import SafeJson
+
+        await credit_model._enable_transaction(
+            transaction_key="cs_test_123",
+            user_id=user_id,
+            metadata=SafeJson({"payment": "completed"}),
+        )
+
+        # Verify notification clearing was called
+        mock_redis_module.get_redis_async.assert_called_once()
+        mock_redis_client.scan_iter.assert_called_once_with(
+            match=f"{INSUFFICIENT_FUNDS_NOTIFIED_PREFIX}:{user_id}:*"
+        )

autogpt_platform/backend/backend/util/settings.py

@@ -658,6 +658,14 @@ class Secrets(UpdateTrackingModel["Secrets"], BaseSettings):
 
     ayrshare_api_key: str = Field(default="", description="Ayrshare API Key")
     ayrshare_jwt_key: str = Field(default="", description="Ayrshare private Key")
+
+    # Langfuse prompt management
+    langfuse_public_key: str = Field(default="", description="Langfuse public key")
+    langfuse_secret_key: str = Field(default="", description="Langfuse secret key")
+    langfuse_host: str = Field(
+        default="https://cloud.langfuse.com", description="Langfuse host URL"
+    )
+
     # Add more secret fields as needed
     model_config = SettingsConfigDict(
         env_file=".env",

autogpt_platform/backend/migrations/20251216153446_add_chat_tables/migration.sql

@@ -0,0 +1,78 @@
+-- CreateTable
+CREATE TABLE "UserBusinessUnderstanding" (
+    "id" TEXT NOT NULL,
+    "createdAt" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    "updatedAt" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    "userId" TEXT NOT NULL,
+    "userName" TEXT,
+    "jobTitle" TEXT,
+    "businessName" TEXT,
+    "industry" TEXT,
+    "businessSize" TEXT,
+    "userRole" TEXT,
+    "keyWorkflows" JSONB,
+    "dailyActivities" JSONB,
+    "painPoints" JSONB,
+    "bottlenecks" JSONB,
+    "manualTasks" JSONB,
+    "automationGoals" JSONB,
+    "currentSoftware" JSONB,
+    "existingAutomation" JSONB,
+    "additionalNotes" TEXT,
+
+    CONSTRAINT "UserBusinessUnderstanding_pkey" PRIMARY KEY ("id")
+);
+
+-- CreateTable
+CREATE TABLE "ChatSession" (
+    "id" TEXT NOT NULL,
+    "createdAt" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    "updatedAt" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    "userId" TEXT,
+    "title" TEXT,
+    "credentials" JSONB NOT NULL DEFAULT '{}',
+    "successfulAgentRuns" JSONB NOT NULL DEFAULT '{}',
+    "successfulAgentSchedules" JSONB NOT NULL DEFAULT '{}',
+    "totalPromptTokens" INTEGER NOT NULL DEFAULT 0,
+    "totalCompletionTokens" INTEGER NOT NULL DEFAULT 0,
+
+    CONSTRAINT "ChatSession_pkey" PRIMARY KEY ("id")
+);
+
+-- CreateTable
+CREATE TABLE "ChatMessage" (
+    "id" TEXT NOT NULL,
+    "createdAt" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    "sessionId" TEXT NOT NULL,
+    "role" TEXT NOT NULL,
+    "content" TEXT,
+    "name" TEXT,
+    "toolCallId" TEXT,
+    "refusal" TEXT,
+    "toolCalls" JSONB,
+    "functionCall" JSONB,
+    "sequence" INTEGER NOT NULL,
+
+    CONSTRAINT "ChatMessage_pkey" PRIMARY KEY ("id")
+);
+
+-- CreateIndex
+CREATE UNIQUE INDEX "UserBusinessUnderstanding_userId_key" ON "UserBusinessUnderstanding"("userId");
+
+-- CreateIndex
+CREATE INDEX "UserBusinessUnderstanding_userId_idx" ON "UserBusinessUnderstanding"("userId");
+
+-- CreateIndex
+CREATE INDEX "ChatSession_userId_updatedAt_idx" ON "ChatSession"("userId", "updatedAt");
+
+-- CreateIndex
+CREATE INDEX "ChatMessage_sessionId_sequence_idx" ON "ChatMessage"("sessionId", "sequence");
+
+-- CreateIndex
+CREATE UNIQUE INDEX "ChatMessage_sessionId_sequence_key" ON "ChatMessage"("sessionId", "sequence");
+
+-- AddForeignKey
+ALTER TABLE "UserBusinessUnderstanding" ADD CONSTRAINT "UserBusinessUnderstanding_userId_fkey" FOREIGN KEY ("userId") REFERENCES "User"("id") ON DELETE CASCADE ON UPDATE CASCADE;
+
+-- AddForeignKey
+ALTER TABLE "ChatMessage" ADD CONSTRAINT "ChatMessage_sessionId_fkey" FOREIGN KEY ("sessionId") REFERENCES "ChatSession"("id") ON DELETE CASCADE ON UPDATE CASCADE;

autogpt_platform/backend/migrations/20251216181700_add_store_embeddings/migration.sql

@@ -0,0 +1,41 @@
+-- Migration: Add pgvector extension and StoreListingEmbedding table
+-- This enables hybrid search combining semantic (embedding) and lexical (tsvector) search
+
+-- Enable pgvector extension for vector similarity search
+CREATE EXTENSION IF NOT EXISTS vector;
+
+-- Create table to store embeddings for store listing versions
+CREATE TABLE "StoreListingEmbedding" (
+    "id" TEXT NOT NULL DEFAULT gen_random_uuid(),
+    "storeListingVersionId" TEXT NOT NULL,
+    "embedding" vector(1536),  -- OpenAI text-embedding-3-small produces 1536 dimensions
+    "searchableText" TEXT,     -- The text that was embedded (for debugging/recomputation)
+    "contentHash" TEXT,        -- MD5 hash of searchable text for change detection
+    "createdAt" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    "updatedAt" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
+
+    CONSTRAINT "StoreListingEmbedding_pkey" PRIMARY KEY ("id")
+);
+
+-- Unique constraint: one embedding per listing version
+CREATE UNIQUE INDEX "StoreListingEmbedding_storeListingVersionId_key"
+    ON "StoreListingEmbedding"("storeListingVersionId");
+
+-- HNSW index for fast approximate nearest neighbor search
+-- Using cosine distance (vector_cosine_ops) which is standard for text embeddings
+CREATE INDEX "StoreListingEmbedding_embedding_idx"
+    ON "StoreListingEmbedding"
+    USING hnsw ("embedding" vector_cosine_ops);
+
+-- Index on content hash for fast lookup during change detection
+CREATE INDEX "StoreListingEmbedding_contentHash_idx"
+    ON "StoreListingEmbedding"("contentHash");
+
+-- Foreign key to StoreListingVersion with CASCADE delete
+-- When a listing version is deleted, its embedding is automatically removed
+ALTER TABLE "StoreListingEmbedding"
+    ADD CONSTRAINT "StoreListingEmbedding_storeListingVersionId_fkey"
+    FOREIGN KEY ("storeListingVersionId")
+    REFERENCES "StoreListingVersion"("id")
+    ON DELETE CASCADE
+    ON UPDATE CASCADE;

autogpt_platform/backend/migrations/20251216181803_enhance_search/migration.sql

@@ -0,0 +1,5 @@
+-- DropIndex
+DROP INDEX "StoreListingEmbedding_embedding_idx";
+
+-- AlterTable
+ALTER TABLE "StoreListingEmbedding" ALTER COLUMN "id" DROP DEFAULT;

autogpt_platform/backend/poetry.lock

@@ -1906,16 +1906,32 @@ httpx = {version = ">=0.26,<0.29", extras = ["http2"]}
 pydantic = ">=1.10,<3"
 pyjwt = ">=2.10.1,<3.0.0"
 
+[[package]]
+name = "gravitas-md2gdocs"
+version = "0.1.0"
+description = "Convert Markdown to Google Docs API requests"
+optional = false
+python-versions = ">=3.10"
+groups = ["main"]
+files = [
+    {file = "gravitas_md2gdocs-0.1.0-py3-none-any.whl", hash = "sha256:0cb0627779fdd65c1604818af4142eea1b25d055060183363de1bae4d9e46508"},
+    {file = "gravitas_md2gdocs-0.1.0.tar.gz", hash = "sha256:bb3122fe9fa35c528f3f00b785d3f1398d350082d5d03f60f56c895bdcc68033"},
+]
+
+[package.extras]
+dev = ["google-auth-oauthlib (>=1.0.0)", "pytest (>=7.0.0)", "pytest-cov (>=4.0.0)", "python-dotenv (>=1.0.0)", "ruff (>=0.1.0)"]
+google = ["google-api-python-client (>=2.0.0)", "google-auth (>=2.0.0)"]
+
 [[package]]
 name = "gravitasml"
-version = "0.1.3"
+version = "0.1.4"
 description = ""
 optional = false
 python-versions = "<4.0,>=3.10"
 groups = ["main"]
 files = [
-    {file = "gravitasml-0.1.3-py3-none-any.whl", hash = "sha256:51ff98b4564b7a61f7796f18d5f2558b919d30b3722579296089645b7bc18b85"},
-    {file = "gravitasml-0.1.3.tar.gz", hash = "sha256:04d240b9fa35878252d57a36032130b6516487468847fcdced1022c032a20f57"},
+    {file = "gravitasml-0.1.4-py3-none-any.whl", hash = "sha256:671a18b11d3d8a0e270c6a80c72cd058458b18d5ef7560d00010e962ab1bca74"},
+    {file = "gravitasml-0.1.4.tar.gz", hash = "sha256:35d0d9fec7431817482d53d9c976e375557c3e041d1eb6928e809324a8c866e3"},
 ]
 
 [package.dependencies]
@@ -2761,6 +2777,33 @@ enabler = ["pytest-enabler (>=2.2)"]
 test = ["pyfakefs", "pytest (>=6,!=8.1.*)"]
 type = ["pygobject-stubs", "pytest-mypy", "shtab", "types-pywin32"]
 
+[[package]]
+name = "langfuse"
+version = "2.60.10"
+description = "A client library for accessing langfuse"
+optional = false
+python-versions = "<4.0,>=3.9"
+groups = ["main"]
+files = [
+    {file = "langfuse-2.60.10-py3-none-any.whl", hash = "sha256:815c6369194aa5b2a24f88eb9952f7c3fc863272c41e90642a71f3bc76f4a11f"},
+    {file = "langfuse-2.60.10.tar.gz", hash = "sha256:a26d0d927a28ee01b2d12bb5b862590b643cc4e60a28de6e2b0c2cfff5dbfc6a"},
+]
+
+[package.dependencies]
+anyio = ">=4.4.0,<5.0.0"
+backoff = ">=1.10.0"
+httpx = ">=0.15.4,<1.0"
+idna = ">=3.7,<4.0"
+packaging = ">=23.2,<25.0"
+pydantic = ">=1.10.7,<3.0"
+requests = ">=2,<3"
+wrapt = ">=1.14,<2.0"
+
+[package.extras]
+langchain = ["langchain (>=0.0.309)"]
+llama-index = ["llama-index (>=0.10.12,<2.0.0)"]
+openai = ["openai (>=0.27.8)"]
+
 [[package]]
 name = "launchdarkly-eventsource"
 version = "1.3.0"
@@ -6906,6 +6949,97 @@ files = [
     {file = "websockets-15.0.1.tar.gz", hash = "sha256:82544de02076bafba038ce055ee6412d68da13ab47f0c60cab827346de828dee"},
 ]
 
