agents.exceptions.ModelBehaviorError: Invalid JSON when parsing

[redpintings]

When I use a support
response_format = {
'type': ' json_object'
}The model is in
When conducting agent testing, the official demo did not run smoothly. Error, please refer to the code below:

The system used： macos
Partial software versions：
openai 1.68.0
openai-agents 0.0.6
pycparser 2.22
pydantic 2.10.6

my code ：
`

-- coding: utf-8 --

@author : ysl

@file : main_bata.py

from agents import Agent, InputGuardrail,GuardrailFunctionOutput, Runner, OpenAIChatCompletionsModel
from pydantic import BaseModel
import asyncio
import os
from config.config import DOUBAO, ZHiPU
from openai import AsyncOpenAI
from agents import set_tracing_disabled

BASE_URL = ZHiPU.get("base_url") or ""
API_KEY = ZHiPU.get("api_key") or ""
MODEL_NAME = ZHiPU.get("model_name") or ""

if not BASE_URL or not API_KEY or not MODEL_NAME:
raise ValueError(
"Please set EXAMPLE_BASE_URL, EXAMPLE_API_KEY, EXAMPLE_MODEL_NAME via env var or code."
)

client = AsyncOpenAI(base_url=BASE_URL, api_key=API_KEY)
set_tracing_disabled(disabled=True)

class HomeworkOutput(BaseModel):
is_homework: bool
reasoning: str

guardrail_agent = Agent(
name="Guardrail check",
instructions="Check if the user is asking about homework.",
output_type=HomeworkOutput,
model=OpenAIChatCompletionsModel(
model=MODEL_NAME,
openai_client=client,
) # 使用豆包 LLM
)

math_tutor_agent = Agent(
name="Math Tutor",
handoff_description="Specialist agent for math questions",
instructions="You provide help with math problems. Explain your reasoning at each step and include examples",
model=OpenAIChatCompletionsModel(model=MODEL_NAME, openai_client=client) # 使用豆包 LLM
)

history_tutor_agent = Agent(
name="History Tutor",
handoff_description="Specialist agent for historical questions",
instructions="You provide assistance with historical queries. Explain important events and context clearly.",
model=OpenAIChatCompletionsModel(model=MODEL_NAME, openai_client=client) # 使用豆包 LLM
)

async def homework_guardrail(ctx, agent, input_data):
result = await Runner.run(guardrail_agent, input_data, context=ctx.context)
final_output = result.final_output_as(HomeworkOutput)
return GuardrailFunctionOutput(
output_info=final_output,
tripwire_triggered=not final_output.is_homework,
)

triage_agent = Agent(
name="Triage Agent",
instructions="You determine which agent to use based on the user's homework question",
handoffs=[history_tutor_agent, math_tutor_agent],
input_guardrails=[
InputGuardrail(guardrail_function=homework_guardrail),
],
model=OpenAIChatCompletionsModel(model=MODEL_NAME, openai_client=client) # 使用豆包 LLM
)

async def main():
result = await Runner.run(triage_agent, "who was the first president of the united states?")
print(result.final_output)

result = await Runner.run(triage_agent, "what is life")
print(result.final_output)

if name == "main":
asyncio.run(main())
`

Error:

`Traceback (most recent call last):
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/.venv/lib/python3.12/site-packages/agents/util/_json.py", line 20, in validate_json
validated = type_adapter.validate_json(json_str, experimental_allow_partial=partial_setting)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/.venv/lib/python3.12/site-packages/pydantic/type_adapter.py", line 446, in validate_json
return self.validator.validate_json(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
pydantic_core._pydantic_core.ValidationError: 1 validation error for HomeworkOutput
Invalid JSON: expected value at line 1 column 1 [type=json_invalid, input_value='The user is not asking a...t of the United States.', input_type=str]
For further information visit https://errors.pydantic.dev/2.10/v/json_invalid

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/main_bata.py", line 80, in
asyncio.run(main())
File "/usr/local/Cellar/python@3.12/3.12.8/Frameworks/Python.framework/Versions/3.12/lib/python3.12/asyncio/runners.py", line 194, in run
return runner.run(main)
^^^^^^^^^^^^^^^^
File "/usr/local/Cellar/python@3.12/3.12.8/Frameworks/Python.framework/Versions/3.12/lib/python3.12/asyncio/runners.py", line 118, in run
return self._loop.run_until_complete(task)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/Cellar/python@3.12/3.12.8/Frameworks/Python.framework/Versions/3.12/lib/python3.12/asyncio/base_events.py", line 686, in run_until_complete
return future.result()
^^^^^^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/main_bata.py", line 73, in main
result = await Runner.run(triage_agent, "who was the first president of the united states?")
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/.venv/lib/python3.12/site-packages/agents/run.py", line 210, in run
input_guardrail_results, turn_result = await asyncio.gather(
^^^^^^^^^^^^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/.venv/lib/python3.12/site-packages/agents/run.py", line 794, in _run_input_guardrails
result = await done
^^^^^^^^^^
File "/usr/local/Cellar/python@3.12/3.12.8/Frameworks/Python.framework/Versions/3.12/lib/python3.12/asyncio/tasks.py", line 631, in _wait_for_one
return f.result() # May raise f.exception().
^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/.venv/lib/python3.12/site-packages/agents/_run_impl.py", line 645, in run_single_input_guardrail
result = await guardrail.run(agent, input, context)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/.venv/lib/python3.12/site-packages/agents/guardrail.py", line 118, in run
output=await output,
^^^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/main_bata.py", line 55, in homework_guardrail
result = await Runner.run(guardrail_agent, input_data, context=ctx.context)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/.venv/lib/python3.12/site-packages/agents/run.py", line 210, in run
input_guardrail_results, turn_result = await asyncio.gather(
^^^^^^^^^^^^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/.venv/lib/python3.12/site-packages/agents/run.py", line 729, in _run_single_turn
return await cls._get_single_step_result_from_response(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/.venv/lib/python3.12/site-packages/agents/run.py", line 761, in _get_single_step_result_from_response
return await RunImpl.execute_tools_and_side_effects(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/.venv/lib/python3.12/site-packages/agents/_run_impl.py", line 265, in execute_tools_and_side_effects
final_output = output_schema.validate_json(potential_final_output_text)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/.venv/lib/python3.12/site-packages/agents/agent_output.py", line 91, in validate_json
validated = _json.validate_json(json_str, self._type_adapter, partial)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/ysl/Downloads/works/workstation/mental_health_agent/.venv/lib/python3.12/site-packages/agents/util/_json.py", line 29, in validate_json
raise ModelBehaviorError(
agents.exceptions.ModelBehaviorError: Invalid JSON when parsing The user is not asking about homework. They are asking a factual question about American history, specifically inquiring about the first president of the United States. for TypeAdapter(HomeworkOutput); 1 validation error for HomeworkOutput
Invalid JSON: expected value at line 1 column 1 [type=json_invalid, input_value='The user is not asking a...t of the United States.', input_type=str]
For further information visit https://errors.pydantic.dev/2.10/v/json_invalid

`

[redpintings]

There are other issues, such as when using a model that does not support “response_format”, its output may be text like this. If I specify in prompt that JSON output is required, then its output：
“”“
``json
{
"is_homework": false,
"reasoning": "The question asks for historical information about a known fact, which is not typically homework-related."
}

”“”

This form, so I feel that more adaptation should be done for this issue

[rm-openai]

If i understand correctly, you're using a non-OpenAI model and the model is producing invalid JSON.

Unfortunately there's not much we can do here. OpenAI structured outputs generally guarantee valid JSON. If you're not using OpenAI, then prompting is your main tool to improve the likelyhood of valid JSON.