Validation Error with create_entities: Missing content Field in Memory Server Response

[fatsandfreedom]

Description

Hello, I am encountering a validation error when interacting with the MCP Memory Server using the create_entities tool. The error indicates that the response from the Memory Server is missing the required content field.

The issue arises when calling the create_entities tool through a Python client. Despite the Memory Server returning what seems to be a valid response, the client throws a validation error, stating that the content field is missing.

Error Output

Here’s the error output from my debug script:

Knowledge Graph MCP Server running on stdio
Error calling Memory Server: 1 validation error for CallToolResult
content
  Field required [type=missing, input_value={'toolResult': [{'name': ... ['Test observation']}]}, input_type=dict]
    For further information visit https://errors.pydantic.dev/2.10/v/missing

Steps to Reproduce

• 1. Run the following Python debug script to test the create_entities tool:

import asyncio
from mcp import ClientSession, StdioServerParameters
from mcp.client.stdio import stdio_client

async def test_memory_server():
    server_params = StdioServerParameters(
        command="npx",
        args=[
            "-y",
            "@modelcontextprotocol/server-memory"
        ]
    )

    async with stdio_client(server_params) as (read, write):
        async with ClientSession(read, write) as session:
            await session.initialize()

            # Example call to create_entities
            command = "create_entities"
            parameters = {
                "entities": [
                    {
                        "name": "Test_Entity",
                        "entityType": "person",
                        "observations": ["Test observation"]
                    }
                ]
            }

            try:
                result = await session.call_tool(command, parameters)
                print(f"Memory Server result for '{command}': {result}")
            except Exception as e:
                print(f"Error calling Memory Server: {e}")

asyncio.run(test_memory_server())

• 2. Observe the error output described above.

What I Expected

I expected the create_entities tool to return a successful response with the following structure:

{
  "content": {
    "toolResult": [
      {
        "name": "Test_Entity",
        "entityType": "person",
        "observations": ["Test observation"]
      }
    ]
  }
}

What I Observed

The actual response from the Memory Server was missing the content field. Instead, it returned:

{
  "toolResult": [
    {
      "name": "Test_Entity",
      "entityType": "person",
      "observations": ["Test observation"]
    }
  ]
}

This mismatch causes the client to throw a validation error.

Environment

Python Version: 3.10
Operating System: Ubuntu 20.04

[ArsenicBismuth]

I also have this issue for memory & puppeteer servers, but not with Brave server. So, it seems server implementation issue?

If successful, my personal AI is capable of calling the tool & reacting to the tool response. But for both memory & puppeteer servers, it stops right after invoking the tool.

[fatsandfreedom]

I also have this issue for memory & puppeteer servers, but not with Brave server. So, it seems server implementation issue?

If successful, my personal AI is capable of calling the tool & reacting to the tool response. But for both memory & puppeteer servers, it stops right after invoking the tool.

In memory server, I used a workaround to address this issue.

async def safe_call_tool(session, command, parameters):
    """Safely call an MCP tool and handle validation errors."""
    try:
        result = await session.call_tool(command, parameters)
        return result
    except ValidationError as e:
        # Extract the actual tool result from the validation error
        raw_data = e.errors() if hasattr(e, 'errors') else None
        if raw_data and len(raw_data) > 0:
            tool_result = raw_data[0]['input']['toolResult']
            return {"toolResult": tool_result}
        # Re-raise the exception if the result cannot be extracted
        raise

[jspahrsummers]

Thanks for flagging! I believe #176 fixes this issue, but please let us know if not.