+[[package]]
+name = "wrapt"
+version = "1.17.3"
+description = "Module for decorators, wrappers and monkey patching."
+optional = false
+python-versions = ">=3.8"
+groups = ["main"]
+files = [
+    {file = "wrapt-1.17.3-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:88bbae4d40d5a46142e70d58bf664a89b6b4befaea7b2ecc14e03cedb8e06c04"},
+    {file = "wrapt-1.17.3-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:e6b13af258d6a9ad602d57d889f83b9d5543acd471eee12eb51f5b01f8eb1bc2"},
+    {file = "wrapt-1.17.3-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:fd341868a4b6714a5962c1af0bd44f7c404ef78720c7de4892901e540417111c"},
+    {file = "wrapt-1.17.3-cp310-cp310-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:f9b2601381be482f70e5d1051a5965c25fb3625455a2bf520b5a077b22afb775"},
+    {file = "wrapt-1.17.3-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:343e44b2a8e60e06a7e0d29c1671a0d9951f59174f3709962b5143f60a2a98bd"},
+    {file = "wrapt-1.17.3-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:33486899acd2d7d3066156b03465b949da3fd41a5da6e394ec49d271baefcf05"},
+    {file = "wrapt-1.17.3-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:e6f40a8aa5a92f150bdb3e1c44b7e98fb7113955b2e5394122fa5532fec4b418"},
+    {file = "wrapt-1.17.3-cp310-cp310-win32.whl", hash = "sha256:a36692b8491d30a8c75f1dfee65bef119d6f39ea84ee04d9f9311f83c5ad9390"},
+    {file = "wrapt-1.17.3-cp310-cp310-win_amd64.whl", hash = "sha256:afd964fd43b10c12213574db492cb8f73b2f0826c8df07a68288f8f19af2ebe6"},
+    {file = "wrapt-1.17.3-cp310-cp310-win_arm64.whl", hash = "sha256:af338aa93554be859173c39c85243970dc6a289fa907402289eeae7543e1ae18"},
+    {file = "wrapt-1.17.3-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:273a736c4645e63ac582c60a56b0acb529ef07f78e08dc6bfadf6a46b19c0da7"},
+    {file = "wrapt-1.17.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:5531d911795e3f935a9c23eb1c8c03c211661a5060aab167065896bbf62a5f85"},
+    {file = "wrapt-1.17.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:0610b46293c59a3adbae3dee552b648b984176f8562ee0dba099a56cfbe4df1f"},
+    {file = "wrapt-1.17.3-cp311-cp311-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:b32888aad8b6e68f83a8fdccbf3165f5469702a7544472bdf41f582970ed3311"},
+    {file = "wrapt-1.17.3-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:8cccf4f81371f257440c88faed6b74f1053eef90807b77e31ca057b2db74edb1"},
+    {file = "wrapt-1.17.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:d8a210b158a34164de8bb68b0e7780041a903d7b00c87e906fb69928bf7890d5"},
+    {file = "wrapt-1.17.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:79573c24a46ce11aab457b472efd8d125e5a51da2d1d24387666cd85f54c05b2"},
+    {file = "wrapt-1.17.3-cp311-cp311-win32.whl", hash = "sha256:c31eebe420a9a5d2887b13000b043ff6ca27c452a9a22fa71f35f118e8d4bf89"},
+    {file = "wrapt-1.17.3-cp311-cp311-win_amd64.whl", hash = "sha256:0b1831115c97f0663cb77aa27d381237e73ad4f721391a9bfb2fe8bc25fa6e77"},
+    {file = "wrapt-1.17.3-cp311-cp311-win_arm64.whl", hash = "sha256:5a7b3c1ee8265eb4c8f1b7d29943f195c00673f5ab60c192eba2d4a7eae5f46a"},
+    {file = "wrapt-1.17.3-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:ab232e7fdb44cdfbf55fc3afa31bcdb0d8980b9b95c38b6405df2acb672af0e0"},
+    {file = "wrapt-1.17.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:9baa544e6acc91130e926e8c802a17f3b16fbea0fd441b5a60f5cf2cc5c3deba"},
+    {file = "wrapt-1.17.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:6b538e31eca1a7ea4605e44f81a48aa24c4632a277431a6ed3f328835901f4fd"},
+    {file = "wrapt-1.17.3-cp312-cp312-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:042ec3bb8f319c147b1301f2393bc19dba6e176b7da446853406d041c36c7828"},
+    {file = "wrapt-1.17.3-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:3af60380ba0b7b5aeb329bc4e402acd25bd877e98b3727b0135cb5c2efdaefe9"},
+    {file = "wrapt-1.17.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:0b02e424deef65c9f7326d8c19220a2c9040c51dc165cddb732f16198c168396"},
+    {file = "wrapt-1.17.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:74afa28374a3c3a11b3b5e5fca0ae03bef8450d6aa3ab3a1e2c30e3a75d023dc"},
+    {file = "wrapt-1.17.3-cp312-cp312-win32.whl", hash = "sha256:4da9f45279fff3543c371d5ababc57a0384f70be244de7759c85a7f989cb4ebe"},
+    {file = "wrapt-1.17.3-cp312-cp312-win_amd64.whl", hash = "sha256:e71d5c6ebac14875668a1e90baf2ea0ef5b7ac7918355850c0908ae82bcb297c"},
+    {file = "wrapt-1.17.3-cp312-cp312-win_arm64.whl", hash = "sha256:604d076c55e2fdd4c1c03d06dc1a31b95130010517b5019db15365ec4a405fc6"},
+    {file = "wrapt-1.17.3-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:a47681378a0439215912ef542c45a783484d4dd82bac412b71e59cf9c0e1cea0"},
+    {file = "wrapt-1.17.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:54a30837587c6ee3cd1a4d1c2ec5d24e77984d44e2f34547e2323ddb4e22eb77"},
+    {file = "wrapt-1.17.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:16ecf15d6af39246fe33e507105d67e4b81d8f8d2c6598ff7e3ca1b8a37213f7"},
+    {file = "wrapt-1.17.3-cp313-cp313-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:6fd1ad24dc235e4ab88cda009e19bf347aabb975e44fd5c2fb22a3f6e4141277"},
+    {file = "wrapt-1.17.3-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:0ed61b7c2d49cee3c027372df5809a59d60cf1b6c2f81ee980a091f3afed6a2d"},
+    {file = "wrapt-1.17.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:423ed5420ad5f5529db9ce89eac09c8a2f97da18eb1c870237e84c5a5c2d60aa"},
+    {file = "wrapt-1.17.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:e01375f275f010fcbf7f643b4279896d04e571889b8a5b3f848423d91bf07050"},
+    {file = "wrapt-1.17.3-cp313-cp313-win32.whl", hash = "sha256:53e5e39ff71b3fc484df8a522c933ea2b7cdd0d5d15ae82e5b23fde87d44cbd8"},
+    {file = "wrapt-1.17.3-cp313-cp313-win_amd64.whl", hash = "sha256:1f0b2f40cf341ee8cc1a97d51ff50dddb9fcc73241b9143ec74b30fc4f44f6cb"},
+    {file = "wrapt-1.17.3-cp313-cp313-win_arm64.whl", hash = "sha256:7425ac3c54430f5fc5e7b6f41d41e704db073309acfc09305816bc6a0b26bb16"},
+    {file = "wrapt-1.17.3-cp314-cp314-macosx_10_13_universal2.whl", hash = "sha256:cf30f6e3c077c8e6a9a7809c94551203c8843e74ba0c960f4a98cd80d4665d39"},
+    {file = "wrapt-1.17.3-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:e228514a06843cae89621384cfe3a80418f3c04aadf8a3b14e46a7be704e4235"},
+    {file = "wrapt-1.17.3-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:5ea5eb3c0c071862997d6f3e02af1d055f381b1d25b286b9d6644b79db77657c"},
+    {file = "wrapt-1.17.3-cp314-cp314-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:281262213373b6d5e4bb4353bc36d1ba4084e6d6b5d242863721ef2bf2c2930b"},
+    {file = "wrapt-1.17.3-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:dc4a8d2b25efb6681ecacad42fca8859f88092d8732b170de6a5dddd80a1c8fa"},
+    {file = "wrapt-1.17.3-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:373342dd05b1d07d752cecbec0c41817231f29f3a89aa8b8843f7b95992ed0c7"},
+    {file = "wrapt-1.17.3-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:d40770d7c0fd5cbed9d84b2c3f2e156431a12c9a37dc6284060fb4bec0b7ffd4"},
+    {file = "wrapt-1.17.3-cp314-cp314-win32.whl", hash = "sha256:fbd3c8319de8e1dc79d346929cd71d523622da527cca14e0c1d257e31c2b8b10"},
+    {file = "wrapt-1.17.3-cp314-cp314-win_amd64.whl", hash = "sha256:e1a4120ae5705f673727d3253de3ed0e016f7cd78dc463db1b31e2463e1f3cf6"},
+    {file = "wrapt-1.17.3-cp314-cp314-win_arm64.whl", hash = "sha256:507553480670cab08a800b9463bdb881b2edeed77dc677b0a5915e6106e91a58"},
+    {file = "wrapt-1.17.3-cp314-cp314t-macosx_10_13_universal2.whl", hash = "sha256:ed7c635ae45cfbc1a7371f708727bf74690daedc49b4dba310590ca0bd28aa8a"},
+    {file = "wrapt-1.17.3-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:249f88ed15503f6492a71f01442abddd73856a0032ae860de6d75ca62eed8067"},
+    {file = "wrapt-1.17.3-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:5a03a38adec8066d5a37bea22f2ba6bbf39fcdefbe2d91419ab864c3fb515454"},
+    {file = "wrapt-1.17.3-cp314-cp314t-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:5d4478d72eb61c36e5b446e375bbc49ed002430d17cdec3cecb36993398e1a9e"},
+    {file = "wrapt-1.17.3-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:223db574bb38637e8230eb14b185565023ab624474df94d2af18f1cdb625216f"},
+    {file = "wrapt-1.17.3-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:e405adefb53a435f01efa7ccdec012c016b5a1d3f35459990afc39b6be4d5056"},
+    {file = "wrapt-1.17.3-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:88547535b787a6c9ce4086917b6e1d291aa8ed914fdd3a838b3539dc95c12804"},
+    {file = "wrapt-1.17.3-cp314-cp314t-win32.whl", hash = "sha256:41b1d2bc74c2cac6f9074df52b2efbef2b30bdfe5f40cb78f8ca22963bc62977"},
+    {file = "wrapt-1.17.3-cp314-cp314t-win_amd64.whl", hash = "sha256:73d496de46cd2cdbdbcce4ae4bcdb4afb6a11234a1df9c085249d55166b95116"},
+    {file = "wrapt-1.17.3-cp314-cp314t-win_arm64.whl", hash = "sha256:f38e60678850c42461d4202739f9bf1e3a737c7ad283638251e79cc49effb6b6"},
+    {file = "wrapt-1.17.3-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:70d86fa5197b8947a2fa70260b48e400bf2ccacdcab97bb7de47e3d1e6312225"},
+    {file = "wrapt-1.17.3-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:df7d30371a2accfe4013e90445f6388c570f103d61019b6b7c57e0265250072a"},
+    {file = "wrapt-1.17.3-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:caea3e9c79d5f0d2c6d9ab96111601797ea5da8e6d0723f77eabb0d4068d2b2f"},
+    {file = "wrapt-1.17.3-cp38-cp38-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:758895b01d546812d1f42204bd443b8c433c44d090248bf22689df673ccafe00"},
+    {file = "wrapt-1.17.3-cp38-cp38-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:02b551d101f31694fc785e58e0720ef7d9a10c4e62c1c9358ce6f63f23e30a56"},
+    {file = "wrapt-1.17.3-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:656873859b3b50eeebe6db8b1455e99d90c26ab058db8e427046dbc35c3140a5"},
+    {file = "wrapt-1.17.3-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:a9a2203361a6e6404f80b99234fe7fb37d1fc73487b5a78dc1aa5b97201e0f22"},
+    {file = "wrapt-1.17.3-cp38-cp38-win32.whl", hash = "sha256:55cbbc356c2842f39bcc553cf695932e8b30e30e797f961860afb308e6b1bb7c"},
+    {file = "wrapt-1.17.3-cp38-cp38-win_amd64.whl", hash = "sha256:ad85e269fe54d506b240d2d7b9f5f2057c2aa9a2ea5b32c66f8902f768117ed2"},
+    {file = "wrapt-1.17.3-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:30ce38e66630599e1193798285706903110d4f057aab3168a34b7fdc85569afc"},
+    {file = "wrapt-1.17.3-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:65d1d00fbfb3ea5f20add88bbc0f815150dbbde3b026e6c24759466c8b5a9ef9"},
+    {file = "wrapt-1.17.3-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:a7c06742645f914f26c7f1fa47b8bc4c91d222f76ee20116c43d5ef0912bba2d"},
+    {file = "wrapt-1.17.3-cp39-cp39-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:7e18f01b0c3e4a07fe6dfdb00e29049ba17eadbc5e7609a2a3a4af83ab7d710a"},
+    {file = "wrapt-1.17.3-cp39-cp39-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:0f5f51a6466667a5a356e6381d362d259125b57f059103dd9fdc8c0cf1d14139"},
+    {file = "wrapt-1.17.3-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:59923aa12d0157f6b82d686c3fd8e1166fa8cdfb3e17b42ce3b6147ff81528df"},
+    {file = "wrapt-1.17.3-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:46acc57b331e0b3bcb3e1ca3b421d65637915cfcd65eb783cb2f78a511193f9b"},
+    {file = "wrapt-1.17.3-cp39-cp39-win32.whl", hash = "sha256:3e62d15d3cfa26e3d0788094de7b64efa75f3a53875cdbccdf78547aed547a81"},
+    {file = "wrapt-1.17.3-cp39-cp39-win_amd64.whl", hash = "sha256:1f23fa283f51c890eda8e34e4937079114c74b4c81d2b2f1f1d94948f5cc3d7f"},
+    {file = "wrapt-1.17.3-cp39-cp39-win_arm64.whl", hash = "sha256:24c2ed34dc222ed754247a2702b1e1e89fdbaa4016f324b4b8f1a802d4ffe87f"},
+    {file = "wrapt-1.17.3-py3-none-any.whl", hash = "sha256:7171ae35d2c33d326ac19dd8facb1e82e5fd04ef8c6c0e394d7af55a55051c22"},
+    {file = "wrapt-1.17.3.tar.gz", hash = "sha256:f66eb08feaa410fe4eebd17f2a2c8e2e46d3476e9f8c783daa8e09e0faa666d0"},
+]
+
 [[package]]
 name = "xattr"
 version = "1.2.0"
@@ -7279,4 +7413,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.1"
 python-versions = ">=3.10,<3.14"
-content-hash = "13b191b2a1989d3321ff713c66ff6f5f4f3b82d15df4d407e0e5dbf87d7522c4"
+content-hash = "e5a35435ea318a13aa53df4124a4151bc8a3dd7dfc8f9738e3da24d7906d555e"

autogpt_platform/backend/pyproject.toml

@@ -27,12 +27,13 @@ google-api-python-client = "^2.177.0"
 google-auth-oauthlib = "^1.2.2"
 google-cloud-storage = "^3.2.0"
 googlemaps = "^4.10.0"
-gravitasml = "^0.1.3"
+gravitasml = "^0.1.4"
 groq = "^0.30.0"
 html2text = "^2024.2.26"
 jinja2 = "^3.1.6"
 jsonref = "^1.1.0"
 jsonschema = "^4.25.0"
+langfuse = "^2.0.0"
 launchdarkly-server-sdk = "^9.12.0"
 mem0ai = "^0.1.115"
 moviepy = "^2.1.2"
@@ -82,6 +83,7 @@ firecrawl-py = "^4.3.6"
 exa-py = "^1.14.20"
 croniter = "^6.0.0"
 stagehand = "^0.5.1"
+gravitas-md2gdocs = "^0.1.0"
 
 [tool.poetry.group.dev.dependencies]
 aiohappyeyeballs = "^2.6.1"

autogpt_platform/backend/schema.prisma

@@ -1,14 +1,15 @@
 datasource db {
-  provider  = "postgresql"
-  url       = env("DATABASE_URL")
-  directUrl = env("DIRECT_URL")
+  provider   = "postgresql"
+  url        = env("DATABASE_URL")
+  directUrl  = env("DIRECT_URL")
+  extensions = [pgvector(map: "vector", schema: "public")]
 }
 
 generator client {
   provider               = "prisma-client-py"
   recursive_type_depth   = -1
   interface              = "asyncio"
-  previewFeatures        = ["views", "fullTextSearch"]
+  previewFeatures        = ["views", "fullTextSearch", "postgresqlExtensions"]
   partial_type_generator = "backend/data/partial_types.py"
 }
 
@@ -53,6 +54,7 @@ model User {
 
   Profile               Profile[]
   UserOnboarding        UserOnboarding?
+  BusinessUnderstanding UserBusinessUnderstanding?
   BuilderSearchHistory  BuilderSearchHistory[]
   StoreListings         StoreListing[]
   StoreListingReviews   StoreListingReview[]
@@ -121,19 +123,109 @@ model UserOnboarding {
   User   User   @relation(fields: [userId], references: [id], onDelete: Cascade)
 }
 
+model UserBusinessUnderstanding {
+  id        String   @id @default(uuid())
+  createdAt DateTime @default(now())
+  updatedAt DateTime @default(now()) @updatedAt
+
+  userId String @unique
+  User   User   @relation(fields: [userId], references: [id], onDelete: Cascade)
+
+  // User info
+  userName String?
+  jobTitle String?
+
+  // Business basics (string columns)
+  businessName String?
+  industry     String?
+  businessSize String? // "1-10", "11-50", "51-200", "201-1000", "1000+"
+  userRole     String? // Role in organization context (e.g., "decision maker", "implementer")
+
+  // Processes & activities (JSON arrays)
+  keyWorkflows    Json?
+  dailyActivities Json?
+
+  // Pain points & goals (JSON arrays)
+  painPoints      Json?
+  bottlenecks     Json?
+  manualTasks     Json?
+  automationGoals Json?
+
+  // Current tools (JSON arrays)
+  currentSoftware    Json?
+  existingAutomation Json?
+
+  additionalNotes String?
+
+  @@index([userId])
+}
+
 model BuilderSearchHistory {
   id        String   @id @default(uuid())
   createdAt DateTime @default(now())
   updatedAt DateTime @default(now()) @updatedAt
 
   searchQuery String
-  filter      String[]  @default([])
-  byCreator   String[]  @default([])
+  filter      String[] @default([])
+  byCreator   String[] @default([])
 
   userId String
   User   User   @relation(fields: [userId], references: [id], onDelete: Cascade)
 }
 
+////////////////////////////////////////////////////////////
+////////////////////////////////////////////////////////////
+////////////////   CHAT SESSION TABLES   ///////////////////
+////////////////////////////////////////////////////////////
+////////////////////////////////////////////////////////////
+
+model ChatSession {
+  id        String   @id @default(uuid())
+  createdAt DateTime @default(now())
+  updatedAt DateTime @default(now()) @updatedAt
+
+  userId String?
+
+  // Session metadata
+  title       String?
+  credentials Json    @default("{}") // Map of provider -> credential metadata
+
+  // Rate limiting counters (stored as JSON maps)
+  successfulAgentRuns      Json @default("{}") // Map of graph_id -> count
+  successfulAgentSchedules Json @default("{}") // Map of graph_id -> count
+
+  // Usage tracking
+  totalPromptTokens     Int @default(0)
+  totalCompletionTokens Int @default(0)
+
+  Messages ChatMessage[]
+
+  @@index([userId, updatedAt])
+}
+
+model ChatMessage {
+  id        String   @id @default(uuid())
+  createdAt DateTime @default(now())
+
+  sessionId String
+  Session   ChatSession @relation(fields: [sessionId], references: [id], onDelete: Cascade)
+
+  // Message content
+  role         String // "user", "assistant", "system", "tool", "function"
+  content      String?
+  name         String?
+  toolCallId   String?
+  refusal      String?
+  toolCalls    Json? // List of tool calls for assistant messages
+  functionCall Json? // Deprecated but kept for compatibility
+
+  // Ordering within session
+  sequence Int
+
+  @@unique([sessionId, sequence])
+  @@index([sessionId, sequence])
+}
+
 // This model describes the Agent Graph/Flow (Multi Agent System).
 model AgentGraph {
   id        String    @default(uuid())
@@ -721,26 +813,26 @@ view StoreAgent {
   storeListingVersionId String
   updated_at            DateTime
 
-  slug                String
-  agent_name          String
-  agent_video         String?
-  agent_output_demo   String?
-  agent_image         String[]
+  slug              String
+  agent_name        String
+  agent_video       String?
+  agent_output_demo String?
+  agent_image       String[]
 
