MCP API logging syntax error: TypeError on /api/mcp/logs/stream endpoint

[Wirasm]

Bug Description

The MCP API crashes when accessing the logs streaming endpoint (/api/mcp/logs/stream) due to incorrect logging syntax mixing standard Python logging with Logfire keyword arguments.

Error Details

TypeError: Logger._log() got an unexpected keyword argument 'error'
  File "/app/src/server/api_routes/mcp_api.py", line 619, in get_logs
    api_logger.error("MCP server logs API failed", error=str(e))

Steps to Reproduce

1. Navigate to the MCP page in the frontend
2. The page attempts to connect to WebSocket endpoint /api/mcp/logs/stream
3. When the endpoint encounters any error, the logging statement fails

Root Cause

The mcp_api.py file has inconsistent logging patterns:

• Line 619: api_logger.error("MCP server logs API failed", error=str(e)) ❌ (Logfire keyword syntax)
• Line 561: api_logger.error("MCP server start API failed - error=%s", str(e)) ✅ (Standard format)
• Line 580: api_logger.error(f"MCP server stop API failed - error={str(e)}") ✅ (F-string format)

The api_logger is created by get_logger() which returns a standard Python logging.Logger that doesn't support keyword arguments like error=.

Affected Code

File: python/src/server/api_routes/mcp_api.py Lines: 619, 637, 701, 826, 837 (all use error= keyword syntax)

Suggested Fix

Replace Logfire-style keyword logging with standard Python logging patterns:

# Current (broken):
api_logger.error("MCP server logs API failed", error=str(e))

# Fix option 1 (f-string):
api_logger.error(f"MCP server logs API failed - error={str(e)}")

# Fix option 2 (format string):
api_logger.error("MCP server logs API failed - error=%s", str(e))

Environment

• This affects all environments where Logfire is not fully configured
• Error surfaces when MCP page is accessed and log streaming fails

Impact

• MCP page completely unusable due to WebSocket connection failure
• All MCP logging functionality broken
• Crashes the entire request handling

Technical Notes

This is a pre-existing issue (introduced in commit 59084036 on 2025-08-13) but surfaces more frequently now due to increased MCP page usage.