fix(autopilot): Include input schema in agent discovery + validate unknown fields

Fixes OPEN-2980

Part 1: Include input schema in find_agent and find_library_agent responses
- Added 'inputs' field to AgentInfo model
- Fetch graph input_schema for marketplace agents via store details
- Fetch graph input_schema for library agents via graph_id
- Graceful fallback to None if schema fetch fails

Part 2: Reject unknown input fields in run_agent
- Check provided inputs against valid schema fields
- Return AgentDetailsResponse with error if unknown fields detected
- Prevents silent failures where agents run with defaults instead of user inputs

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

@@ -1883,127 +1883,84 @@ async def _generate_llm_continuation(
     This is called by background tasks to continue the conversation
     after a tool result is saved. The response is saved to the database
     so users see it when they refresh or poll.
-
-    Includes retry logic with exponential backoff for transient API errors.
     """
-    retry_count = 0
-    max_retries = MAX_RETRIES
-    last_error: Exception | None = None
+    try:
+        # Load fresh session from DB (bypass cache to get the updated tool result)
+        await invalidate_session_cache(session_id)
+        session = await get_chat_session(session_id, user_id)
+        if not session:
+            logger.error(f"Session {session_id} not found for LLM continuation")
+            return
 
-    while retry_count <= max_retries:
-        try:
-            # Load fresh session from DB (bypass cache to get the updated tool result)
-            await invalidate_session_cache(session_id)
-            session = await get_chat_session(session_id, user_id)
-            if not session:
-                logger.error(f"Session {session_id} not found for LLM continuation")
+        # Build system prompt
+        system_prompt, _ = await _build_system_prompt(user_id)
+
+        # Build messages in OpenAI format
+        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
+
+        # Build extra_body for tracing
+        extra_body: dict[str, Any] = {
+            "posthogProperties": {
+                "environment": settings.config.app_env.value,
+            },
+        }
+        if user_id:
+            extra_body["user"] = user_id[:128]
+            extra_body["posthogDistinctId"] = user_id
+        if session_id:
+            extra_body["session_id"] = session_id[:128]
+
+        # Make non-streaming LLM call (no tools - just text response)
+        from typing import cast
+
+        from openai.types.chat import ChatCompletionMessageParam
+
+        # No tools parameter = text-only response (no tool calls)
+        response = await client.chat.completions.create(
+            model=config.model,
+            messages=cast(list[ChatCompletionMessageParam], messages),
+            extra_body=extra_body,
+        )
+
+        if response.choices and response.choices[0].message.content:
+            assistant_content = response.choices[0].message.content
+
+            # Reload session from DB to avoid race condition with user messages
+            # that may have been sent while we were generating the LLM response
+            fresh_session = await get_chat_session(session_id, user_id)
+            if not fresh_session:
+                logger.error(
+                    f"Session {session_id} disappeared during LLM continuation"
+                )
                 return
 
-            # Build system prompt
-            system_prompt, _ = await _build_system_prompt(user_id)
-
-            # Build messages in OpenAI format
-            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
-
-            # Build extra_body for tracing
-            extra_body: dict[str, Any] = {
-                "posthogProperties": {
-                    "environment": settings.config.app_env.value,
-                },
-            }
-            if user_id:
-                extra_body["user"] = user_id[:128]
-                extra_body["posthogDistinctId"] = user_id
-            if session_id:
-                extra_body["session_id"] = session_id[:128]
-
-            from typing import cast
-
-            from openai.types.chat import ChatCompletionMessageParam
-
-            # Include tools with tool_choice="none" to allow the provider to validate
-            # tool interactions in the message history without allowing new tool calls.
-            # Some providers (especially Anthropic via OpenRouter) require the tools
-            # schema to be present when the conversation contains tool_use/tool_result.
-            response = await client.chat.completions.create(
-                model=config.model,
-                messages=cast(list[ChatCompletionMessageParam], messages),
-                tools=tools,
-                tool_choice="none",
-                extra_body=extra_body,
+            # Save assistant message to database
+            assistant_message = ChatMessage(
+                role="assistant",
+                content=assistant_content,
             )
+            fresh_session.messages.append(assistant_message)
 