-  featured         Boolean                  @default(false)
-  creator_username String?
-  creator_avatar   String?
-  sub_heading      String
-  description      String
-  categories       String[]
-  search           Unsupported("tsvector")? @default(dbgenerated("''::tsvector"))
-  runs                Int
-  rating              Float
-  versions            String[]
-  agentGraphVersions  String[]
-  agentGraphId        String
-  is_available        Boolean                  @default(true)
-  useForOnboarding    Boolean                  @default(false)
+  featured           Boolean                  @default(false)
+  creator_username   String?
+  creator_avatar     String?
+  sub_heading        String
+  description        String
+  categories         String[]
+  search             Unsupported("tsvector")? @default(dbgenerated("''::tsvector"))
+  runs               Int
+  rating             Float
+  versions           String[]
+  agentGraphVersions String[]
+  agentGraphId       String
+  is_available       Boolean                  @default(true)
+  useForOnboarding   Boolean                  @default(false)
 
   // Materialized views used (refreshed every 15 minutes via pg_cron):
   // - mv_agent_run_counts - Pre-aggregated agent execution counts by agentGraphId
@@ -856,14 +948,14 @@ model StoreListingVersion {
   AgentGraph        AgentGraph @relation(fields: [agentGraphId, agentGraphVersion], references: [id, version])
 
   // Content fields
-  name                String
-  subHeading          String
-  videoUrl            String?
-  agentOutputDemoUrl  String?
-  imageUrls           String[]
-  description         String
-  instructions        String?
-  categories          String[]
+  name               String
+  subHeading         String
+  videoUrl           String?
+  agentOutputDemoUrl String?
+  imageUrls          String[]
+  description        String
+  instructions       String?
+  categories         String[]
 
   isFeatured Boolean @default(false)
 
@@ -899,6 +991,9 @@ model StoreListingVersion {
   // Reviews for this specific version
   Reviews StoreListingReview[]
 
+  // Embedding for semantic search (one-to-one)
+  Embedding StoreListingEmbedding?
+
   @@unique([storeListingId, version])
   @@index([storeListingId, submissionStatus, isAvailable])
   @@index([submissionStatus])
@@ -924,6 +1019,24 @@ model StoreListingReview {
   @@index([reviewByUserId])
 }
 
+// Stores vector embeddings for semantic search of store listings
+// Uses pgvector extension for efficient similarity search
+model StoreListingEmbedding {
+  id        String   @id @default(uuid())
+  createdAt DateTime @default(now())
+  updatedAt DateTime @default(now()) @updatedAt
+
+  storeListingVersionId String              @unique
+  StoreListingVersion   StoreListingVersion @relation(fields: [storeListingVersionId], references: [id], onDelete: Cascade)
+
+  // pgvector embedding - stored as Unsupported type since Prisma doesn't natively support vector
+  embedding      Unsupported("vector(1536)")?
+  searchableText String? // The text that was embedded (for debugging/recomputation)
+  contentHash    String? // MD5 hash for change detection
+
+  @@index([contentHash])
+}
+
 enum SubmissionStatus {
   DRAFT // Being prepared, not yet submitted
   PENDING // Submitted, awaiting review
@@ -998,16 +1111,16 @@ model OAuthApplication {
   updatedAt DateTime @updatedAt
 
   // Application metadata
-  name               String
-  description        String?
-  logoUrl            String? // URL to app logo stored in GCS
-  clientId           String @unique
-  clientSecret       String // Hashed with Scrypt (same as API keys)
-  clientSecretSalt   String // Salt for Scrypt hashing
+  name             String
+  description      String?
+  logoUrl          String? // URL to app logo stored in GCS
+  clientId         String  @unique
+  clientSecret     String // Hashed with Scrypt (same as API keys)
+  clientSecretSalt String // Salt for Scrypt hashing
 
   // OAuth configuration
   redirectUris String[] // Allowed callback URLs
-  grantTypes   String[] @default(["authorization_code", "refresh_token"])
+  grantTypes   String[]           @default(["authorization_code", "refresh_token"])
   scopes       APIKeyPermission[] // Which permissions the app can request
 
   // Application management

autogpt_platform/backend/test/blocks/test_google_docs_format_text_color.py

@@ -0,0 +1,113 @@
+from unittest.mock import Mock
+
+from backend.blocks.google.docs import GoogleDocsFormatTextBlock
+
+
+def _make_mock_docs_service() -> Mock:
+    service = Mock()
+    # Ensure chained call exists: service.documents().batchUpdate(...).execute()
+    service.documents.return_value.batchUpdate.return_value.execute.return_value = {}
+    return service
+
+
+def test_format_text_parses_shorthand_hex_color():
+    block = GoogleDocsFormatTextBlock()
+    service = _make_mock_docs_service()
+
+    result = block._format_text(
+        service,
+        document_id="doc_1",
+        start_index=1,
+        end_index=2,
+        bold=False,
+        italic=False,
+        underline=False,
+        font_size=0,
+        foreground_color="#FFF",
+    )
+
+    assert result["success"] is True
+
+    # Verify request body contains correct rgbColor for white.
+    _, kwargs = service.documents.return_value.batchUpdate.call_args
+    requests = kwargs["body"]["requests"]
+    rgb = requests[0]["updateTextStyle"]["textStyle"]["foregroundColor"]["color"][
+        "rgbColor"
+    ]
+    assert rgb == {"red": 1.0, "green": 1.0, "blue": 1.0}
+
+
+def test_format_text_parses_full_hex_color():
+    block = GoogleDocsFormatTextBlock()
+    service = _make_mock_docs_service()
+
+    result = block._format_text(
+        service,
+        document_id="doc_1",
+        start_index=1,
+        end_index=2,
+        bold=False,
+        italic=False,
+        underline=False,
+        font_size=0,
+        foreground_color="#FF0000",
+    )
+
+    assert result["success"] is True
+
+    _, kwargs = service.documents.return_value.batchUpdate.call_args
+    requests = kwargs["body"]["requests"]
+    rgb = requests[0]["updateTextStyle"]["textStyle"]["foregroundColor"]["color"][
+        "rgbColor"
+    ]
+    assert rgb == {"red": 1.0, "green": 0.0, "blue": 0.0}
+
+
+def test_format_text_ignores_invalid_color_when_other_fields_present():
+    block = GoogleDocsFormatTextBlock()
+    service = _make_mock_docs_service()
+
+    result = block._format_text(
+        service,
+        document_id="doc_1",
+        start_index=1,
+        end_index=2,
+        bold=True,
+        italic=False,
+        underline=False,
+        font_size=0,
+        foreground_color="#GGG",
+    )
+
+    assert result["success"] is True
+    assert "warning" in result
+
+    # Should still apply bold, but should NOT include foregroundColor in textStyle.
+    _, kwargs = service.documents.return_value.batchUpdate.call_args
+    requests = kwargs["body"]["requests"]
+    text_style = requests[0]["updateTextStyle"]["textStyle"]
+    fields = requests[0]["updateTextStyle"]["fields"]
+
+    assert text_style == {"bold": True}
+    assert fields == "bold"
+
+
+def test_format_text_invalid_color_only_does_not_call_api():
+    block = GoogleDocsFormatTextBlock()
+    service = _make_mock_docs_service()
+
+    result = block._format_text(
+        service,
+        document_id="doc_1",
+        start_index=1,
+        end_index=2,
+        bold=False,
+        italic=False,
+        underline=False,
+        font_size=0,
+        foreground_color="#F",
+    )
+
+    assert result["success"] is False
+    assert "Invalid foreground_color" in result["message"]
+    service.documents.return_value.batchUpdate.assert_not_called()

autogpt_platform/backend/test/blocks/test_youtube.py

@@ -37,6 +37,18 @@ class TestTranscribeYoutubeVideoBlock:
         video_id = self.youtube_block.extract_video_id(url)
         assert video_id == "dQw4w9WgXcQ"
 
+    def test_extract_video_id_shorts_url(self):
+        """Test extracting video ID from YouTube Shorts URL."""
+        url = "https://www.youtube.com/shorts/dtUqwMu3e-g"
+        video_id = self.youtube_block.extract_video_id(url)
+        assert video_id == "dtUqwMu3e-g"
+
+    def test_extract_video_id_shorts_url_with_params(self):
+        """Test extracting video ID from YouTube Shorts URL with query parameters."""
+        url = "https://www.youtube.com/shorts/dtUqwMu3e-g?feature=share"
+        video_id = self.youtube_block.extract_video_id(url)
+        assert video_id == "dtUqwMu3e-g"
+
     @patch("backend.blocks.youtube.YouTubeTranscriptApi")
     def test_get_transcript_english_available(self, mock_api_class):
         """Test getting transcript when English is available."""

autogpt_platform/cloudflare_worker.js

@@ -0,0 +1,146 @@
+/**
+ * Cloudflare Workers Script for docs.agpt.co → agpt.co/docs migration
+ * 
+ * Deploy this script to handle all redirects with a single JavaScript file.
+ * No rule limits, easy to maintain, handles all edge cases.
+ */
+
+// URL mapping for special cases that don't follow patterns
+const SPECIAL_MAPPINGS = {
+  // Root page
+  '/': '/docs/platform',
+  
+  // Special cases that don't follow standard patterns
+  '/platform/d_id/': '/docs/integrations/block-integrations/d-id',
+  '/platform/blocks/blocks/': '/docs/integrations',
+  '/platform/blocks/decoder_block/': '/docs/integrations/block-integrations/text-decoder',
+  '/platform/blocks/http': '/docs/integrations/block-integrations/send-web-request',
+  '/platform/blocks/llm/': '/docs/integrations/block-integrations/ai-and-llm',
+  '/platform/blocks/time_blocks': '/docs/integrations/block-integrations/time-and-date',
+  '/platform/blocks/text_to_speech_block': '/docs/integrations/block-integrations/text-to-speech',
+  '/platform/blocks/ai_shortform_video_block': '/docs/integrations/block-integrations/ai-shortform-video',
+  '/platform/blocks/replicate_flux_advanced': '/docs/integrations/block-integrations/replicate-flux-advanced',
+  '/platform/blocks/flux_kontext': '/docs/integrations/block-integrations/flux-kontext',
+  '/platform/blocks/ai_condition/': '/docs/integrations/block-integrations/ai-condition',
+  '/platform/blocks/email_block': '/docs/integrations/block-integrations/email',
+  '/platform/blocks/google_maps': '/docs/integrations/block-integrations/google-maps',
+  '/platform/blocks/google/gmail': '/docs/integrations/block-integrations/gmail',
+  '/platform/blocks/github/issues/': '/docs/integrations/block-integrations/github-issues',
+  '/platform/blocks/github/repo/': '/docs/integrations/block-integrations/github-repo',
+  '/platform/blocks/github/pull_requests': '/docs/integrations/block-integrations/github-pull-requests',
+  '/platform/blocks/twitter/twitter': '/docs/integrations/block-integrations/twitter',
+  '/classic/setup/': '/docs/classic/setup/setting-up-autogpt-classic',
+  '/code-of-conduct/': '/docs/classic/help-us-improve-autogpt/code-of-conduct',
+  '/contributing/': '/docs/classic/contributing',
+  '/contribute/': '/docs/contribute',
+  '/forge/components/introduction/': '/docs/classic/forge/introduction'
+};
+
+/**
+ * Transform path by replacing underscores with hyphens and removing trailing slashes
+ */
+function transformPath(path) {
+  return path.replace(/_/g, '-').replace(/\/$/, '');
+}
+
+/**
+ * Handle docs.agpt.co redirects
+ */
+function handleDocsRedirect(url) {
+  const pathname = url.pathname;
+  
+  // Check special mappings first
+  if (SPECIAL_MAPPINGS[pathname]) {
+    return `https://agpt.co${SPECIAL_MAPPINGS[pathname]}`;
+  }
+  
+  // Pattern-based redirects
+  
+  // Platform blocks: /platform/blocks/* → /docs/integrations/block-integrations/*
+  if (pathname.startsWith('/platform/blocks/')) {
+    const blockName = pathname.substring('/platform/blocks/'.length);
+    const transformedName = transformPath(blockName);
+    return `https://agpt.co/docs/integrations/block-integrations/${transformedName}`;
+  }
+  
+  // Platform contributing: /platform/contributing/* → /docs/platform/contributing/*
+  if (pathname.startsWith('/platform/contributing/')) {
+    const subPath = pathname.substring('/platform/contributing/'.length);
+    return `https://agpt.co/docs/platform/contributing/${subPath}`;
+  }
+  
+  // Platform general: /platform/* → /docs/platform/* (with underscore→hyphen)
+  if (pathname.startsWith('/platform/')) {
+    const subPath = pathname.substring('/platform/'.length);
+    const transformedPath = transformPath(subPath);
+    return `https://agpt.co/docs/platform/${transformedPath}`;
+  }
+  
+  // Forge components: /forge/components/* → /docs/classic/forge/introduction/*
+  if (pathname.startsWith('/forge/components/')) {
+    const subPath = pathname.substring('/forge/components/'.length);
+    return `https://agpt.co/docs/classic/forge/introduction/${subPath}`;
+  }
+  
+  // Forge general: /forge/* → /docs/classic/forge/*
+  if (pathname.startsWith('/forge/')) {
+    const subPath = pathname.substring('/forge/'.length);
+    return `https://agpt.co/docs/classic/forge/${subPath}`;
+  }
+  
+  // Classic: /classic/* → /docs/classic/*
+  if (pathname.startsWith('/classic/')) {
+    const subPath = pathname.substring('/classic/'.length);
+    return `https://agpt.co/docs/classic/${subPath}`;
+  }
+  
+  // Default fallback
+  return 'https://agpt.co/docs/';
+}
+
+/**
+ * Main Worker function
+ */
+export default {
+  async fetch(request, env, ctx) {
+    const url = new URL(request.url);
+    
+    // Only handle docs.agpt.co requests
+    if (url.hostname === 'docs.agpt.co') {
+      const redirectUrl = handleDocsRedirect(url);
+      
+      return new Response(null, {
+        status: 301,
+        headers: {
+          'Location': redirectUrl,
+          'Cache-Control': 'max-age=300' // Cache redirects for 5 minutes
+        }
+      });
+    }
+    
+    // For non-docs requests, pass through or return 404
+    return new Response('Not Found', { status: 404 });
+  }
+};
+
+// Test function for local development
+export function testRedirects() {
+  const testCases = [
+    'https://docs.agpt.co/',
+    'https://docs.agpt.co/platform/getting-started/',
+    'https://docs.agpt.co/platform/advanced_setup/',
+    'https://docs.agpt.co/platform/blocks/basic/',
+    'https://docs.agpt.co/platform/blocks/ai_condition/',
+    'https://docs.agpt.co/classic/setup/',
+    'https://docs.agpt.co/forge/components/agents/',
+    'https://docs.agpt.co/contributing/',
+    'https://docs.agpt.co/unknown-page'
+  ];
+  
+  console.log('Testing redirects:');
+  testCases.forEach(testUrl => {
+    const url = new URL(testUrl);
+    const result = handleDocsRedirect(url);
+    console.log(`${testUrl} → ${result}`);
+  });
+}

autogpt_platform/frontend/package.json

@@ -46,13 +46,14 @@
     "@radix-ui/react-scroll-area": "1.2.10",
     "@radix-ui/react-select": "2.2.6",
     "@radix-ui/react-separator": "1.1.7",
+    "@radix-ui/react-slider": "1.3.6",
     "@radix-ui/react-slot": "1.2.3",
     "@radix-ui/react-switch": "1.2.6",
     "@radix-ui/react-tabs": "1.1.13",
     "@radix-ui/react-toast": "1.2.15",
     "@radix-ui/react-tooltip": "1.2.8",
-    "@rjsf/core": "5.24.13",
-    "@rjsf/utils": "5.24.13",
+    "@rjsf/core": "6.1.2",
+    "@rjsf/utils": "6.1.2",
     "@rjsf/validator-ajv8": "5.24.13",
     "@sentry/nextjs": "10.27.0",
     "@supabase/ssr": "0.7.0",
@@ -69,6 +70,7 @@
     "cmdk": "1.1.1",
     "cookie": "1.0.2",
     "date-fns": "4.1.0",
+    "dexie": "4.2.1",
     "dotenv": "17.2.3",
     "elliptic": "6.6.1",
     "embla-carousel-react": "8.6.0",

autogpt_platform/frontend/pnpm-lock.yaml

@@ -62,6 +62,9 @@ importers:
       '@radix-ui/react-separator':
         specifier: 1.1.7
         version: 1.1.7(@types/react-dom@18.3.5(@types/react@18.3.17))(@types/react@18.3.17)(react-dom@18.3.1(react@18.3.1))(react@18.3.1)
+      '@radix-ui/react-slider':
+        specifier: 1.3.6
+        version: 1.3.6(@types/react-dom@18.3.5(@types/react@18.3.17))(@types/react@18.3.17)(react-dom@18.3.1(react@18.3.1))(react@18.3.1)
       '@radix-ui/react-slot':
         specifier: 1.2.3
         version: 1.2.3(@types/react@18.3.17)(react@18.3.1)
@@ -78,14 +81,14 @@ importers:
         specifier: 1.2.8
         version: 1.2.8(@types/react-dom@18.3.5(@types/react@18.3.17))(@types/react@18.3.17)(react-dom@18.3.1(react@18.3.1))(react@18.3.1)
       '@rjsf/core':
-        specifier: 5.24.13
-        version: 5.24.13(@rjsf/utils@5.24.13(react@18.3.1))(react@18.3.1)
+        specifier: 6.1.2
+        version: 6.1.2(@rjsf/utils@6.1.2(react@18.3.1))(react@18.3.1)
       '@rjsf/utils':
-        specifier: 5.24.13
-        version: 5.24.13(react@18.3.1)
+        specifier: 6.1.2
+        version: 6.1.2(react@18.3.1)
       '@rjsf/validator-ajv8':
         specifier: 5.24.13
-        version: 5.24.13(@rjsf/utils@5.24.13(react@18.3.1))
+        version: 5.24.13(@rjsf/utils@6.1.2(react@18.3.1))
       '@sentry/nextjs':
         specifier: 10.27.0
         version: 10.27.0(@opentelemetry/context-async-hooks@2.2.0(@opentelemetry/api@1.9.0))(@opentelemetry/core@2.2.0(@opentelemetry/api@1.9.0))(@opentelemetry/sdk-trace-base@2.2.0(@opentelemetry/api@1.9.0))(next@15.4.10(@babel/core@7.28.4)(@opentelemetry/api@1.9.0)(@playwright/test@1.56.1)(react-dom@18.3.1(react@18.3.1))(react@18.3.1))(react@18.3.1)(webpack@5.101.3(esbuild@0.25.9))
@@ -131,6 +134,9 @@ importers:
       date-fns:
         specifier: 4.1.0
         version: 4.1.0
+      dexie:
+        specifier: 4.2.1
+        version: 4.2.1
       dotenv:
         specifier: 17.2.3
         version: 17.2.3
@@ -2307,6 +2313,19 @@ packages:
       '@types/react-dom':
         optional: true
 
