TheAudioDB MCP has clear, well-structured tool definitions with consistent naming (verb_noun pattern: search_*, get_*) and solid descriptions (avg 180 chars, within 10-1024 baseline). All 4 tools have complete input schemas with typed properties and required arrays. However, output schemas are entirely undocumented, LLMs cannot predict response structure for planning downstream calls. Parameter descriptions are present but lack constraints (no enums, ranges, or format specs). Error handling returns generic error objects without recovery guidance. No tool annotations (readOnlyHint, etc.) despite all tools being read-only. Missing pagination guidance for potentially large result sets (search_artist, search_album could return many results).

• Document output schemas for all tools. Add a 'returns' section to each tool description or create a separate schema definition. Example: search_artist returns {found: boolean, artists: [{id: string, name: string, genre: string, style: string, mood: string, formedYear: number, country: string, website: string, biography: string, thumb: string}]}.
• Add pagination support to search_artist and search_album. Introduce optional limit (default 20, max 100) and offset parameters. Return total_count in response so agents know if more results exist.
• Enhance error messages with recovery guidance. Instead of {error: 'artist not found'}, return {error: 'artist not found', suggestion: 'Try search_artist() with a partial name or different spelling'}.
• Add readOnlyHint: true to all tool definitions in the McpToolDefinition interface. This signals to agents that these tools are safe to retry and have no side effects.
• Clarify parameter constraints in descriptions. E.g., 'artist' param: 'Artist or band name (case-insensitive, supports partial matching). E.g. "Coldplay" or "cold".' For IDs: 'TheAudioDB numeric artist ID (e.g. "111239"). Obtain from search_artist() response.'