-            if response.choices and response.choices[0].message.content:
-                assistant_content = response.choices[0].message.content
+            # Save to database (not cache) to persist the response
+            await upsert_chat_session(fresh_session)
 
-                # Reload session from DB to avoid race condition with user messages
-                # that may have been sent while we were generating the LLM response.
-                # Invalidate cache first to ensure we get fresh data from the database.
-                await invalidate_session_cache(session_id)
-                fresh_session = await get_chat_session(session_id, user_id)
-                if not fresh_session:
-                    logger.error(
-                        f"Session {session_id} disappeared during LLM continuation"
-                    )
-                    return
+            # Invalidate cache so next poll/refresh gets fresh data
+            await invalidate_session_cache(session_id)
 
-                # Save assistant message to database
-                assistant_message = ChatMessage(
-                    role="assistant",
-                    content=assistant_content,
-                )
-                fresh_session.messages.append(assistant_message)
+            logger.info(
+                f"Generated LLM continuation for session {session_id}, "
+                f"response length: {len(assistant_content)}"
+            )
+        else:
+            logger.warning(f"LLM continuation returned empty response for {session_id}")
 
-                # Save to database (not cache) to persist the response
-                await upsert_chat_session(fresh_session)
-
-                # Invalidate cache so next poll/refresh gets fresh data
-                await invalidate_session_cache(session_id)
-
-                logger.info(
-                    f"Generated LLM continuation for session {session_id}, "
-                    f"response length: {len(assistant_content)}"
-                )
-                return  # Success - exit the retry loop
-            else:
-                logger.warning(
-                    f"LLM continuation returned empty response for {session_id}"
-                )
-                return  # Empty response is not retryable
-
-        except Exception as e:
-            last_error = e
-            if _is_retryable_error(e) and retry_count < max_retries:
-                retry_count += 1
-                delay = min(
-                    BASE_DELAY_SECONDS * (2 ** (retry_count - 1)),
-                    MAX_DELAY_SECONDS,
-                )
-                logger.warning(
-                    f"Retryable error in LLM continuation for {session_id}: {e!s}. "
-                    f"Retrying in {delay:.1f}s (attempt {retry_count}/{max_retries})"
-                )
-                await asyncio.sleep(delay)
-                continue
-            else:
-                # Non-retryable error or max retries exceeded
-                logger.error(
-                    f"Failed to generate LLM continuation for {session_id}: {e!s}",
-                    exc_info=True,
-                )
-                break
-
-    # If we get here, all retries failed
-    if last_error:
-        logger.error(
-            f"LLM continuation failed after {retry_count} retries for {session_id}. "
-            f"Last error: {last_error!s}"
-        )
+    except Exception as e:
+        logger.error(f"Failed to generate LLM continuation: {e}", exc_info=True)

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

@@ -1,13 +1,11 @@
 """Shared agent search functionality for find_agent and find_library_agent tools."""
 
-import asyncio
 import logging
-from typing import Literal
+from typing import Any, Literal
 
 from backend.api.features.library import db as library_db
 from backend.api.features.store import db as store_db
 from backend.data import graph as graph_db
-from backend.data.graph import GraphModel
 from backend.util.exceptions import DatabaseError, NotFoundError
 
 from .models import (
@@ -17,13 +15,50 @@ from .models import (
     NoResultsResponse,
     ToolResponseBase,
 )
-from .utils import fetch_graph_from_store_slug
 
 logger = logging.getLogger(__name__)
 
 SearchSource = Literal["marketplace", "library"]
 