+  '@radix-ui/react-slider@1.3.6':
+    resolution: {integrity: sha512-JPYb1GuM1bxfjMRlNLE+BcmBC8onfCi60Blk7OBqi2MLTFdS+8401U4uFjnwkOr49BLmXxLC6JHkvAsx5OJvHw==}
+    peerDependencies:
+      '@types/react': '*'
+      '@types/react-dom': '*'
+      react: ^16.8 || ^17.0 || ^18.0 || ^19.0 || ^19.0.0-rc
+      react-dom: ^16.8 || ^17.0 || ^18.0 || ^19.0 || ^19.0.0-rc
+    peerDependenciesMeta:
+      '@types/react':
+        optional: true
+      '@types/react-dom':
+        optional: true
+
   '@radix-ui/react-slot@1.2.3':
     resolution: {integrity: sha512-aeNmHnBxbi2St0au6VBVC7JXFlhLlOnvIIlePNniyUNAClzmtAUEY8/pBiK3iHjufOlwA+c20/8jngo7xcrg8A==}
     peerDependencies:
@@ -2476,18 +2495,18 @@ packages:
       react-redux:
         optional: true
 
-  '@rjsf/core@5.24.13':
-    resolution: {integrity: sha512-ONTr14s7LFIjx2VRFLuOpagL76sM/HPy6/OhdBfq6UukINmTIs6+aFN0GgcR0aXQHFDXQ7f/fel0o/SO05Htdg==}
-    engines: {node: '>=14'}
+  '@rjsf/core@6.1.2':
+    resolution: {integrity: sha512-fcEO6kArMcVIzTBoBxNStqxzAL417NDw049nmNx11pIcMwUnU5sAkSW18c8kgZOT6v1xaZhQrY+X5cBzzHy9+g==}
+    engines: {node: '>=20'}
     peerDependencies:
-      '@rjsf/utils': ^5.24.x
-      react: ^16.14.0 || >=17
+      '@rjsf/utils': ^6.x
+      react: '>=18'
 
-  '@rjsf/utils@5.24.13':
-    resolution: {integrity: sha512-rNF8tDxIwTtXzz5O/U23QU73nlhgQNYJ+Sv5BAwQOIyhIE2Z3S5tUiSVMwZHt0julkv/Ryfwi+qsD4FiE5rOuw==}
-    engines: {node: '>=14'}
+  '@rjsf/utils@6.1.2':
+    resolution: {integrity: sha512-Px3FIkE1KK0745Qng9v88RZ0O7hcLf/1JUu0j00g+r6C8Zyokna42Hz/5TKyyQSKJqgVYcj2Z47YroVLenUM3A==}
+    engines: {node: '>=20'}
     peerDependencies:
-      react: ^16.14.0 || >=17
+      react: '>=18'
 
   '@rjsf/validator-ajv8@5.24.13':
     resolution: {integrity: sha512-oWHP7YK581M8I5cF1t+UXFavnv+bhcqjtL1a7MG/Kaffi0EwhgcYjODrD8SsnrhncsEYMqSECr4ZOEoirnEUWw==}
@@ -3640,6 +3659,9 @@ packages:
   '@webassemblyjs/wast-printer@1.14.1':
     resolution: {integrity: sha512-kPSSXE6De1XOR820C90RIo2ogvZG+c3KiHzqUoO/F34Y2shGzesfqv7o57xrxovZJH/MetF5UjroJ/R/3isoiw==}
 
+  '@x0k/json-schema-merge@1.0.2':
+    resolution: {integrity: sha512-1734qiJHNX3+cJGDMMw2yz7R+7kpbAtl5NdPs1c/0gO5kYT6s4dMbLXiIfpZNsOYhGZI3aH7FWrj4Zxz7epXNg==}
+
   '@xtuc/ieee754@1.2.0':
     resolution: {integrity: sha512-DX8nKgqcGwsc0eJSqYt5lwP4DH5FlHnmuWWBRy7X0NcaGR0ZtuyeESgMwTYVEtxmsNGY+qit4QYT/MIYTOTPeA==}
 
@@ -4156,12 +4178,6 @@ packages:
   compare-versions@6.1.1:
     resolution: {integrity: sha512-4hm4VPpIecmlg59CHXnRDnqGplJFrbLG4aFEl5vl6cK1u76ws3LLvX7ikFnTDl5vo39sjWD6AaDPYodJp/NNHg==}
 
-  compute-gcd@1.2.1:
-    resolution: {integrity: sha512-TwMbxBNz0l71+8Sc4czv13h4kEqnchV9igQZBi6QUaz09dnz13juGnnaWWJTRsP3brxOoxeB4SA2WELLw1hCtg==}
-
-  compute-lcm@1.1.2:
-    resolution: {integrity: sha512-OFNPdQAXnQhDSKioX8/XYT6sdUlXwpeMjfd6ApxMJfyZ4GxmLR1xvMERctlYhlHwIiz6CSpBc2+qYKjHGZw4TQ==}
-
   concat-map@0.0.1:
     resolution: {integrity: sha512-/Srv4dswyQNBfohGpz9o6Yb3Gz3SrUDqBH5rTuhGR7ahtlbYKnVxw2bCFMRljaA7EXHaXZ8wsHdodFvbkhKmqg==}
 
@@ -4428,6 +4444,9 @@ packages:
   devlop@1.1.0:
     resolution: {integrity: sha512-RWmIqhcFf1lRYBvNmr7qTNuyCt/7/ns2jbpp1+PalgE/rDQcBT0fioSMUpJ93irlUhC5hrg4cYqe6U+0ImW0rA==}
 
+  dexie@4.2.1:
+    resolution: {integrity: sha512-Ckej0NS6jxQ4Po3OrSQBFddayRhTCic2DoCAG5zacOfOVB9P2Q5Xc5uL/nVa7ZVs+HdMnvUPzLFCB/JwpB6Csg==}
+
   didyoumean@1.2.2:
     resolution: {integrity: sha512-gxtyfqMg7GKyhQmb056K7M3xszy/myH8w+B4RT+QXBQsvAOdc3XymqDDPHx1BgPgsdAA5SIifona89YtRATDzw==}
 
@@ -4823,6 +4842,9 @@ packages:
   fast-uri@3.0.6:
     resolution: {integrity: sha512-Atfo14OibSv5wAp4VWNsFYE1AchQRTv9cBGWET4pZWHzYshFSS9NQI6I57rdKn9croWVMbYFbLhJ+yJvmZIIHw==}
 
+  fast-uri@3.1.0:
+    resolution: {integrity: sha512-iPeeDKJSWf4IEOasVVrknXpaBV0IApz/gp7S2bb7Z4Lljbl2MGJRqInZiUrQwV16cpzw/D3S5j5Julj/gT52AA==}
+
   fastq@1.19.1:
     resolution: {integrity: sha512-GwLTyxkCXjXbxqIhTsMI2Nui8huMPtnxg7krajPJAjnEG/iiOS7i+zCtWGZR9G0NBKbXKh6X9m9UIsYX/N6vvQ==}
 
@@ -5471,13 +5493,6 @@ packages:
   json-parse-even-better-errors@2.3.1:
     resolution: {integrity: sha512-xyFwyhro/JEof6Ghe2iz2NcXoj2sloNsWr/XsERDK/oiPCfaNhl5ONfp+jQdAZRQQ0IJWNzH9zIZF7li91kh2w==}
 
-  json-schema-compare@0.2.2:
-    resolution: {integrity: sha512-c4WYmDKyJXhs7WWvAWm3uIYnfyWFoIp+JEoX34rctVvEkMYCPGhXtvmFFXiffBbxfZsvQ0RNnV5H7GvDF5HCqQ==}
-
-  json-schema-merge-allof@0.8.1:
-    resolution: {integrity: sha512-CTUKmIlPJbsWfzRRnOXz+0MjIqvnleIXwFTzz+t9T86HnYX/Rozria6ZVGLktAU9e+NygNljveP+yxqtQp/Q4w==}
-    engines: {node: '>=12.0.0'}
-
   json-schema-traverse@0.4.1:
     resolution: {integrity: sha512-xbbCH5dCYU5T8LcEhhuh7HJ88HXuW3qsI3Y0zOZFKfZEHcpWiHU/Jxzk629Brsab/mMiHQti9wMP+845RPe3Vg==}
 
@@ -5593,6 +5608,9 @@ packages:
   lodash-es@4.17.21:
     resolution: {integrity: sha512-mKnC+QJ9pWVzv+C4/U3rRsHapFfHvQFoFB92e52xeyGMcX6/OlIl78je1u8vePzYZSkkogMPJ2yjxxsb89cxyw==}
 
+  lodash-es@4.17.22:
+    resolution: {integrity: sha512-XEawp1t0gxSi9x01glktRZ5HDy0HXqrM0x5pXQM98EaI0NxO6jVM7omDOxsuEo5UIASAnm2bRp1Jt/e0a2XU8Q==}
+
   lodash.camelcase@4.3.0:
     resolution: {integrity: sha512-TwuEnCnxbc3rAvhf/LbG7tJUDzhqXyFnv3dtzLOPgCG/hODL7WFnsbwktkD7yUV0RrreP/l1PALq/YSg6VvjlA==}
 
@@ -5682,11 +5700,14 @@ packages:
   markdown-table@3.0.4:
     resolution: {integrity: sha512-wiYz4+JrLyb/DqW2hkFJxP7Vd7JuTDm77fvbM8VfEQdmSMqcImWeeRbHwZjBjIFki/VaMK2BhFi7oUUZeM5bqw==}
 
-  markdown-to-jsx@7.7.13:
-    resolution: {integrity: sha512-DiueEq2bttFcSxUs85GJcQVrOr0+VVsPfj9AEUPqmExJ3f8P/iQNvZHltV4tm1XVhu1kl0vWBZWT3l99izRMaA==}
+  markdown-to-jsx@8.0.0:
+    resolution: {integrity: sha512-hWEaRxeCDjes1CVUQqU+Ov0mCqBqkGhLKjL98KdbwHSgEWZZSJQeGlJQatVfeZ3RaxrfTrZZ3eczl2dhp5c/pA==}
     engines: {node: '>= 10'}
     peerDependencies:
       react: '>= 0.14.0'
+    peerDependenciesMeta:
+      react:
+        optional: true
 
   math-intrinsics@1.1.0:
     resolution: {integrity: sha512-/IXtbwEk5HTPyEwyKX6hGkYXxM9nbj64B+ilVJnC/R6B0pH5G4V3b0pVbL7DBj4tkhBAppbQUlf6F6Xl9LHu1g==}
@@ -7572,21 +7593,6 @@ packages:
     resolution: {integrity: sha512-b+1eJOlsR9K8HJpow9Ok3fiWOWSIcIzXodvv0rQjVoOVNpWMpxf1wZNpt4y9h10odCNrqnYp1OBzRktckBe3sA==}
     hasBin: true
 
-  validate.io-array@1.0.6:
-    resolution: {integrity: sha512-DeOy7CnPEziggrOO5CZhVKJw6S3Yi7e9e65R1Nl/RTN1vTQKnzjfvks0/8kQ40FP/dsjRAOd4hxmJ7uLa6vxkg==}
-
-  validate.io-function@1.0.2:
-    resolution: {integrity: sha512-LlFybRJEriSuBnUhQyG5bwglhh50EpTL2ul23MPIuR1odjO7XaMLFV8vHGwp7AZciFxtYOeiSCT5st+XSPONiQ==}
-
-  validate.io-integer-array@1.0.0:
-    resolution: {integrity: sha512-mTrMk/1ytQHtCY0oNO3dztafHYyGU88KL+jRxWuzfOmQb+4qqnWmI+gykvGp8usKZOM0H7keJHEbRaFiYA0VrA==}
-
-  validate.io-integer@1.0.5:
-    resolution: {integrity: sha512-22izsYSLojN/P6bppBqhgUDjCkr5RY2jd+N2a3DCAUey8ydvrZ/OkGvFPR7qfOpwR2LC5p4Ngzxz36g5Vgr/hQ==}
-
-  validate.io-number@1.0.3:
-    resolution: {integrity: sha512-kRAyotcbNaSYoDnXvb4MHg/0a1egJdLwS6oJ38TJY7aw9n93Fl/3blIXdyYvPOp55CNxywooG/3BcrwNrBpcSg==}
-
   validator@13.15.20:
     resolution: {integrity: sha512-KxPOq3V2LmfQPP4eqf3Mq/zrT0Dqp2Vmx2Bn285LwVahLc+CsxOM0crBHczm8ijlcjZ0Q5Xd6LW3z3odTPnlrw==}
     engines: {node: '>= 0.10'}
@@ -9897,6 +9903,25 @@ snapshots:
       '@types/react': 18.3.17
       '@types/react-dom': 18.3.5(@types/react@18.3.17)
 
+  '@radix-ui/react-slider@1.3.6(@types/react-dom@18.3.5(@types/react@18.3.17))(@types/react@18.3.17)(react-dom@18.3.1(react@18.3.1))(react@18.3.1)':
+    dependencies:
+      '@radix-ui/number': 1.1.1
+      '@radix-ui/primitive': 1.1.3
+      '@radix-ui/react-collection': 1.1.7(@types/react-dom@18.3.5(@types/react@18.3.17))(@types/react@18.3.17)(react-dom@18.3.1(react@18.3.1))(react@18.3.1)
+      '@radix-ui/react-compose-refs': 1.1.2(@types/react@18.3.17)(react@18.3.1)
+      '@radix-ui/react-context': 1.1.2(@types/react@18.3.17)(react@18.3.1)
+      '@radix-ui/react-direction': 1.1.1(@types/react@18.3.17)(react@18.3.1)
+      '@radix-ui/react-primitive': 2.1.3(@types/react-dom@18.3.5(@types/react@18.3.17))(@types/react@18.3.17)(react-dom@18.3.1(react@18.3.1))(react@18.3.1)
+      '@radix-ui/react-use-controllable-state': 1.2.2(@types/react@18.3.17)(react@18.3.1)
+      '@radix-ui/react-use-layout-effect': 1.1.1(@types/react@18.3.17)(react@18.3.1)
+      '@radix-ui/react-use-previous': 1.1.1(@types/react@18.3.17)(react@18.3.1)
+      '@radix-ui/react-use-size': 1.1.1(@types/react@18.3.17)(react@18.3.1)
+      react: 18.3.1
+      react-dom: 18.3.1(react@18.3.1)
+    optionalDependencies:
+      '@types/react': 18.3.17
+      '@types/react-dom': 18.3.5(@types/react@18.3.17)
+
   '@radix-ui/react-slot@1.2.3(@types/react@18.3.17)(react@18.3.1)':
     dependencies:
       '@radix-ui/react-compose-refs': 1.1.2(@types/react@18.3.17)(react@18.3.1)
@@ -10059,27 +10084,28 @@ snapshots:
       react: 18.3.1
       react-redux: 9.2.0(@types/react@18.3.17)(react@18.3.1)(redux@5.0.1)
 
