[BUG] Recent activity tool call fails with pydantic validation error

[chishaku]

Bug Description

Prompt: get recent notes from today and summarize

Error

DEBUG Getting function recent_activity
DEBUG Running: recent_activity(timeframe=today, type=note)
DEBUG Calling MCP Tool 'recent_activity' with args: {'tool_description': 'Get recent activity from across the knowledge base.\n    \n    Timeframe supports natural language formats like:\n    - "2 days ago"  \n    - "last
      week"\n    - "yesterday" \n    - "today"\n    - "3 weeks ago"\n    Or standard formats like "7d"\n    ', 'timeframe': 'today', 'type': 'note'}
ERROR    Failed to call MCP tool 'recent_activity': Error from MCP tool 'recent_activity': [TextContent(type='text', text="Error executing tool recent_activity: 1 validation error for recent_activityArguments\ntype\n  Input
         should be a valid list [type=list_type, input_value='note', input_type=str]\n    For further information visit https://errors.pydantic.dev/2.11/v/list_type", annotations=None)]
         Traceback (most recent call last):
           File "/usr/local/google/home/jeffwayne/.cache/uv/environments-v2/tpgm-9150c9e49cf98506/lib/python3.12/site-packages/agno/utils/mcp.py", line 37, in call_tool
             raise Exception(f"Error from MCP tool '{tool_name}': {result.content}")
         Exception: Error from MCP tool 'recent_activity': [TextContent(type='text', text="Error executing tool recent_activity: 1 validation error for recent_activityArguments\ntype\n  Input should be a valid list
         [type=list_type, input_value='note', input_type=str]\n    For further information visit https://errors.pydantic.dev/2.11/v/list_type", annotations=None)]

Steps To Reproduce

Steps to reproduce the behavior:

1. Install version 0.12.2
2. Run command uvx basic-memory --project=notes mcp
3. Use tool/feature recent activity
4. See error description above

[phernandez]

claude: can you track down what caused this error?

[github-actions]

Claude Code Analysis

The bug and solution are now clear. Let me provide a comprehensive analysis.

Issue Analysis

The bug in the recent_activity tool is occurring because:

1. The tool function defines type as Optional[List[SearchItemType]] (a list of enum values)

2. When a user calls the tool with type=note, it's passing a string instead of a list

3. In the tool implementation, it tries to iterate through type with:

   params["type"] = [type.value if isinstance(type, SearchItemType) else type for type in type]

4. When type is a string like "note", Python treats it as an iterable of characters, causing the code to iterate through each character ('n', 'o', 't', 'e')

5. The validation error occurs because the API endpoint expects a list but gets a string

Root Cause

The main issue is a type mismatch: the function expects List[SearchItemType] but is receiving a plain string. There's also a secondary issue that "note" isn't a valid SearchItemType - the valid values are "entity", "observation", and "relation".

Potential Fixes

There are two main approaches to fix this issue:

1. Fix the parameter handling in the tool implementation

@mcp.tool(description="...")
async def recent_activity(
    type: Optional[Union[List[SearchItemType], SearchItemType, str]] = None,
    # other parameters...
) -> GraphContext:
    # ...
    
    # Handle different type inputs: string, enum, or list
    if type:
        if isinstance(type, str):
            # Convert single string to list
            params["type"] = [type]
        elif isinstance(type, SearchItemType):
            # Convert single enum to list with value
            params["type"] = [type.value]
        else:
            # Handle list of values (could be strings or enums)
            params["type"] = [
                t.value if isinstance(t, SearchItemType) else t for t in type
            ]

2. Add parameter validation and mapping

@mcp.tool(description="...")
async def recent_activity(
    type: Optional[Union[List[str], str]] = None,
    # other parameters...
) -> GraphContext:
    # ...
    
    # Map user-friendly names to actual SearchItemType values
    type_mapping = {
        "note": "entity",  # Map "note" to "entity"
        "entity": "entity",
        "observation": "observation",
        "relation": "relation"
    }
    
    if type:
        if isinstance(type, str):
            # Map and convert single string to list
            mapped_type = type_mapping.get(type, type)
            params["type"] = [mapped_type]
        else:
            # Map and handle list of string values
            params["type"] = [type_mapping.get(t, t) for t in type]

Recommended Solution

The best solution is to improve the parameter handling to accept either a string or a list and properly convert it to the expected format:

@mcp.tool(description="...")
async def recent_activity(
    type: Optional[Union[List[str], str]] = None,
    # other parameters...
) -> GraphContext:
    # ...
    
    # Handle type parameter properly
    if type:
        # Convert single string to list
        if isinstance(type, str):
            params["type"] = [type]
        # Already a list, keep as is
        else:
            params["type"] = type

This approach is simple and doesn't require changing APIs. It makes the tool more flexible by accepting both string and list parameters, which is a common expectation for CLI tools.

Files That Need Modification

Only one file needs to be modified:

• /src/basic_memory/mcp/tools/recent_activity.py

The specific changes needed are:

1. Update the type annotation to accept either a string or a list: type: Optional[Union[List[str], str]] = None
2. Modify the parameter handling code to handle both cases
3. Make sure valid SearchItemType values are validated or mapped

This fix will make the tool more robust and ensure it can handle different input formats from users.

---

🤖 Generated with Claude Code GitHub Action