 
+async def _fetch_input_schema_for_store_agent(
+    creator: str, slug: str
+) -> dict[str, Any] | None:
+    """Fetch input schema for a marketplace agent."""
+    try:
+        store_agent = await store_db.get_store_agent_details(creator, slug)
+        graph_meta = await store_db.get_available_graph(
+            store_agent.store_listing_version_id
+        )
+        graph = await graph_db.get_graph(
+            graph_id=graph_meta.id,
+            version=graph_meta.version,
+            user_id=None,
+            include_subgraphs=False,
+        )
+        return graph.input_schema if graph else None
+    except Exception as e:
+        logger.warning(f"Failed to fetch input schema for {creator}/{slug}: {e}")
+        return None
+
+
+async def _fetch_input_schema_for_library_agent(
+    graph_id: str, user_id: str
+) -> dict[str, Any] | None:
+    """Fetch input schema for a library agent."""
+    try:
+        graph = await graph_db.get_graph(
+            graph_id=graph_id,
+            version=None,  # Get latest version
+            user_id=user_id,
+            include_subgraphs=False,
+        )
+        return graph.input_schema if graph else None
+    except Exception as e:
+        logger.warning(f"Failed to fetch input schema for graph {graph_id}: {e}")
+        return None
+
+
 async def search_agents(
     query: str,
     source: SearchSource,
@@ -58,28 +93,11 @@ async def search_agents(
         if source == "marketplace":
             logger.info(f"Searching marketplace for: {query}")
             results = await store_db.get_store_agents(search_query=query, page_size=5)
-
-            # Fetch all graphs in parallel for better performance
-            async def fetch_marketplace_graph(
-                creator: str, slug: str
-            ) -> GraphModel | None:
-                try:
-                    graph, _ = await fetch_graph_from_store_slug(creator, slug)
-                    return graph
-                except Exception as e:
-                    logger.warning(
-                        f"Failed to fetch input schema for {creator}/{slug}: {e}"
-                    )
-                    return None
-
-            graphs = await asyncio.gather(
-                *(
-                    fetch_marketplace_graph(agent.creator, agent.slug)
-                    for agent in results.agents
+            for agent in results.agents:
+                # Fetch input schema for the agent
+                input_schema = await _fetch_input_schema_for_store_agent(
+                    agent.creator, agent.slug
                 )
-            )
-
-            for agent, graph in zip(results.agents, graphs):
                 agents.append(
                     AgentInfo(
                         id=f"{agent.creator}/{agent.slug}",
@@ -92,7 +110,7 @@ async def search_agents(
                         rating=agent.rating,
                         runs=agent.runs,
                         is_featured=False,
-                        inputs=graph.input_schema if graph else None,
+                        inputs=input_schema,
                     )
                 )
         else:  # library
@@ -102,32 +120,11 @@ async def search_agents(
                 search_term=query,
                 page_size=10,
             )
-
-            # Fetch all graphs in parallel for better performance
-            # (list_library_agents doesn't include nodes for performance)
-            async def fetch_library_graph(
-                graph_id: str, graph_version: int
-            ) -> GraphModel | None:
-                try:
-                    return await graph_db.get_graph(
-                        graph_id=graph_id,
-                        version=graph_version,
-                        user_id=user_id,
-                    )
-                except Exception as e:
-                    logger.warning(
-                        f"Failed to fetch input schema for graph {graph_id}: {e}"
-                    )
-                    return None
-
-            graphs = await asyncio.gather(
-                *(
-                    fetch_library_graph(agent.graph_id, agent.graph_version)
-                    for agent in results.agents
+            for agent in results.agents:
+                # Fetch input schema for the agent
+                input_schema = await _fetch_input_schema_for_library_agent(
+                    agent.graph_id, user_id  # type: ignore[arg-type]
                 )
-            )
-
-            for agent, graph in zip(results.agents, graphs):
                 agents.append(
                     AgentInfo(
                         id=agent.id,
@@ -141,7 +138,7 @@ async def search_agents(
                         has_external_trigger=agent.has_external_trigger,
                         new_output=agent.new_output,
                         graph_id=agent.graph_id,
-                        inputs=graph.input_schema if graph else None,
+                        inputs=input_schema,
                     )
                 )
         logger.info(f"Found {len(agents)} agents in {source}")