-  '@rjsf/core@5.24.13(@rjsf/utils@5.24.13(react@18.3.1))(react@18.3.1)':
+  '@rjsf/core@6.1.2(@rjsf/utils@6.1.2(react@18.3.1))(react@18.3.1)':
     dependencies:
-      '@rjsf/utils': 5.24.13(react@18.3.1)
+      '@rjsf/utils': 6.1.2(react@18.3.1)
       lodash: 4.17.21
-      lodash-es: 4.17.21
-      markdown-to-jsx: 7.7.13(react@18.3.1)
+      lodash-es: 4.17.22
+      markdown-to-jsx: 8.0.0(react@18.3.1)
       prop-types: 15.8.1
       react: 18.3.1
 
-  '@rjsf/utils@5.24.13(react@18.3.1)':
+  '@rjsf/utils@6.1.2(react@18.3.1)':
     dependencies:
-      json-schema-merge-allof: 0.8.1
+      '@x0k/json-schema-merge': 1.0.2
+      fast-uri: 3.1.0
       jsonpointer: 5.0.1
       lodash: 4.17.21
-      lodash-es: 4.17.21
+      lodash-es: 4.17.22
       react: 18.3.1
       react-is: 18.3.1
 
-  '@rjsf/validator-ajv8@5.24.13(@rjsf/utils@5.24.13(react@18.3.1))':
+  '@rjsf/validator-ajv8@5.24.13(@rjsf/utils@6.1.2(react@18.3.1))':
     dependencies:
-      '@rjsf/utils': 5.24.13(react@18.3.1)
+      '@rjsf/utils': 6.1.2(react@18.3.1)
       ajv: 8.17.1
       ajv-formats: 2.1.1(ajv@8.17.1)
       lodash: 4.17.21
@@ -11496,6 +11522,10 @@ snapshots:
       '@webassemblyjs/ast': 1.14.1
       '@xtuc/long': 4.2.2
 
+  '@x0k/json-schema-merge@1.0.2':
+    dependencies:
+      '@types/json-schema': 7.0.15
+
   '@xtuc/ieee754@1.2.0': {}
 
   '@xtuc/long@4.2.2': {}
@@ -12035,19 +12065,6 @@ snapshots:
 
   compare-versions@6.1.1: {}
 
-  compute-gcd@1.2.1:
-    dependencies:
-      validate.io-array: 1.0.6
-      validate.io-function: 1.0.2
-      validate.io-integer-array: 1.0.0
-
-  compute-lcm@1.1.2:
-    dependencies:
-      compute-gcd: 1.2.1
-      validate.io-array: 1.0.6
-      validate.io-function: 1.0.2
-      validate.io-integer-array: 1.0.0
-
   concat-map@0.0.1: {}
 
   concurrently@9.2.1:
@@ -12323,6 +12340,8 @@ snapshots:
     dependencies:
       dequal: 2.0.3
 
+  dexie@4.2.1: {}
+
   didyoumean@1.2.2: {}
 
   diffie-hellman@5.0.3:
@@ -12924,6 +12943,8 @@ snapshots:
 
   fast-uri@3.0.6: {}
 
+  fast-uri@3.1.0: {}
+
   fastq@1.19.1:
     dependencies:
       reusify: 1.1.0
@@ -13011,7 +13032,7 @@ snapshots:
       minimatch: 3.1.2
       node-abort-controller: 3.1.1
       schema-utils: 3.3.0
-      semver: 7.7.2
+      semver: 7.7.3
       tapable: 2.2.3
       typescript: 5.9.3
       webpack: 5.101.3(esbuild@0.25.9)
@@ -13633,16 +13654,6 @@ snapshots:
 
   json-parse-even-better-errors@2.3.1: {}
 
-  json-schema-compare@0.2.2:
-    dependencies:
-      lodash: 4.17.21
-
-  json-schema-merge-allof@0.8.1:
-    dependencies:
-      compute-lcm: 1.1.2
-      json-schema-compare: 0.2.2
-      lodash: 4.17.21
-
   json-schema-traverse@0.4.1: {}
 
   json-schema-traverse@1.0.0: {}
@@ -13758,6 +13769,8 @@ snapshots:
 
   lodash-es@4.17.21: {}
 
+  lodash-es@4.17.22: {}
+
   lodash.camelcase@4.3.0: {}
 
   lodash.debounce@4.0.8: {}
@@ -13837,8 +13850,8 @@ snapshots:
 
   markdown-table@3.0.4: {}
 
-  markdown-to-jsx@7.7.13(react@18.3.1):
-    dependencies:
+  markdown-to-jsx@8.0.0(react@18.3.1):
+    optionalDependencies:
       react: 18.3.1
 
   math-intrinsics@1.1.0: {}
@@ -16194,21 +16207,6 @@ snapshots:
 
   uuid@9.0.1: {}
 
-  validate.io-array@1.0.6: {}
-
-  validate.io-function@1.0.2: {}
-
-  validate.io-integer-array@1.0.0:
-    dependencies:
-      validate.io-array: 1.0.6
-      validate.io-integer: 1.0.5
-
-  validate.io-integer@1.0.5:
-    dependencies:
-      validate.io-number: 1.0.3
-
-  validate.io-number@1.0.3: {}
-
   validator@13.15.20: {}
 
   vaul@1.1.2(@types/react-dom@18.3.5(@types/react@18.3.17))(@types/react@18.3.17)(react-dom@18.3.1(react@18.3.1))(react@18.3.1):

autogpt_platform/frontend/src/app/(no-navbar)/onboarding/5-run/components/AgentOnboardingCredentials/AgentOnboardingCredentials.tsx

@@ -1,6 +1,6 @@
-import { CredentialsInput } from "@/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/modals/CredentialsInputs/CredentialsInputs";
 import { CredentialsMetaInput } from "@/app/api/__generated__/models/credentialsMetaInput";
 import { GraphMeta } from "@/app/api/__generated__/models/graphMeta";
+import { CredentialsInput } from "@/components/contextual/CredentialsInputs/CredentialsInputs";
 import { useState } from "react";
 import { getSchemaDefaultCredentials } from "../../helpers";
 import { areAllCredentialsSet, getCredentialFields } from "./helpers";

autogpt_platform/frontend/src/app/(no-navbar)/onboarding/5-run/page.tsx

@@ -1,12 +1,12 @@
 "use client";
 
-import { RunAgentInputs } from "@/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/modals/RunAgentInputs/RunAgentInputs";
 import {
   Card,
   CardContent,
   CardHeader,
   CardTitle,
 } from "@/components/__legacy__/ui/card";
+import { RunAgentInputs } from "@/components/contextual/RunAgentInputs/RunAgentInputs";
 import { ErrorCard } from "@/components/molecules/ErrorCard/ErrorCard";
 import { CircleNotchIcon } from "@phosphor-icons/react/dist/ssr";
 import { Play } from "lucide-react";

autogpt_platform/frontend/src/app/(platform)/PlatformLayoutContent.tsx

@@ -0,0 +1,48 @@
+"use client";
+
+import { ChatDrawer } from "@/components/contextual/Chat/ChatDrawer";
+import { usePathname } from "next/navigation";
+import { Children, ReactNode } from "react";
+
+interface PlatformLayoutContentProps {
+  children: ReactNode;
+}
+
+export function PlatformLayoutContent({
+  children,
+}: PlatformLayoutContentProps) {
+  const pathname = usePathname();
+  const isAuthPage =
+    pathname?.includes("/login") || pathname?.includes("/signup");
+
+  // Extract Navbar, AdminImpersonationBanner, and page content from children
+  const childrenArray = Children.toArray(children);
+  const navbar = childrenArray[0];
+  const adminBanner = childrenArray[1];
+  const pageContent = childrenArray.slice(2);
+
+  // For login/signup pages, use a simpler layout that doesn't interfere with centering
+  if (isAuthPage) {
+    return (
+      <main className="flex min-h-screen w-full flex-col">
+        {navbar}
+        {adminBanner}
+        <section className="flex-1">{pageContent}</section>
+        {/* ChatDrawer must always be rendered to maintain consistent hook count */}
+        <ChatDrawer />
+      </main>
+    );
+  }
+
+  // For logged-in pages, use the drawer layout
+  return (
+    <main className="flex h-screen w-full flex-col overflow-hidden">
+      {navbar}
+      {adminBanner}
+      <section className="flex min-h-0 flex-1 overflow-auto">
+        {pageContent}
+      </section>
+      <ChatDrawer />
+    </main>
+  );
+}

autogpt_platform/frontend/src/app/(platform)/auth/integrations/oauth_callback/route.ts

@@ -1,4 +1,4 @@
-import { OAuthPopupResultMessage } from "@/components/renderers/input-renderer/fields/CredentialField/models/OAuthCredentialModal/useOAuthCredentialModal";
+import { OAuthPopupResultMessage } from "./types";
 import { NextResponse } from "next/server";
 
 // This route is intended to be used as the callback for integration OAuth flows,

autogpt_platform/frontend/src/app/(platform)/auth/integrations/oauth_callback/types.ts

@@ -0,0 +1,11 @@
+export type OAuthPopupResultMessage = { message_type: "oauth_popup_result" } & (
+  | {
+      success: true;
+      code: string;
+      state: string;
+    }
+  | {
+      success: false;
+      message: string;
+    }
+);

autogpt_platform/frontend/src/app/(platform)/auth/integrations/setup-wizard/page.tsx

@@ -8,7 +8,7 @@ import { AuthCard } from "@/components/auth/AuthCard";
 import { Text } from "@/components/atoms/Text/Text";
 import { Button } from "@/components/atoms/Button/Button";
 import { ErrorCard } from "@/components/molecules/ErrorCard/ErrorCard";
-import { CredentialsInput } from "@/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/modals/CredentialsInputs/CredentialsInputs";
+import { CredentialsInput } from "@/components/contextual/CredentialsInputs/CredentialsInputs";
 import type {
   BlockIOCredentialsSubSchema,
   CredentialsMetaInput,

autogpt_platform/frontend/src/app/(platform)/build/components/BuilderActions/components/AgentOutputs/AgentOutputs.tsx

@@ -1,11 +1,6 @@
 import { BlockUIType } from "@/app/(platform)/build/components/types";
 import { useGraphStore } from "@/app/(platform)/build/stores/graphStore";
 import { useNodeStore } from "@/app/(platform)/build/stores/nodeStore";
-import {
-  globalRegistry,
-  OutputActions,
-  OutputItem,
-} from "@/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/selected-views/OutputRenderers";
 import { Label } from "@/components/__legacy__/ui/label";
 import { ScrollArea } from "@/components/__legacy__/ui/scroll-area";
 import {
@@ -16,16 +11,21 @@ import {
   SheetTitle,
   SheetTrigger,
 } from "@/components/__legacy__/ui/sheet";
+import { Button } from "@/components/atoms/Button/Button";
 import {
   Tooltip,
   TooltipContent,
   TooltipProvider,
   TooltipTrigger,
 } from "@/components/atoms/Tooltip/BaseTooltip";
+import {
+  globalRegistry,
+  OutputActions,
+  OutputItem,
+} from "@/components/contextual/OutputRenderers";
 import { BookOpenIcon } from "@phosphor-icons/react";
 import { useMemo } from "react";
 import { useShallow } from "zustand/react/shallow";
-import { BuilderActionButton } from "../BuilderActionButton";
 
 export const AgentOutputs = ({ flowID }: { flowID: string | null }) => {
   const hasOutputs = useGraphStore(useShallow((state) => state.hasOutputs));
@@ -76,9 +76,13 @@ export const AgentOutputs = ({ flowID }: { flowID: string | null }) => {
         <Tooltip>
           <TooltipTrigger asChild>
             <SheetTrigger asChild>
-              <BuilderActionButton disabled={!flowID || !hasOutputs()}>
-                <BookOpenIcon className="size-6" />
-              </BuilderActionButton>
+              <Button
+                variant="outline"
+                size="icon"
+                disabled={!flowID || !hasOutputs()}
+              >
+                <BookOpenIcon className="size-4" />
+              </Button>
             </SheetTrigger>
           </TooltipTrigger>
           <TooltipContent>

autogpt_platform/frontend/src/app/(platform)/build/components/BuilderActions/components/BuilderActionButton.tsx

@@ -1,37 +0,0 @@
-import { Button } from "@/components/atoms/Button/Button";
-import { ButtonProps } from "@/components/atoms/Button/helpers";
-import { cn } from "@/lib/utils";
-import { CircleNotchIcon } from "@phosphor-icons/react";
-
-export const BuilderActionButton = ({
-  children,
-  className,
-  isLoading,
-  ...props
-}: ButtonProps & { isLoading?: boolean }) => {
-  return (
-    <Button
-      variant="icon"
-      size={"small"}
-      className={cn(
-        "relative h-12 w-12 min-w-0 text-lg",
-        "bg-gradient-to-br from-zinc-50 to-zinc-200",
-        "border border-zinc-200",
-        "shadow-[inset_0_3px_0_0_rgba(255,255,255,0.5),0_2px_4px_0_rgba(0,0,0,0.2)]",
-        "dark:shadow-[inset_0_1px_0_0_rgba(255,255,255,0.1),0_2px_4px_0_rgba(0,0,0,0.4)]",
-        "hover:shadow-[inset_0_1px_0_0_rgba(255,255,255,0.5),0_1px_2px_0_rgba(0,0,0,0.2)]",
-        "active:shadow-[inset_0_2px_4px_0_rgba(0,0,0,0.2)]",
-        "transition-all duration-150",
-        "disabled:cursor-not-allowed disabled:opacity-50",
-        className,
-      )}
-      {...props}
-    >
-      {!isLoading ? (
-        children
-      ) : (
-        <CircleNotchIcon className="size-6 animate-spin" />
-      )}
-    </Button>
-  );
-};

autogpt_platform/frontend/src/app/(platform)/build/components/BuilderActions/components/PublishToMarketplace/PublishToMarketplace.tsx

@@ -1,12 +1,12 @@
-import { ShareIcon } from "@phosphor-icons/react";
-import { BuilderActionButton } from "../BuilderActionButton";
+import { Button } from "@/components/atoms/Button/Button";
 import {
   Tooltip,
   TooltipContent,
   TooltipTrigger,
 } from "@/components/atoms/Tooltip/BaseTooltip";
-import { usePublishToMarketplace } from "./usePublishToMarketplace";
 import { PublishAgentModal } from "@/components/contextual/PublishAgentModal/PublishAgentModal";
+import { ShareIcon } from "@phosphor-icons/react";
+import { usePublishToMarketplace } from "./usePublishToMarketplace";
 
 export const PublishToMarketplace = ({ flowID }: { flowID: string | null }) => {
   const { handlePublishToMarketplace, publishState, handleStateChange } =
@@ -16,12 +16,14 @@ export const PublishToMarketplace = ({ flowID }: { flowID: string | null }) => {
     <>
       <Tooltip>
         <TooltipTrigger asChild>
-          <BuilderActionButton
+          <Button
+            variant="outline"
+            size="icon"
             onClick={handlePublishToMarketplace}
             disabled={!flowID}
           >
-            <ShareIcon className="size-6 drop-shadow-sm" />
-          </BuilderActionButton>
+            <ShareIcon className="size-4" />
+          </Button>
         </TooltipTrigger>
         <TooltipContent>Publish to Marketplace</TooltipContent>
       </Tooltip>
@@ -30,6 +32,7 @@ export const PublishToMarketplace = ({ flowID }: { flowID: string | null }) => {
         targetState={publishState}
         onStateChange={handleStateChange}
         preSelectedAgentId={flowID || undefined}
+        showTrigger={false}
       />
     </>
   );

autogpt_platform/frontend/src/app/(platform)/build/components/BuilderActions/components/RunGraph/RunGraph.tsx

