Two tools with minimal descriptions and no input parameter descriptions. add_note has a basic description (54 chars) but lacks parameter detail. read_notes has a typo-ridden description (54 chars, 'return all notes') and no parameters to describe. Neither tool documents output schema or error handling. No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Schemas are inferred from function signatures rather than explicitly registered. No pagination, no error guidance, no recovery paths.

• Add explicit parameter descriptions in tool definitions. For add_note, document 'message' as: 'The note content to append (plain text, max 1000 chars). Will be saved on a new line.'
• Document output schemas. add_note should return {"status": "success", "message": "Note saved"}. read_notes should return {"notes": [string], "count": int, "last_updated": ISO8601_string}.
• Add tool annotations: @mcp.tool(annotations={"destructiveHint": true}) for add_note; @mcp.tool(annotations={"readOnlyHint": true}) for read_notes.
• Implement error handling with recovery guidance. Catch file I/O exceptions and return: {"error": "Failed to save note: disk full", "recovery": "Check available disk space and retry"}.
• Fix typo in read_notes description and expand to 100+ chars explaining when to use it vs. get_latest_note resource.