langgraph-tool-calling

langgraph1.2.16/langchain-openai1.7.2

• 真实LangGraph1.2.16 StateGraph执行LLM→tools→LLM，模型产生add_items(17,25)，实际ToolMessage为42，随后最终回答，退出0；不是用预制AIMessage或FakeChatModel替代推理。 网关分支未执行：独立真实探针确认当前测试Key的Gemini组429后停止新增请求，没有伪造网关tool_call。
• 模型为本次隔离Ollama原生导入的明确别名tutorial-qwen3-proof；权重和原始请求/响应留档。
• 网关分支未发起：当前测试Key的Gemini组独立探针429后停止新增请求。

明确限制

• 只证明对应SDK和本地模型的本题任务，不证明QuickRouter公网账户、Claude或所有模型均可用。
• 本次小样例通过不能推断复杂工具、多轮高风险任务也可靠。
• 仅本地工具回路有真实成功，未验证公网工具协议，不伪造tool_call或网关成功。

原始输出

run-execution.json
{
  "exitCode": 0,
  "stdout": "{\n  \"langgraph\": \"1.2.16\",\n  \"messages\": [\n    {\n      \"content\": \"Use add_items exactly once for item totals, then answer with the numeric tool result.\",\n      \"additional_kwargs\": {},\n      \"response_metadata\": {},\n      \"type\": \"system\",\n      \"name\": null,\n      \"id\": \"a8e9f3ed-e27b-462d-8289-473926bcfb6c\"\n    },\n    {\n      \"content\": \"There are 17 blue items and 25 red items. Use add_items to calculate total.\",\n      \"additional_kwargs\": {},\n      \"response_metadata\": {},\n      \"type\": \"human\",\n      \"name\": null,\n      \"id\": \"3934e9b1-3558-4485-91ce-e14c36850201\"\n    },\n    {\n      \"content\": \"\",\n      \"additional_kwargs\": {\n        \"refusal\": null\n      },\n      \"response_metadata\": {\n        \"token_usage\": {\n          \"completion_tokens\": 203,\n          \"prompt_tokens\": 199,\n          \"total_tokens\": 402,\n          \"completion_tokens_details\": null,\n          \"prompt_tokens_details\": {\n            \"audio_tokens\": null,\n            \"cache_write_tokens\": null,\n            \"cached_tokens\": 3,\n            \"image_tokens\": null,\n            \"text_tokens\": null\n          }\n        },\n        \"model_provider\": \"openai\",\n        \"model_name\": \"tutorial-qwen3-proof\",\n        \"system_fingerprint\": \"fp_ollama\",\n        \"id\": \"chatcmpl-902\",\n        \"finish_reason\": \"tool_calls\",\n        \"logprobs\": null\n      },\n      \"type\": \"ai\",\n      \"name\": null,\n      \"id\": \"lc_run--01a16947-d0a0-7525-96a1-4c1481d2dae0-0\",\n      \"tool_calls\": [\n        {\n          \"name\": \"add_items\",\n          \"args\": {\n            \"blue\": 17,\n            \"red\": 25\n          },\n          \"id\": \"2n5xVPE2uWnxKdp8KGHzbxEAGNwyUehS\",\n          \"type\": \"tool_call\"\n        }\n      ],\n      \"invalid_tool_calls\": [],\n      \"usage_metadata\": {\n        \"input_tokens\": 199,\n        \"output_tokens\": 203,\n        \"total_tokens\": 402,\n        \"input_token_details\": {\n          \"cache_read\": 3\n        },\n        \"output_token_details\": {}\n      }\n    },\n    {\n      \"content\": \"42\",\n      \"additional_kwargs\": {},\n      \"response_metadata\": {},\n      \"type\": \"tool\",\n      \"name\": null,\n      \"id\": \"5db52c89-0161-4a07-ae07-2bd29984dc18\",\n      \"tool_call_id\": \"2n5xVPE2uWnxKdp8KGHzbxEAGNwyUehS\",\n      \"artifact\": null,\n      \"status\": \"success\"\n    },\n    {\n      \"content\": \"The total number of blue and red items is 42.\",\n      \"additional_kwargs\": {\n        \"refusal\": null\n      },\n      \"response_metadata\": {\n        \"token_usage\": {\n          \"completion_tokens\": 190,\n          \"prompt_tokens\": 242,\n          \"total_tokens\": 432,\n          \"completion_tokens_details\": null,\n          \"prompt_tokens_details\": {\n            \"audio_tokens\": null,\n            \"cache_write_tokens\": null,\n            \"cached_tokens\": 199,\n            \"image_tokens\": null,\n            \"text_tokens\": null\n          }\n        },\n        \"model_provider\": \"openai\",\n        \"model_name\": \"tutorial-qwen3-proof\",\n        \"system_fingerprint\": \"fp_ollama\",\n        \"id\": \"chatcmpl-801\",\n        \"finish_reason\": \"stop\",\n        \"logprobs\": null\n      },\n      \"type\": \"ai\",\n      \"name\": null,\n      \"id\": \"lc_run--01a16947-d7be-76b2-a78a-fd11e86bdc0d-0\",\n      \"tool_calls\": [],\n      \"invalid_tool_calls\": [],\n      \"usage_metadata\": {\n        \"input_tokens\": 242,\n        \"output_tokens\": 190,\n        \"total_tokens\": 432,\n        \"input_token_details\": {\n          \"cache_read\": 199\n        },\n        \"output_token_details\": {}\n      }\n    }\n  ]\n}\n",
  "stderr": "",
  "command": [
    "/Volumes/Mac Storage/Projects/quickrouter-tutorial30-proof-20261023/coding/.venv/bin/python",
    "run.py"
  ]
}