@@ -1,15 +1,14 @@
-import { useRunGraph } from "./useRunGraph";
 import { useGraphStore } from "@/app/(platform)/build/stores/graphStore";
-import { useShallow } from "zustand/react/shallow";
-import { PlayIcon, StopIcon } from "@phosphor-icons/react";
-import { cn } from "@/lib/utils";
-import { RunInputDialog } from "../RunInputDialog/RunInputDialog";
+import { Button } from "@/components/atoms/Button/Button";
 import {
   Tooltip,
   TooltipContent,
   TooltipTrigger,
 } from "@/components/atoms/Tooltip/BaseTooltip";
-import { BuilderActionButton } from "../BuilderActionButton";
+import { PlayIcon, StopIcon } from "@phosphor-icons/react";
+import { useShallow } from "zustand/react/shallow";
+import { RunInputDialog } from "../RunInputDialog/RunInputDialog";
+import { useRunGraph } from "./useRunGraph";
 
 export const RunGraph = ({ flowID }: { flowID: string | null }) => {
   const {
@@ -29,21 +28,19 @@ export const RunGraph = ({ flowID }: { flowID: string | null }) => {
     <>
       <Tooltip>
         <TooltipTrigger asChild>
-          <BuilderActionButton
-            className={cn(
-              isGraphRunning &&
-                "border-red-500 bg-gradient-to-br from-red-400 to-red-500 shadow-[inset_0_2px_0_0_rgba(255,255,255,0.5),0_2px_4px_0_rgba(0,0,0,0.2)]",
-            )}
+          <Button
+            size="icon"
+            variant={isGraphRunning ? "destructive" : "primary"}
             onClick={isGraphRunning ? handleStopGraph : handleRunGraph}
             disabled={!flowID || isExecutingGraph || isTerminatingGraph}
-            isLoading={isExecutingGraph || isTerminatingGraph || isSaving}
+            loading={isExecutingGraph || isTerminatingGraph || isSaving}
           >
             {!isGraphRunning ? (
-              <PlayIcon className="size-6 drop-shadow-sm" />
+              <PlayIcon className="size-4" />
             ) : (
-              <StopIcon className="size-6 drop-shadow-sm" />
+              <StopIcon className="size-4" />
             )}
-          </BuilderActionButton>
+          </Button>
         </TooltipTrigger>
         <TooltipContent>
           {isGraphRunning ? "Stop agent" : "Run agent"}

autogpt_platform/frontend/src/app/(platform)/build/components/BuilderActions/components/RunInputDialog/RunInputDialog.tsx

@@ -5,7 +5,7 @@ import { useGraphStore } from "@/app/(platform)/build/stores/graphStore";
 import { Button } from "@/components/atoms/Button/Button";
 import { ClockIcon, PlayIcon } from "@phosphor-icons/react";
 import { Text } from "@/components/atoms/Text/Text";
-import { FormRenderer } from "@/components/renderers/input-renderer/FormRenderer";
+import { FormRenderer } from "@/components/renderers/InputRenderer/FormRenderer";
 import { useRunInputDialog } from "./useRunInputDialog";
 import { CronSchedulerDialog } from "../CronSchedulerDialog/CronSchedulerDialog";
 

autogpt_platform/frontend/src/app/(platform)/build/components/BuilderActions/components/RunInputDialog/useRunInputDialog.ts

@@ -8,7 +8,7 @@ import {
 import { parseAsInteger, parseAsString, useQueryStates } from "nuqs";
 import { useMemo, useState } from "react";
 import { uiSchema } from "../../../FlowEditor/nodes/uiSchema";
-import { isCredentialFieldSchema } from "@/components/renderers/input-renderer/fields/CredentialField/helpers";
+import { isCredentialFieldSchema } from "@/components/renderers/InputRenderer/custom/CredentialField/helpers";
 
 export const useRunInputDialog = ({
   setIsOpen,

autogpt_platform/frontend/src/app/(platform)/build/components/BuilderActions/components/ScheduleGraph/ScheduleGraph.tsx

@@ -1,14 +1,14 @@
-import { ClockIcon } from "@phosphor-icons/react";
-import { RunInputDialog } from "../RunInputDialog/RunInputDialog";
-import { useScheduleGraph } from "./useScheduleGraph";
+import { Button } from "@/components/atoms/Button/Button";
 import {
   Tooltip,
   TooltipContent,
   TooltipProvider,
   TooltipTrigger,
 } from "@/components/atoms/Tooltip/BaseTooltip";
+import { ClockIcon } from "@phosphor-icons/react";
 import { CronSchedulerDialog } from "../CronSchedulerDialog/CronSchedulerDialog";
-import { BuilderActionButton } from "../BuilderActionButton";
+import { RunInputDialog } from "../RunInputDialog/RunInputDialog";
+import { useScheduleGraph } from "./useScheduleGraph";
 
 export const ScheduleGraph = ({ flowID }: { flowID: string | null }) => {
   const {
@@ -23,12 +23,14 @@ export const ScheduleGraph = ({ flowID }: { flowID: string | null }) => {
       <TooltipProvider>
         <Tooltip>
           <TooltipTrigger asChild>
-            <BuilderActionButton
+            <Button
+              variant="outline"
+              size="icon"
               onClick={handleScheduleGraph}
               disabled={!flowID}
             >
-              <ClockIcon className="size-6" />
-            </BuilderActionButton>
+              <ClockIcon className="size-4" />
+            </Button>
           </TooltipTrigger>
           <TooltipContent>
             <p>Schedule Graph</p>

autogpt_platform/frontend/src/app/(platform)/build/components/DraftRecoveryDialog/DraftRecoveryPopup.tsx

@@ -0,0 +1,160 @@
+"use client";
+
+import { Button } from "@/components/atoms/Button/Button";
+import { ClockCounterClockwiseIcon, XIcon } from "@phosphor-icons/react";
+import { cn } from "@/lib/utils";
+import { formatTimeAgo } from "@/lib/utils/time";
+import {
+  Tooltip,
+  TooltipContent,
+  TooltipTrigger,
+} from "@/components/atoms/Tooltip/BaseTooltip";
+import { useDraftRecoveryPopup } from "./useDraftRecoveryPopup";
+import { Text } from "@/components/atoms/Text/Text";
+import { AnimatePresence, motion } from "framer-motion";
+import { DraftDiff } from "@/lib/dexie/draft-utils";
+
+interface DraftRecoveryPopupProps {
+  isInitialLoadComplete: boolean;
+}
+
+function formatDiffSummary(diff: DraftDiff | null): string {
+  if (!diff) return "";
+
+  const parts: string[] = [];
+
+  // Node changes
+  const nodeChanges: string[] = [];
+  if (diff.nodes.added > 0) nodeChanges.push(`+${diff.nodes.added}`);
+  if (diff.nodes.removed > 0) nodeChanges.push(`-${diff.nodes.removed}`);
+  if (diff.nodes.modified > 0) nodeChanges.push(`~${diff.nodes.modified}`);
+
+  if (nodeChanges.length > 0) {
+    parts.push(
+      `${nodeChanges.join("/")} block${diff.nodes.added + diff.nodes.removed + diff.nodes.modified !== 1 ? "s" : ""}`,
+    );
+  }
+
+  // Edge changes
+  const edgeChanges: string[] = [];
+  if (diff.edges.added > 0) edgeChanges.push(`+${diff.edges.added}`);
+  if (diff.edges.removed > 0) edgeChanges.push(`-${diff.edges.removed}`);
+  if (diff.edges.modified > 0) edgeChanges.push(`~${diff.edges.modified}`);
+
+  if (edgeChanges.length > 0) {
+    parts.push(
+      `${edgeChanges.join("/")} connection${diff.edges.added + diff.edges.removed + diff.edges.modified !== 1 ? "s" : ""}`,
+    );
+  }
+
+  return parts.join(", ");
+}
+
+export function DraftRecoveryPopup({
+  isInitialLoadComplete,
+}: DraftRecoveryPopupProps) {
+  const {
+    isOpen,
+    popupRef,
+    nodeCount,
+    edgeCount,
+    diff,
+    savedAt,
+    onLoad,
+    onDiscard,
+  } = useDraftRecoveryPopup(isInitialLoadComplete);
+
+  const diffSummary = formatDiffSummary(diff);
+
+  return (
+    <AnimatePresence>
+      {isOpen && (
+        <motion.div
+          ref={popupRef}
+          className={cn("absolute left-1/2 top-4 z-50")}
+          initial={{
+            opacity: 0,
+            x: "-50%",
+            y: "-150%",
+            scale: 0.5,
+            filter: "blur(20px)",
+          }}
+          animate={{
+            opacity: 1,
+            x: "-50%",
+            y: "0%",
+            scale: 1,
+            filter: "blur(0px)",
+          }}
+          exit={{
+            opacity: 0,
+            y: "-150%",
+            scale: 0.5,
+            filter: "blur(20px)",
+            transition: { duration: 0.4, type: "spring", bounce: 0.2 },
+          }}
+          transition={{ duration: 0.2, type: "spring", bounce: 0.2 }}
+        >
+          <div
+            className={cn(
+              "flex items-center gap-3 rounded-xlarge border border-amber-200 bg-amber-50 px-4 py-3 shadow-lg",
+            )}
+          >
+            <div className="flex items-center gap-2 text-amber-700 dark:text-amber-300">
+              <ClockCounterClockwiseIcon className="h-5 w-5" weight="fill" />
+            </div>
+
+            <div className="flex flex-col">
+              <Text
+                variant="small-medium"
+                className="text-amber-900 dark:text-amber-100"
+              >
+                Unsaved changes found
+              </Text>
+              <Text
+                variant="small"
+                className="text-amber-700 dark:text-amber-400"
+              >
+                {diffSummary ||
+                  `${nodeCount} block${nodeCount !== 1 ? "s" : ""}, ${edgeCount} connection${edgeCount !== 1 ? "s" : ""}`}{" "}
+                • {formatTimeAgo(new Date(savedAt).toISOString())}
+              </Text>
+            </div>
+
+            <div className="ml-2 flex items-center gap-2">
+              <Tooltip delayDuration={10}>
+                <TooltipTrigger asChild>
+                  <Button
+                    variant="primary"
+                    size="small"
+                    onClick={onLoad}
+                    className="aspect-square min-w-0 p-1.5"
+                  >
+                    <ClockCounterClockwiseIcon size={20} weight="fill" />
+                    <span className="sr-only">Restore changes</span>
+                  </Button>
+                </TooltipTrigger>
+                <TooltipContent>Restore changes</TooltipContent>
+              </Tooltip>
+              <Tooltip delayDuration={10}>
+                <TooltipTrigger asChild>
+                  <Button
+                    variant="destructive"
+                    size="icon"
+                    onClick={onDiscard}
+                    aria-label="Discard changes"
+                    className="aspect-square min-w-0 p-1.5"
+                  >
+                    <XIcon size={20} />
+                    <span className="sr-only">Discard changes</span>
+                  </Button>
+                </TooltipTrigger>
+                <TooltipContent>Discard changes</TooltipContent>
+              </Tooltip>
+            </div>
+          </div>
+        </motion.div>
+      )}
+    </AnimatePresence>
+  );
+}

autogpt_platform/frontend/src/app/(platform)/build/components/DraftRecoveryDialog/useDraftRecoveryPopup.tsx

@@ -0,0 +1,63 @@
+import { useEffect, useRef } from "react";
+import { useDraftManager } from "../FlowEditor/Flow/useDraftManager";
+
+export const useDraftRecoveryPopup = (isInitialLoadComplete: boolean) => {
+  const popupRef = useRef<HTMLDivElement>(null);
+
+  const {
+    isRecoveryOpen: isOpen,
+    savedAt,
+    nodeCount,
+    edgeCount,
+    diff,
+    loadDraft: onLoad,
+    discardDraft: onDiscard,
+  } = useDraftManager(isInitialLoadComplete);
+
+  useEffect(() => {
+    if (!isOpen) return;
+
+    const handleClickOutside = (event: MouseEvent) => {
+      if (
+        popupRef.current &&
+        !popupRef.current.contains(event.target as Node)
+      ) {
+        onDiscard();
+      }
+    };
+
+    const timeoutId = setTimeout(() => {
+      document.addEventListener("mousedown", handleClickOutside);
+    }, 100);
+
+    return () => {
+      clearTimeout(timeoutId);
+      document.removeEventListener("mousedown", handleClickOutside);
+    };
+  }, [isOpen, onDiscard]);
+
+  useEffect(() => {
+    if (!isOpen) return;
+
+    const handleKeyDown = (event: KeyboardEvent) => {
+      if (event.key === "Escape") {
+        onDiscard();
+      }
+    };
+
+    document.addEventListener("keydown", handleKeyDown);
+    return () => {
+      document.removeEventListener("keydown", handleKeyDown);
+    };
+  }, [isOpen, onDiscard]);
+  return {
+    popupRef,
+    isOpen,
+    nodeCount,
+    edgeCount,
+    diff,
+    savedAt,
+    onLoad,
+    onDiscard,
+  };
+};

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/Flow/Flow.tsx

@@ -1,26 +1,27 @@
-import { ReactFlow, Background } from "@xyflow/react";
-import NewControlPanel from "../../NewControlPanel/NewControlPanel";
-import CustomEdge from "../edges/CustomEdge";
-import { useFlow } from "./useFlow";
-import { useShallow } from "zustand/react/shallow";
-import { useNodeStore } from "../../../stores/nodeStore";
-import { useMemo, useEffect, useCallback } from "react";
-import { CustomNode } from "../nodes/CustomNode/CustomNode";
-import { useCustomEdge } from "../edges/useCustomEdge";
-import { useFlowRealtime } from "./useFlowRealtime";
-import { GraphLoadingBox } from "./components/GraphLoadingBox";
-import { BuilderActions } from "../../BuilderActions/BuilderActions";
-import { RunningBackground } from "./components/RunningBackground";
-import { useGraphStore } from "../../../stores/graphStore";
-import { useCopyPaste } from "./useCopyPaste";
-import { FloatingReviewsPanel } from "@/components/organisms/FloatingReviewsPanel/FloatingReviewsPanel";
-import { parseAsString, useQueryStates } from "nuqs";
-import { CustomControls } from "./components/CustomControl";
 import { useGetV1GetSpecificGraph } from "@/app/api/__generated__/endpoints/graphs/graphs";
 import { okData } from "@/app/api/helpers";
+import { FloatingReviewsPanel } from "@/components/organisms/FloatingReviewsPanel/FloatingReviewsPanel";
+import { Background, ReactFlow } from "@xyflow/react";
+import { parseAsString, useQueryStates } from "nuqs";
+import { useCallback, useMemo } from "react";
+import { useShallow } from "zustand/react/shallow";
+import { useGraphStore } from "../../../stores/graphStore";
+import { useNodeStore } from "../../../stores/nodeStore";
+import { BuilderActions } from "../../BuilderActions/BuilderActions";
+import { DraftRecoveryPopup } from "../../DraftRecoveryDialog/DraftRecoveryPopup";
+import { FloatingSafeModeToggle } from "../../FloatingSafeModeToogle";
+import NewControlPanel from "../../NewControlPanel/NewControlPanel";
+import CustomEdge from "../edges/CustomEdge";
+import { useCustomEdge } from "../edges/useCustomEdge";
+import { CustomNode } from "../nodes/CustomNode/CustomNode";
+import { CustomControls } from "./components/CustomControl";
+import { GraphLoadingBox } from "./components/GraphLoadingBox";
+import { RunningBackground } from "./components/RunningBackground";
 import { TriggerAgentBanner } from "./components/TriggerAgentBanner";
 import { resolveCollisions } from "./helpers/resolve-collision";
-import { FloatingSafeModeToggle } from "../../FloatingSafeModeToogle";
+import { useCopyPaste } from "./useCopyPaste";
+import { useFlow } from "./useFlow";
+import { useFlowRealtime } from "./useFlowRealtime";
 
 export const Flow = () => {
   const [{ flowID, flowExecutionID }] = useQueryStates({
@@ -41,14 +42,18 @@ export const Flow = () => {
 
   const nodes = useNodeStore(useShallow((state) => state.nodes));
   const setNodes = useNodeStore(useShallow((state) => state.setNodes));
+
   const onNodesChange = useNodeStore(
     useShallow((state) => state.onNodesChange),
   );
+
   const hasWebhookNodes = useNodeStore(
     useShallow((state) => state.hasWebhookNodes()),
   );
+
   const nodeTypes = useMemo(() => ({ custom: CustomNode }), []);
   const edgeTypes = useMemo(() => ({ custom: CustomEdge }), []);
+
   const onNodeDragStop = useCallback(() => {
     setNodes(
       resolveCollisions(nodes, {
@@ -60,29 +65,26 @@ export const Flow = () => {
   }, [setNodes, nodes]);
   const { edges, onConnect, onEdgesChange } = useCustomEdge();
 
