Originally posted by wigging July 7, 2026
I have an agent that uses an MCP server for one of its tools as demonstrated below.
agent = Agent(
client,
instructions,
name="My Agent",
description="Agent for the system",
tools=[my_mcp],
)
This agent is served with the A2A protocol as shown next.
agent_card = AgentCard(
name="My Agent",
description="Agent for the system",
)
request_handler = DefaultRequestHandler(
agent_executor=A2AExecutor(agent),
task_store=InMemoryTaskStore(),
agent_card=agent_card,
)
app = Starlette(
routes=[
*create_agent_card_routes(agent_card),
*create_jsonrpc_routes(request_handler, "/"),
]
)
uvicorn.run(app)
To test this agent I have some client code as shown below. This connects to the remote A2A agent which, for demo purposes, is running on localhost port 8000.
# client.py
a2a_url = "http://127.0.0.1:8000"
agent = A2AAgent(name="My A2A Agent", url=a2a_url)
user_prompt = "what is status of the system"
print(f"\nUser Prompt:\n{user_prompt}")
response = await agent.run(user_prompt)
print(f"\nRemote Agent Response:\n\033[96m{response}\033[0m")
Everything works fine but I noticed the A2A server gives the following output when the client sends a request:
A2AExecutor does not yet support content type: function_call. Omitted.
A2AExecutor does not yet support content type: function_result. Omitted.
I think this is caused by the MCP tool on the remote agent. Is this something I need to worry about or should I just ignore it? Is there a way to prevent these A2AExecutor log statements from displaying in the server output?
Discussed in #6940
Originally posted by wigging July 7, 2026
I have an agent that uses an MCP server for one of its tools as demonstrated below.
This agent is served with the A2A protocol as shown next.
To test this agent I have some client code as shown below. This connects to the remote A2A agent which, for demo purposes, is running on localhost port 8000.
Everything works fine but I noticed the A2A server gives the following output when the client sends a request:
I think this is caused by the MCP tool on the remote agent. Is this something I need to worry about or should I just ignore it? Is there a way to prevent these A2AExecutor log statements from displaying in the server output?