Fix: Handle Exception case in lowlevel server _handle_message

Problem

The _handle_message method in src/mcp/server/lowlevel/server.py had a critical unhandled case for Exception types. Despite the function signature explicitly accepting Exception as a valid message type, the match statement was missing this case, potentially leading to silent failures in production environments.

Critical TODO was present:

# TODO(Marcelo): We should be checking if message is Exception here.
match message:  # type: ignore[reportMatchNotExhaustive]

Solution

🔧 Fixed the unhandled Exception case by:

1. Added proper Exception handling in the match statement:

   • Added case Exception() as error: to handle Exception instances
   • Used logger.exception() following CLAUDE.md guidelines for proper exception logging with traceback
   • Respected the raise_exceptions parameter to maintain backward compatibility

2. Removed the TODO comment and type ignore directive as the issue is now resolved

3. Added comprehensive test coverage in tests/server/test_exception_handling.py:

   • Test for proper exception logging when raise_exceptions=False
   • Test for exception re-raising when raise_exceptions=True
   • Test to ensure exceptions don't crash the server when handled gracefully

Files Changed

• src/mcp/server/lowlevel/server.py - Fixed Exception handling in _handle_message
• tests/server/test_exception_handling.py - Added comprehensive test coverage

Commit Info

• Branch: fix/exception-handling-lowlevel-server
• Files changed: 2 files, 126 insertions(+), 2 deletions(-)

Reviewers: @jerome3o-anthropic @jspahrsummers

This PR resolves the most critical outstanding TODO in the codebase and significantly improves the robustness of the MCP server implementation.