-  // We use this hook to load the graph and convert them into custom nodes and edges.
-  const { onDragOver, onDrop, isFlowContentLoading, isLocked, setIsLocked } =
-    useFlow();
+  // for loading purpose
+  const {
+    onDragOver,
+    onDrop,
+    isFlowContentLoading,
+    isInitialLoadComplete,
+    isLocked,
+    setIsLocked,
+  } = useFlow();
 
   // This hook is used for websocket realtime updates.
   useFlowRealtime();
 
   // Copy/paste functionality
-  const handleCopyPaste = useCopyPaste();
+  useCopyPaste();
 
-  useEffect(() => {
-    const handleKeyDown = (event: KeyboardEvent) => {
-      handleCopyPaste(event);
-    };
-
-    window.addEventListener("keydown", handleKeyDown);
-    return () => {
-      window.removeEventListener("keydown", handleKeyDown);
-    };
-  }, [handleCopyPaste]);
   const isGraphRunning = useGraphStore(
     useShallow((state) => state.isGraphRunning),
   );
+
   return (
     <div className="flex h-full w-full dark:bg-slate-900">
       <div className="relative flex-1">
@@ -102,6 +104,7 @@ export const Flow = () => {
           nodesDraggable={!isLocked}
           nodesConnectable={!isLocked}
           elementsSelectable={!isLocked}
+          deleteKeyCode={["Backspace", "Delete"]}
         >
           <Background />
           <CustomControls setIsLocked={setIsLocked} isLocked={isLocked} />
@@ -115,6 +118,7 @@ export const Flow = () => {
               className="right-2 top-32 p-2"
             />
           )}
+          <DraftRecoveryPopup isInitialLoadComplete={isInitialLoadComplete} />
         </ReactFlow>
       </div>
       {/* TODO: Need to update it in future - also do not send executionId as prop - rather use useQueryState inside the component */}

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/Flow/helpers/resolve-collision.ts

@@ -48,8 +48,6 @@ export const resolveCollisions: CollisionAlgorithm = (
     const width = (node.width ?? node.measured?.width ?? 0) + margin * 2;
     const height = (node.height ?? node.measured?.height ?? 0) + margin * 2;
 
-    console.log("width", width);
-    console.log("height", height);
     const x = node.position.x - margin;
     const y = node.position.y - margin;
 

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/Flow/useCopyPaste.ts

@@ -1,4 +1,4 @@
-import { useCallback } from "react";
+import { useCallback, useEffect } from "react";
 import { useReactFlow } from "@xyflow/react";
 import { v4 as uuidv4 } from "uuid";
 import { useNodeStore } from "../../../stores/nodeStore";
@@ -151,5 +151,16 @@ export function useCopyPaste() {
     [getViewport, toast],
   );
 
+  useEffect(() => {
+    const handleKeyDown = (event: KeyboardEvent) => {
+      handleCopyPaste(event);
+    };
+
+    window.addEventListener("keydown", handleKeyDown);
+    return () => {
+      window.removeEventListener("keydown", handleKeyDown);
+    };
+  }, [handleCopyPaste]);
+
   return handleCopyPaste;
 }

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/Flow/useDraftManager.ts

@@ -0,0 +1,319 @@
+import { useState, useCallback, useEffect, useRef } from "react";
+import { parseAsString, parseAsInteger, useQueryStates } from "nuqs";
+import {
+  draftService,
+  getTempFlowId,
+  getOrCreateTempFlowId,
+  DraftData,
+} from "@/services/builder-draft/draft-service";
+import { BuilderDraft } from "@/lib/dexie/db";
+import {
+  cleanNodes,
+  cleanEdges,
+  calculateDraftDiff,
+  DraftDiff,
+} from "@/lib/dexie/draft-utils";
+import { useNodeStore } from "../../../stores/nodeStore";
+import { useEdgeStore } from "../../../stores/edgeStore";
+import { useGraphStore } from "../../../stores/graphStore";
+import { useHistoryStore } from "../../../stores/historyStore";
+import isEqual from "lodash/isEqual";
+
+const AUTO_SAVE_INTERVAL_MS = 15000; // 15 seconds
+
+interface DraftRecoveryState {
+  isOpen: boolean;
+  draft: BuilderDraft | null;
+  diff: DraftDiff | null;
+}
+
+/**
+ * Consolidated hook for draft persistence and recovery
+ * - Auto-saves builder state every 15 seconds
+ * - Saves on beforeunload event
+ * - Checks for and manages unsaved drafts on load
+ */
+export function useDraftManager(isInitialLoadComplete: boolean) {
+  const [state, setState] = useState<DraftRecoveryState>({
+    isOpen: false,
+    draft: null,
+    diff: null,
+  });
+
+  const [{ flowID, flowVersion }] = useQueryStates({
+    flowID: parseAsString,
+    flowVersion: parseAsInteger,
+  });
+
+  const lastSavedStateRef = useRef<DraftData | null>(null);
+  const saveTimeoutRef = useRef<NodeJS.Timeout | null>(null);
+  const isDirtyRef = useRef(false);
+  const hasCheckedForDraft = useRef(false);
+
+  const getEffectiveFlowId = useCallback((): string => {
+    return flowID || getOrCreateTempFlowId();
+  }, [flowID]);
+
+  const getCurrentState = useCallback((): DraftData => {
+    const nodes = useNodeStore.getState().nodes;
+    const edges = useEdgeStore.getState().edges;
+    const nodeCounter = useNodeStore.getState().nodeCounter;
+    const graphStore = useGraphStore.getState();
+
+    return {
+      nodes,
+      edges,
+      graphSchemas: {
+        input: graphStore.inputSchema,
+        credentials: graphStore.credentialsInputSchema,
+        output: graphStore.outputSchema,
+      },
+      nodeCounter,
+      flowVersion: flowVersion ?? undefined,
+    };
+  }, [flowVersion]);
+
+  const cleanStateForComparison = useCallback((stateData: DraftData) => {
+    return {
+      nodes: cleanNodes(stateData.nodes),
+      edges: cleanEdges(stateData.edges),
+    };
+  }, []);
+
+  const hasChanges = useCallback((): boolean => {
+    const currentState = getCurrentState();
+
+    if (!lastSavedStateRef.current) {
+      return currentState.nodes.length > 0;
+    }
+
+    const currentClean = cleanStateForComparison(currentState);
+    const lastClean = cleanStateForComparison(lastSavedStateRef.current);
+
+    return !isEqual(currentClean, lastClean);
+  }, [getCurrentState, cleanStateForComparison]);
+
+  const saveDraft = useCallback(async () => {
+    const effectiveFlowId = getEffectiveFlowId();
+    const currentState = getCurrentState();
+
+    if (currentState.nodes.length === 0 && currentState.edges.length === 0) {
+      return;
+    }
+
+    if (!hasChanges()) {
+      return;
+    }
+
+    try {
+      await draftService.saveDraft(effectiveFlowId, currentState);
+      lastSavedStateRef.current = currentState;
+      isDirtyRef.current = false;
+    } catch (error) {
+      console.error("[DraftPersistence] Failed to save draft:", error);
+    }
+  }, [getEffectiveFlowId, getCurrentState, hasChanges]);
+
+  const scheduleSave = useCallback(() => {
+    isDirtyRef.current = true;
+
+    if (saveTimeoutRef.current) {
+      clearTimeout(saveTimeoutRef.current);
+    }
+
+    saveTimeoutRef.current = setTimeout(() => {
+      saveDraft();
+    }, AUTO_SAVE_INTERVAL_MS);
+  }, [saveDraft]);
+
+  useEffect(() => {
+    const unsubscribeNodes = useNodeStore.subscribe((storeState, prevState) => {
+      if (storeState.nodes !== prevState.nodes) {
+        scheduleSave();
+      }
+    });
+
+    const unsubscribeEdges = useEdgeStore.subscribe((storeState, prevState) => {
+      if (storeState.edges !== prevState.edges) {
+        scheduleSave();
+      }
+    });
+
+    return () => {
+      unsubscribeNodes();
+      unsubscribeEdges();
+    };
+  }, [scheduleSave]);
+
+  useEffect(() => {
+    const handleBeforeUnload = () => {
+      if (isDirtyRef.current) {
+        const effectiveFlowId = getEffectiveFlowId();
+        const currentState = getCurrentState();
+
+        if (
+          currentState.nodes.length === 0 &&
+          currentState.edges.length === 0
+        ) {
+          return;
+        }
+
+        draftService.saveDraft(effectiveFlowId, currentState).catch(() => {
+          // Ignore errors on unload
+        });
+      }
+    };
+
+    window.addEventListener("beforeunload", handleBeforeUnload);
+    return () => {
+      window.removeEventListener("beforeunload", handleBeforeUnload);
+    };
+  }, [getEffectiveFlowId, getCurrentState]);
+
+  useEffect(() => {
+    return () => {
+      if (saveTimeoutRef.current) {
+        clearTimeout(saveTimeoutRef.current);
+      }
+      if (isDirtyRef.current) {
+        saveDraft();
+      }
+    };
+  }, [saveDraft]);
+
+  useEffect(() => {
+    draftService.cleanupExpired().catch((error) => {
+      console.error(
+        "[DraftPersistence] Failed to cleanup expired drafts:",
+        error,
+      );
+    });
+  }, []);
+
+  const checkForDraft = useCallback(async () => {
+    const effectiveFlowId = flowID || getTempFlowId();
+
+    if (!effectiveFlowId) {
+      return;
+    }
+
+    try {
+      const draft = await draftService.loadDraft(effectiveFlowId);
+
+      if (!draft) {
+        return;
+      }
+
+      const currentNodes = useNodeStore.getState().nodes;
+      const currentEdges = useEdgeStore.getState().edges;
+
+      const isDifferent = draftService.isDraftDifferent(
+        draft,
+        currentNodes,
+        currentEdges,
+      );
+
+      if (isDifferent && (draft.nodes.length > 0 || draft.edges.length > 0)) {
+        const diff = calculateDraftDiff(
+          draft.nodes,
+          draft.edges,
+          currentNodes,
+          currentEdges,
+        );
+        setState({
+          isOpen: true,
+          draft,
+          diff,
+        });
+      } else {
+        await draftService.deleteDraft(effectiveFlowId);
+      }
+    } catch (error) {
+      console.error("[DraftRecovery] Failed to check for draft:", error);
+    }
+  }, [flowID]);
+
+  useEffect(() => {
+    if (isInitialLoadComplete && !hasCheckedForDraft.current) {
+      hasCheckedForDraft.current = true;
+      checkForDraft();
+    }
+  }, [isInitialLoadComplete, checkForDraft]);
+
+  useEffect(() => {
+    hasCheckedForDraft.current = false;
+    setState({
+      isOpen: false,
+      draft: null,
+      diff: null,
+    });
+  }, [flowID]);
+
+  const loadDraft = useCallback(async () => {
+    if (!state.draft) return;
+
+    const { draft } = state;
+
+    try {
+      useNodeStore.getState().setNodes(draft.nodes);
+      useEdgeStore.getState().setEdges(draft.edges);
+      draft.nodes.forEach((node) => {
+        useNodeStore.getState().syncHardcodedValuesWithHandleIds(node.id);
+      });
+
+      if (draft.nodeCounter !== undefined) {
+        useNodeStore.setState({ nodeCounter: draft.nodeCounter });
+      }
+
+      if (draft.graphSchemas) {
+        useGraphStore
+          .getState()
+          .setGraphSchemas(
+            draft.graphSchemas.input as Record<string, unknown> | null,
+            draft.graphSchemas.credentials as Record<string, unknown> | null,
+            draft.graphSchemas.output as Record<string, unknown> | null,
+          );
+      }
+
+      setTimeout(() => {
+        useHistoryStore.getState().initializeHistory();
+      }, 100);
+
+      await draftService.deleteDraft(draft.id);
+
+      setState({
+        isOpen: false,
+        draft: null,
+        diff: null,
+      });
+    } catch (error) {
+      console.error("[DraftRecovery] Failed to load draft:", error);
+    }
+  }, [state.draft]);
+
+  const discardDraft = useCallback(async () => {
+    if (!state.draft) {
+      setState({ isOpen: false, draft: null, diff: null });
+      return;
+    }
+
+    try {
+      await draftService.deleteDraft(state.draft.id);
+    } catch (error) {
+      console.error("[DraftRecovery] Failed to discard draft:", error);
+    }
+
+    setState({ isOpen: false, draft: null, diff: null });
+  }, [state.draft]);
+
+  return {
+    // Recovery popup props
+    isRecoveryOpen: state.isOpen,
+    savedAt: state.draft?.savedAt ?? 0,
+    nodeCount: state.draft?.nodes.length ?? 0,
+    edgeCount: state.draft?.edges.length ?? 0,
+    diff: state.diff,
+    loadDraft,
+    discardDraft,
+  };
+}

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/Flow/useFlow.ts

@@ -21,6 +21,7 @@ import { AgentExecutionStatus } from "@/app/api/__generated__/models/agentExecut
 export const useFlow = () => {
   const [isLocked, setIsLocked] = useState(false);
   const [hasAutoFramed, setHasAutoFramed] = useState(false);
+  const [isInitialLoadComplete, setIsInitialLoadComplete] = useState(false);
   const addNodes = useNodeStore(useShallow((state) => state.addNodes));
   const addLinks = useEdgeStore(useShallow((state) => state.addLinks));
   const updateNodeStatus = useNodeStore(
@@ -120,6 +121,14 @@ export const useFlow = () => {
     if (customNodes.length > 0) {
       useNodeStore.getState().setNodes([]);
       addNodes(customNodes);
+
+      // Sync hardcoded values with handle IDs.
+      // If a key–value field has a key without a value, the backend omits it from hardcoded values.
+      // But if a handleId exists for that key, it causes inconsistency.
+      // This ensures hardcoded values stay in sync with handle IDs.
+      customNodes.forEach((node) => {
+        useNodeStore.getState().syncHardcodedValuesWithHandleIds(node.id);
+      });
     }
   }, [customNodes, addNodes]);
 
@@ -174,11 +183,23 @@ export const useFlow = () => {
     if (customNodes.length > 0 && graph?.links) {
       const timer = setTimeout(() => {
         useHistoryStore.getState().initializeHistory();
+        // Mark initial load as complete after history is initialized
+        setIsInitialLoadComplete(true);
       }, 100);
       return () => clearTimeout(timer);
     }
   }, [customNodes, graph?.links]);
 
+  // Also mark as complete for new flows (no flowID) after a short delay
+  useEffect(() => {
+    if (!flowID && !isGraphLoading && !isBlocksLoading) {
+      const timer = setTimeout(() => {
+        setIsInitialLoadComplete(true);
+      }, 200);
+      return () => clearTimeout(timer);
+    }
+  }, [flowID, isGraphLoading, isBlocksLoading]);
+
   useEffect(() => {
     return () => {
       useNodeStore.getState().setNodes([]);
@@ -217,6 +238,7 @@ export const useFlow = () => {
 
   useEffect(() => {
     setHasAutoFramed(false);
+    setIsInitialLoadComplete(false);
   }, [flowID, flowVersion]);
 
   // Drag and drop block from block menu
@@ -253,6 +275,7 @@ export const useFlow = () => {
 
   return {
     isFlowContentLoading: isGraphLoading || isBlocksLoading,
+    isInitialLoadComplete,
     onDragOver,
     onDrop,
     isLocked,