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

@@ -32,8 +32,6 @@ class ResponseType(str, Enum):
     OPERATION_STARTED = "operation_started"
     OPERATION_PENDING = "operation_pending"
     OPERATION_IN_PROGRESS = "operation_in_progress"
-    # Input validation
-    INPUT_VALIDATION_ERROR = "input_validation_error"
 
 
 # Base response model
@@ -194,20 +192,6 @@ class ErrorResponse(ToolResponseBase):
     details: dict[str, Any] | None = None
 
 
-class InputValidationErrorResponse(ToolResponseBase):
-    """Response when run_agent receives unknown input fields."""
-
-    type: ResponseType = ResponseType.INPUT_VALIDATION_ERROR
-    unrecognized_fields: list[str] = Field(
-        description="List of input field names that were not recognized"
-    )
-    inputs: dict[str, Any] = Field(
-        description="The agent's valid input schema for reference"
-    )
-    graph_id: str | None = None
-    graph_version: int | None = None
-
-
 # Agent output models
 class ExecutionOutputInfo(BaseModel):
     """Summary of a single execution's outputs."""

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

@@ -30,7 +30,6 @@ from .models import (
     ErrorResponse,
     ExecutionOptions,
     ExecutionStartedResponse,
-    InputValidationErrorResponse,
     SetupInfo,
     SetupRequirementsResponse,
     ToolResponseBase,
@@ -274,22 +273,6 @@ class RunAgentTool(BaseTool):
             input_properties = graph.input_schema.get("properties", {})
             required_fields = set(graph.input_schema.get("required", []))
             provided_inputs = set(params.inputs.keys())
-            valid_fields = set(input_properties.keys())
-
-            # Check for unknown input fields
-            unrecognized_fields = provided_inputs - valid_fields
-            if unrecognized_fields:
-                return InputValidationErrorResponse(
-                    message=(
-                        f"Unknown input field(s) provided: {', '.join(sorted(unrecognized_fields))}. "
-                        f"Agent was not executed. Please use the correct field names from the schema."
-                    ),
-                    session_id=session_id,
-                    unrecognized_fields=sorted(unrecognized_fields),
-                    inputs=graph.input_schema,
-                    graph_id=graph.id,
-                    graph_version=graph.version,
-                )
 
             # If agent has inputs but none were provided AND use_defaults is not set,
             # always show what's available first so user can decide
@@ -327,6 +310,26 @@ class RunAgentTool(BaseTool):
                     graph_version=graph.version,
                 )
 
+            # Check for unknown input fields - reject to prevent silent failures
+            valid_fields = set(input_properties.keys())
+            unknown_fields = provided_inputs - valid_fields
+            if unknown_fields:
+                credentials = extract_credentials_from_schema(
+                    graph.credentials_input_schema
+                )
+                return AgentDetailsResponse(
+                    message=(
+                        f"Unknown input field(s) provided: {', '.join(sorted(unknown_fields))}. "
+                        f"Agent was not executed. "
+                        f"Valid input fields are: {', '.join(sorted(valid_fields)) or 'none'}."
+                    ),
+                    session_id=session_id,
+                    agent=self._build_agent_details(graph, credentials),
+                    user_authenticated=True,
+                    graph_id=graph.id,
+                    graph_version=graph.version,
+                )
+
             # Step 4: Execute or Schedule
             if is_schedule:
                 return await self._schedule_agent(

