AI & Agent Dev Bug Sandbox Back to Radar

Agents Silently Fail To Produce Structured_response When Model Omits Tool Call

Using create_agent with response_format=ToolStrategy and a model returning an AIMessage without tool_calls results in a response missing the structured_response key instead of raising an error (handle_errors=False) or retrying (handle_errors=True), causing silent downstream failures.

highConfidence 92%Langchain Affected V1.2.13

Origin Analysis

The output handling conditional in langchain.agents.factory only processes ToolStrategy when output is an AIMessage and output.tool_calls is truthy. An AIMessage with no tool_calls bypasses all branches and is returned unchanged, skipping structured output validation/error handling.

Define a fake chat model that returns AIMessage(content='NOT a tool call') with no tool_calls. Bind tools and call create_agent(model=model, tools=[fake_tool], response_format=ToolStrategy(Parrot, handle_errors=False)). Invoke with a user message. The try/except does not catch StructuredOutputValidationError, and the response lacks the 'structured_response' key, causing assertion failure.

Fixing Code Block

0x7F 45 4C 46 02 01 01 000x90 12 AB 04 C1 7E 33 08
0xA4 1C 90 3E 77 0B D2 580x11 F6 6A 2C 80 19 DE 44
0x00 00 00 00 03 00 3E 000x01 00 B7 4D 55 90 0C 61
0x6C 69 62 63 2E 73 6F 2E0x36 00 8E 17 42 AA 05 F3
0xDE C0 DE 01 8B 45 FC 890xC7 5D 20 6F 91 03 E4 78
0x13 37 00 4F 9A 2D E1 700x08 B4 5C 22 71 D0 4A 16
0x48 8D 3D 22 01 00 00 E80x91 FF 0A 63 2E 57 B8 04
0x5A 91 C3 0E 44 70 1B 880x02 6F 39 C4 15 80 ED 27
0xFF 15 40 20 60 00 48 890xC3 90 7A 01 4C 28 B1 56
0x0F 1F 44 00 00 55 48 890xE5 C3 18 6D 92 0B 33 A7

Edge Case Audit

This change may cause additional retry loops for models that legitimately stop without making a structured output tool call, especially if handle_errors=True and the model repeatedly fails; ensure the agent has a retry limit or use handle_errors=False with explicit outer error handling. The new NoStructuredOutputError will propagate in async/concurrent agents and may require updating tests that currently allow missing structured_response. In LangGraph checkpoints/streaming, the extra ToolMessage may alter message flow. Rollback: remove the added elif and the new class/import; the existing ToolStrategy branch for tool_calls remains unchanged.