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/edges/useCustomEdge.ts

@@ -1,12 +1,17 @@
-import { Connection as RFConnection, EdgeChange } from "@xyflow/react";
+import {
+  Connection as RFConnection,
+  EdgeChange,
+  applyEdgeChanges,
+} from "@xyflow/react";
 import { useEdgeStore } from "@/app/(platform)/build/stores/edgeStore";
 import { useCallback } from "react";
 import { useNodeStore } from "../../../stores/nodeStore";
+import { CustomEdge } from "./CustomEdge";
 
 export const useCustomEdge = () => {
   const edges = useEdgeStore((s) => s.edges);
   const addEdge = useEdgeStore((s) => s.addEdge);
-  const removeEdge = useEdgeStore((s) => s.removeEdge);
+  const setEdges = useEdgeStore((s) => s.setEdges);
 
   const onConnect = useCallback(
     (conn: RFConnection) => {
@@ -45,14 +50,10 @@ export const useCustomEdge = () => {
   );
 
   const onEdgesChange = useCallback(
-    (changes: EdgeChange[]) => {
-      changes.forEach((change) => {
-        if (change.type === "remove") {
-          removeEdge(change.id);
-        }
-      });
+    (changes: EdgeChange<CustomEdge>[]) => {
+      setEdges(applyEdgeChanges(changes, edges));
     },
-    [removeEdge],
+    [edges, setEdges],
   );
 
   return { edges, onConnect, onEdgesChange };

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/handlers/NodeHandle.tsx

@@ -1,26 +1,32 @@
 import { CircleIcon } from "@phosphor-icons/react";
 import { Handle, Position } from "@xyflow/react";
+import { useEdgeStore } from "../../../stores/edgeStore";
+import { cleanUpHandleId } from "@/components/renderers/InputRenderer/helpers";
+import { cn } from "@/lib/utils";
 
-const NodeHandle = ({
+const InputNodeHandle = ({
   handleId,
-  isConnected,
-  side,
+  nodeId,
 }: {
   handleId: string;
-  isConnected: boolean;
-  side: "left" | "right";
+  nodeId: string;
 }) => {
+  const cleanedHandleId = cleanUpHandleId(handleId);
+  const isInputConnected = useEdgeStore((state) =>
+    state.isInputConnected(nodeId ?? "", cleanedHandleId),
+  );
+
   return (
     <Handle
-      type={side === "left" ? "target" : "source"}
-      position={side === "left" ? Position.Left : Position.Right}
-      id={handleId}
-      className={side === "left" ? "-ml-4 mr-2" : "-mr-2 ml-2"}
+      type={"target"}
+      position={Position.Left}
+      id={cleanedHandleId}
+      className={"-ml-6 mr-2"}
     >
       <div className="pointer-events-none">
         <CircleIcon
           size={16}
-          weight={isConnected ? "fill" : "duotone"}
+          weight={isInputConnected ? "fill" : "duotone"}
           className={"text-gray-400 opacity-100"}
         />
       </div>
@@ -28,4 +34,35 @@ const NodeHandle = ({
   );
 };
 
-export default NodeHandle;
+const OutputNodeHandle = ({
+  field_name,
+  nodeId,
+  hexColor,
+}: {
+  field_name: string;
+  nodeId: string;
+  hexColor: string;
+}) => {
+  const isOutputConnected = useEdgeStore((state) =>
+    state.isOutputConnected(nodeId, field_name),
+  );
+  return (
+    <Handle
+      type={"source"}
+      position={Position.Right}
+      id={field_name}
+      className={"-mr-2 ml-2"}
+    >
+      <div className="pointer-events-none">
+        <CircleIcon
+          size={16}
+          weight={"duotone"}
+          color={isOutputConnected ? hexColor : "gray"}
+          className={cn("text-gray-400 opacity-100")}
+        />
+      </div>
+    </Handle>
+  );
+};
+
+export { InputNodeHandle, OutputNodeHandle };

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/handlers/helpers.ts

@@ -1,31 +1,4 @@
-/**
- * Handle ID Types for different input structures
- *
- * Examples:
- * SIMPLE: "message"
- * NESTED: "config.api_key"
- * ARRAY: "items_$_0", "items_$_1"
- * KEY_VALUE: "headers_#_Authorization", "params_#_limit"
- *
- * Note: All handle IDs are sanitized to remove spaces and special characters.
- * Spaces become underscores, and special characters are removed.
- * Example: "user name" becomes "user_name", "email@domain.com" becomes "emaildomaincom"
- */
-export enum HandleIdType {
-  SIMPLE = "SIMPLE",
-  NESTED = "NESTED",
-  ARRAY = "ARRAY",
-  KEY_VALUE = "KEY_VALUE",
-}
-
-const fromRjsfId = (id: string): string => {
-  if (!id) return "";
-  const parts = id.split("_");
-  const filtered = parts.filter(
-    (p) => p !== "root" && p !== "properties" && p.length > 0,
-  );
-  return filtered.join("_") || "";
-};
+// Here we are handling single level of nesting, if need more in future then i will update it
 
 const sanitizeForHandleId = (str: string): string => {
   if (!str) return "";
@@ -38,51 +11,53 @@ const sanitizeForHandleId = (str: string): string => {
     .replace(/^_|_$/g, ""); // Remove leading/trailing underscores
 };
 
-export const generateHandleId = (
+const cleanTitleId = (id: string): string => {
+  if (!id) return "";
+
+  if (id.endsWith("_title")) {
+    id = id.slice(0, -6);
+  }
+  const parts = id.split("_");
+  const filtered = parts.filter(
+    (p) => p !== "root" && p !== "properties" && p.length > 0,
+  );
+  const filtered_id = filtered.join("_") || "";
+  return filtered_id;
+};
+
+export const generateHandleIdFromTitleId = (
   fieldKey: string,
-  nestedValues: string[] = [],
-  type: HandleIdType = HandleIdType.SIMPLE,
+  {
+    isObjectProperty,
+    isAdditionalProperty,
+    isArrayItem,
+  }: {
+    isArrayItem?: boolean;
+    isObjectProperty?: boolean;
+    isAdditionalProperty?: boolean;
+  } = {
+    isArrayItem: false,
+    isObjectProperty: false,
+    isAdditionalProperty: false,
+  },
 ): string => {
   if (!fieldKey) return "";
 
-  fieldKey = fromRjsfId(fieldKey);
-  fieldKey = sanitizeForHandleId(fieldKey);
+  const filteredKey = cleanTitleId(fieldKey);
+  if (isAdditionalProperty || isArrayItem) {
+    return filteredKey;
+  }
+  const cleanedKey = sanitizeForHandleId(filteredKey);
 
-  if (type === HandleIdType.SIMPLE || nestedValues.length === 0) {
-    return fieldKey;
+  if (isObjectProperty) {
+    // "config_api_key" -> "config.api_key"
+    const parts = cleanedKey.split("_");
+    if (parts.length >= 2) {
+      const baseName = parts[0];
+      const propertyName = parts.slice(1).join("_");
+      return `${baseName}.${propertyName}`;
+    }
   }
 
-  const sanitizedNestedValues = nestedValues.map((value) =>
-    sanitizeForHandleId(value),
-  );
-
-  switch (type) {
-    case HandleIdType.NESTED:
-      return [fieldKey, ...sanitizedNestedValues].join(".");
-
-    case HandleIdType.ARRAY:
-      return [fieldKey, ...sanitizedNestedValues].join("_$_");
-
-    case HandleIdType.KEY_VALUE:
-      return [fieldKey, ...sanitizedNestedValues].join("_#_");
-
-    default:
-      return fieldKey;
-  }
-};
-
-export const parseKeyValueHandleId = (
-  handleId: string,
-  type: HandleIdType,
-): string => {
-  if (type === HandleIdType.KEY_VALUE) {
-    return handleId.split("_#_")[1];
-  } else if (type === HandleIdType.ARRAY) {
-    return handleId.split("_$_")[1];
-  } else if (type === HandleIdType.NESTED) {
-    return handleId.split(".")[1];
-  } else if (type === HandleIdType.SIMPLE) {
-    return handleId.split("_")[1];
-  }
-  return "";
+  return cleanedKey;
 };

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/nodes/CustomNode/CustomNode.tsx

@@ -10,7 +10,7 @@ import { NodeExecutionResult } from "@/app/api/__generated__/models/nodeExecutio
 import { NodeContainer } from "./components/NodeContainer";
 import { NodeHeader } from "./components/NodeHeader";
 import { FormCreator } from "../FormCreator";
-import { preprocessInputSchema } from "@/components/renderers/input-renderer/utils/input-schema-pre-processor";
+import { preprocessInputSchema } from "@/components/renderers/InputRenderer/utils/input-schema-pre-processor";
 import { OutputHandler } from "../OutputHandler";
 import { NodeAdvancedToggle } from "./components/NodeAdvancedToggle";
 import { NodeDataRenderer } from "./components/NodeOutput/NodeOutput";
@@ -99,7 +99,7 @@ export const CustomNode: React.FC<NodeProps<CustomNode>> = React.memo(
             nodeId={nodeId}
             uiType={data.uiType}
             className={cn(
-              "bg-white pr-6",
+              "bg-white px-4",
               isWebhook && "pointer-events-none opacity-50",
             )}
             showHandles={showHandles}

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/nodes/CustomNode/components/NodeAdvancedToggle.tsx

@@ -8,7 +8,7 @@ export const NodeAdvancedToggle = ({ nodeId }: { nodeId: string }) => {
   );
   const setShowAdvanced = useNodeStore((state) => state.setShowAdvanced);
   return (
-    <div className="flex items-center justify-between gap-2 rounded-b-xlarge border-t border-slate-200/50 bg-white px-5 py-3.5">
+    <div className="flex items-center justify-between gap-2 rounded-b-xlarge border-t border-zinc-200 bg-white px-5 py-3.5">
       <Text variant="body" className="font-medium text-slate-700">
         Advanced
       </Text>

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/nodes/CustomNode/components/NodeContainer.tsx

@@ -22,7 +22,7 @@ export const NodeContainer = ({
   return (
     <div
       className={cn(
-        "z-12 max-w-[370px] rounded-xlarge ring-1 ring-slate-200/60",
+        "z-12 w-[350px] rounded-xlarge ring-1 ring-slate-200/60",
         selected && "shadow-lg ring-2 ring-slate-200",
         status && nodeStyleBasedOnStatus[status],
         hasErrors ? nodeStyleBasedOnStatus[AgentExecutionStatus.FAILED] : "",

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/nodes/CustomNode/components/NodeHeader.tsx

@@ -23,7 +23,9 @@ export const NodeHeader = ({
   const updateNodeData = useNodeStore((state) => state.updateNodeData);
   const title = (data.metadata?.customized_name as string) || data.title;
   const [isEditingTitle, setIsEditingTitle] = useState(false);
-  const [editedTitle, setEditedTitle] = useState(title);
+  const [editedTitle, setEditedTitle] = useState(
+    beautifyString(title).replace("Block", "").trim(),
+  );
 
   const handleTitleEdit = () => {
     updateNodeData(nodeId, {
@@ -41,7 +43,7 @@ export const NodeHeader = ({
   };
 
   return (
-    <div className="flex h-auto flex-col gap-1 rounded-xlarge border-b border-slate-200/50 bg-gradient-to-r from-slate-50/80 to-white/90 px-4 py-4 pt-3">
+    <div className="flex h-auto flex-col gap-1 rounded-xlarge border-b border-zinc-200 bg-gradient-to-r from-slate-50/80 to-white/90 px-4 py-4 pt-3">
       {/* Title row with context menu */}
       <div className="flex items-start justify-between gap-2">
         <div className="flex min-w-0 flex-1 items-center gap-2">
@@ -68,12 +70,12 @@ export const NodeHeader = ({
                   <TooltipTrigger asChild>
                     <div>
                       <Text variant="large-semibold" className="line-clamp-1">
-                        {beautifyString(title)}
+                        {beautifyString(title).replace("Block", "").trim()}
                       </Text>
                     </div>
                   </TooltipTrigger>
                   <TooltipContent>
-                    <p>{beautifyString(title)}</p>
+                    <p>{beautifyString(title).replace("Block", "").trim()}</p>
                   </TooltipContent>
                 </Tooltip>
               </TooltipProvider>

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/nodes/CustomNode/components/NodeOutput/NodeOutput.tsx

@@ -23,7 +23,7 @@ export const NodeDataRenderer = ({ nodeId }: { nodeId: string }) => {
   }
 
   return (
-    <div className="flex flex-col gap-3 rounded-b-xl border-t border-slate-200/50 px-4 py-4">
+    <div className="flex flex-col gap-3 rounded-b-xl border-t border-zinc-200 px-4 py-4">
       <div className="flex items-center justify-between">
         <Text variant="body-medium" className="!font-semibold text-slate-700">
           Node Output

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/nodes/CustomNode/components/NodeOutput/components/ContentRenderer.tsx

@@ -1,7 +1,7 @@
 "use client";
 
-import type { OutputMetadata } from "@/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/selected-views/OutputRenderers";
-import { globalRegistry } from "@/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/selected-views/OutputRenderers";
+import type { OutputMetadata } from "@/components/contextual/OutputRenderers";
+import { globalRegistry } from "@/components/contextual/OutputRenderers";
 
 export const TextRenderer: React.FC<{
   value: any;

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/nodes/CustomNode/components/NodeOutput/components/NodeDataViewer/NodeDataViewer.tsx

@@ -1,7 +1,3 @@
-import {
-  OutputActions,
-  OutputItem,
-} from "@/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/selected-views/OutputRenderers";
 import { ScrollArea } from "@/components/__legacy__/ui/scroll-area";
 import { Button } from "@/components/atoms/Button/Button";
 import { Text } from "@/components/atoms/Text/Text";
@@ -11,6 +7,10 @@ import {
   TooltipProvider,
   TooltipTrigger,
 } from "@/components/atoms/Tooltip/BaseTooltip";
+import {
+  OutputActions,
+  OutputItem,
+} from "@/components/contextual/OutputRenderers";
 import { Dialog } from "@/components/molecules/Dialog/Dialog";
 import { beautifyString } from "@/lib/utils";
 import {

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/nodes/CustomNode/components/NodeOutput/components/NodeDataViewer/useNodeDataViewer.ts

@@ -1,6 +1,6 @@
-import type { OutputMetadata } from "@/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/selected-views/OutputRenderers";
-import { globalRegistry } from "@/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/selected-views/OutputRenderers";
-import { downloadOutputs } from "@/app/(platform)/library/agents/[id]/components/NewAgentLibraryView/components/selected-views/OutputRenderers/utils/download";
+import type { OutputMetadata } from "@/components/contextual/OutputRenderers";
+import { globalRegistry } from "@/components/contextual/OutputRenderers";
+import { downloadOutputs } from "@/components/contextual/OutputRenderers/utils/download";
 import { useToast } from "@/components/molecules/Toast/use-toast";
 import { beautifyString } from "@/lib/utils";
 import React, { useMemo, useState } from "react";

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/nodes/CustomNode/components/StickyNoteBlock.tsx

@@ -1,6 +1,6 @@
 import { useMemo } from "react";
 import { FormCreator } from "../../FormCreator";
-import { preprocessInputSchema } from "@/components/renderers/input-renderer/utils/input-schema-pre-processor";
+import { preprocessInputSchema } from "@/components/renderers/InputRenderer/utils/input-schema-pre-processor";
 import { CustomNodeData } from "../CustomNode";
 import { Text } from "@/components/atoms/Text/Text";
 import { cn } from "@/lib/utils";

autogpt_platform/frontend/src/app/(platform)/build/components/FlowEditor/nodes/CustomNode/components/WebhookDisclaimer.tsx

@@ -1,10 +1,10 @@
-import { Alert, AlertDescription } from "@/components/molecules/Alert/Alert";
-import { Text } from "@/components/atoms/Text/Text";
-import Link from "next/link";
 import { useGetV2GetLibraryAgentByGraphId } from "@/app/api/__generated__/endpoints/library/library";
 import { LibraryAgent } from "@/app/api/__generated__/models/libraryAgent";
-import { useQueryStates, parseAsString } from "nuqs";
-import { isValidUUID } from "@/app/(platform)/chat/helpers";
+import { Text } from "@/components/atoms/Text/Text";
+import { isValidUUID } from "@/components/contextual/Chat/helpers";
+import { Alert, AlertDescription } from "@/components/molecules/Alert/Alert";
+import Link from "next/link";
+import { parseAsString, useQueryStates } from "nuqs";
 
 export const WebhookDisclaimer = ({ nodeId }: { nodeId: string }) => {
   const [{ flowID }] = useQueryStates({