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

@@ -402,42 +402,3 @@ async def test_run_agent_schedule_without_name(setup_test_data):
     # Should return error about missing schedule_name
     assert result_data.get("type") == "error"
     assert "schedule_name" in result_data["message"].lower()
-
-
-@pytest.mark.asyncio(loop_scope="session")
-async def test_run_agent_rejects_unknown_input_fields(setup_test_data):
-    """Test that run_agent returns input_validation_error for unknown input fields."""
-    user = setup_test_data["user"]
-    store_submission = setup_test_data["store_submission"]
-
-    tool = RunAgentTool()
-    agent_marketplace_id = f"{user.email.split('@')[0]}/{store_submission.slug}"
-    session = make_session(user_id=user.id)
-
-    # Execute with unknown input field names
-    response = await tool.execute(
-        user_id=user.id,
-        session_id=str(uuid.uuid4()),
-        tool_call_id=str(uuid.uuid4()),
-        username_agent_slug=agent_marketplace_id,
-        inputs={
-            "unknown_field": "some value",
-            "another_unknown": "another value",
-        },
-        session=session,
-    )
-
-    assert response is not None
-    assert hasattr(response, "output")
-    assert isinstance(response.output, str)
-    result_data = orjson.loads(response.output)
-
-    # Should return input_validation_error type with unrecognized fields
-    assert result_data.get("type") == "input_validation_error"
-    assert "unrecognized_fields" in result_data
-    assert set(result_data["unrecognized_fields"]) == {
-        "another_unknown",
-        "unknown_field",
-    }
-    assert "inputs" in result_data  # Contains the valid schema
-    assert "Agent was not executed" in result_data["message"]

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

@@ -4,8 +4,6 @@ import logging
 from collections import defaultdict
 from typing import Any
 
-from pydantic_core import PydanticUndefined
-
 from backend.api.features.chat.model import ChatSession
 from backend.data.block import get_block
 from backend.data.execution import ExecutionContext
@@ -75,22 +73,15 @@ class RunBlockTool(BaseTool):
         self,
         user_id: str,
         block: Any,
-        input_data: dict[str, Any] | None = None,
     ) -> tuple[dict[str, CredentialsMetaInput], list[CredentialsMetaInput]]:
         """
         Check if user has required credentials for a block.
 
-        Args:
-            user_id: User ID
-            block: Block to check credentials for
-            input_data: Input data for the block (used to determine provider via discriminator)
-
         Returns:
             tuple[matched_credentials, missing_credentials]
         """
         matched_credentials: dict[str, CredentialsMetaInput] = {}
         missing_credentials: list[CredentialsMetaInput] = []
-        input_data = input_data or {}
 
         # Get credential field info from block's input schema
         credentials_fields_info = block.input_schema.get_credentials_fields_info()
@@ -103,33 +94,14 @@ class RunBlockTool(BaseTool):
         available_creds = await creds_manager.store.get_all_creds(user_id)
 
         for field_name, field_info in credentials_fields_info.items():
-            effective_field_info = field_info
-            if field_info.discriminator and field_info.discriminator_mapping:
-                # Get discriminator from input, falling back to schema default
-                discriminator_value = input_data.get(field_info.discriminator)
-                if discriminator_value is None:
-                    field = block.input_schema.model_fields.get(
-                        field_info.discriminator
-                    )
-                    if field and field.default is not PydanticUndefined:
-                        discriminator_value = field.default
-
-                if (
-                    discriminator_value
-                    and discriminator_value in field_info.discriminator_mapping
-                ):
-                    effective_field_info = field_info.discriminate(discriminator_value)
-                    logger.debug(
-                        f"Discriminated provider for {field_name}: "
-                        f"{discriminator_value} -> {effective_field_info.provider}"
-                    )
-
+            # 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 effective_field_info.provider
-                    and cred.type in effective_field_info.supported_types
+                    if cred.provider in field_info.provider
+                    and cred.type in field_info.supported_types
                 ),
                 None,
             )
@@ -143,8 +115,8 @@ class RunBlockTool(BaseTool):
                 )
             else:
                 # Create a placeholder for the missing credential
-                provider = next(iter(effective_field_info.provider), "unknown")
-                cred_type = next(iter(effective_field_info.supported_types), "api_key")
+                provider = next(iter(field_info.provider), "unknown")
+                cred_type = next(iter(field_info.supported_types), "api_key")
                 missing_credentials.append(
                     CredentialsMetaInput(
                         id=field_name,
@@ -212,9 +184,10 @@ class RunBlockTool(BaseTool):
 
         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, input_data
+            user_id, block
         )
 
         if missing_credentials: