[BUG] Bedrock Converse validation failure when using CAMEL ChatAgent with PubMedToolkit function tools

[HelloJocelynLu]

Required prerequisites

• I have read the documentation https://camel-ai.github.io/camel/camel.html.
• I have searched the Issue Tracker and Discussions that this hasn't already been reported. (+1 or comment there if it has.)
• Consider asking first in a Discussion.

What version of camel are you using?

0.2.90a6

System information

OS: macOS (Darwin 25.3.0, arm64)
Describe how the library was installed: git clone from source, master branch
3.11.14 | packaged by conda-forge | (main, Oct 22 2025, 22:56:31) [Clang 19.1.7 ] darwin
0.2.90a6

Problem description

I am seeing a Bedrock Converse validation failure when using CAMEL ChatAgent with PubMedToolkit function tools.

Reproducible example code

The Python snippets:

import os
from pathlib import Path
from typing import Any

import boto3
from botocore.config import Config
from dotenv import load_dotenv

from camel.agents import ChatAgent
from camel.models import ModelFactory
from camel.toolkits import PubMedToolkit
from camel.types import ModelPlatformType


def build_bedrock_client() -> Any:
    gateway_url = os.getenv("AI_GATEWAY_URL") or os.getenv("BEDROCK_ENDPOINT_URL")
    api_key = os.getenv("AI_GATEWAY_KEY") or os.getenv("AWS_BEARER_TOKEN_BEDROCK")

    if not gateway_url or not api_key:
        load_dotenv(dotenv_path=Path(".env"))
        gateway_url = (
            gateway_url
            or os.getenv("AI_GATEWAY_URL")
            or os.getenv("BEDROCK_ENDPOINT_URL")
        )
        api_key = (
            api_key
            or os.getenv("AI_GATEWAY_KEY")
            or os.getenv("AWS_BEARER_TOKEN_BEDROCK")
        )

    if not gateway_url or not api_key:
        raise ValueError(
            "Missing gateway config. Set AI_GATEWAY_URL and AI_GATEWAY_KEY in env/.env"
        )

    endpoint = gateway_url if gateway_url.endswith("/bedrock") else f"{gateway_url}/bedrock"
    os.environ["AWS_BEARER_TOKEN_BEDROCK"] = api_key

    return boto3.client(
        service_name="bedrock-runtime",
        region_name="us-east-1",
        endpoint_url=endpoint,
        aws_access_key_id="",
        aws_secret_access_key="",
        config=Config(retries={"max_attempts": 0}),
    )


def main() -> None:
    model_id = os.getenv("BEDROCK_MODEL_ID", "amazon.nova-lite-v1:0")
    client = build_bedrock_client()

    model = ModelFactory.create(
        model_platform=ModelPlatformType.AWS_BEDROCK_CONVERSE,
        model_type=model_id,
        model_config_dict={"temperature": 0.0, "max_tokens": 256},
        bedrock_client=client,
    )

    toolkit = PubMedToolkit(timeout=10)
    agent = ChatAgent(
        system_message=(
            "You must call the search_papers tool exactly once before answering."
        ),
        model=model,
        tools=toolkit.get_tools(),
        retry_attempts=1,
        step_timeout=120,
    )

    prompt = (
        "Find papers on YIBEUIWD2u and give one-sentence summary. "
        "Use search_papers first."
    )
    response = agent.step(prompt)
    print(response.msgs[0].content)

if __name__ == "__main__":
    main()

Command lines:

python test.py

Steps to reproduce:

1. Name the reproducible code to test.py
2. Run python test.py

Traceback

Traceback (most recent call last):
  File "/Users/xxx/test.py", line 80, in <module>
    main()
  File "/Users/xxx/test.py", line 76, in main
    response = agent.step(prompt)
               ^^^^^^^^^^^^^^^^^^
  File "/Users/xxx/camel/camel/agents/chat_agent.py", line 2872, in step
    return future.result(timeout=self.step_timeout)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/xxx/miniconda3/envs/agent/lib/python3.11/concurrent/futures/_base.py", line 456, in result
    return self.__get_result()
           ^^^^^^^^^^^^^^^^^^^
  File "/Users/xxx/miniconda3/envs/agent/lib/python3.11/concurrent/futures/_base.py", line 401, in __get_result
    raise self._exception
  File "/Users/xxx/miniconda3/envs/agent/lib/python3.11/concurrent/futures/thread.py", line 58, in run
    result = self.fn(*self.args, **self.kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/xxx/Downloads/AZ/camel/camel/agents/chat_agent.py", line 2957, in _step_impl
    response = self._get_model_response(
               ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/xxx/camel/camel/agents/chat_agent.py", line 3590, in _get_model_response
    response = self.model_backend.run(
               ^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/xxx/camel/camel/models/model_manager.py", line 239, in run
    raise exc
  File "/Users/xxx/camel/camel/models/model_manager.py", line 229, in run
    response = self.current_model.run(messages, response_format, tools)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/xxx/camel/camel/models/base_model.py", line 214, in wrapped_run
    result = original_run(self, messages, *args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/xxx/camel/camel/models/base_model.py", line 873, in run
    result = self._run(messages, response_format, tools)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/xxx/camel/camel/models/aws_bedrock_converse_model.py", line 866, in _run
    response = self.bedrock_client.converse(**request)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/xxx/miniconda3/envs/agent/lib/python3.11/site-packages/botocore/client.py", line 602, in _api_call
    return self._make_api_call(operation_name, kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/xxx/miniconda3/envs/agent/lib/python3.11/site-packages/botocore/context.py", line 123, in wrapper
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/Users/xxx/miniconda3/envs/agent/lib/python3.11/site-packages/botocore/client.py", line 1078, in _make_api_call
    raise error_class(parsed_response, operation_name)
botocore.errorfactory.ValidationException: An error occurred (ValidationException) when calling the Converse operation: The format of the value at messages.2.content.0.toolResult.content.0.json is invalid. Provide a json object for the field and try again.

Expected behavior

It should call search_papers in PubMedToolkit without errors.

Additional context

In search_papers when nothing is found am empty list will return and toolResult.content[].json is being sent as a non-object (for example a list) when a tool returns a top-level list, which Bedrock rejects.
code pointer:

camel/camel/toolkits/pubmed_toolkit.py

Lines 150 to 156 in d84c72c

	results = []
	for paper_id in paper_ids:
	paper_details = self.get_paper_details(paper_id)
	if paper_details:
	results.append(paper_details)
	return results

Proposed fix:

1. In CAMEL’s Bedrock Converse adapter (not only in PubMedToolkit), normalize every tool result before building toolResult.content[].json.
2. Wrap the PubMedToolkit returns with a dict

[HelloJocelynLu]

@voidborne-d would you like to take this one as well? I think you might be interested in