The server defines 7 tools with consistent naming patterns (verb_noun: search_works, search_people, get_work, list_work_files, etc.), which is good. All tools have basic descriptions and input schemas are present with type definitions. However, descriptions are quite short (typically 12-20 characters), falling below the 50-200 character optimum for LLM selection guidance. Parameter descriptions are minimal and lack important details like constraints, format expectations, or dependency hints. Output schemas are not explicitly documented in the visible code, tool implementations reference output shapes but these are not exported for agent visibility. Error handling is minimal with no recovery guidance. The schema definitions use proper JSON types (string, number) but lack constraints (no enums, no min/max ranges, no patterns). This is a competent but baseline implementation.

• Expand each tool description to 50-200 characters. Example: 'search_works' → 'Search IMSLP for works by title, composer, or keyword. Returns paginated results with work IDs for use with get_work. Limit results to 1-100.'
• Add descriptions to all parameters. For 'limit': 'Maximum number of results per page (1 - 100, default 20). Higher limits may exceed API rate limits.' For 'offset': 'Starting position for pagination. Use with limit for multi-page results.'
• Add constraints to numeric parameters: use JSON Schema minValue/maxValue and document in descriptions. Example: limit with minValue=1, maxValue=100.
• Document output schemas explicitly in tool descriptions or separate schema docs. Specify what fields are returned (e.g., 'Returns: {id, title, composer, url, year}') so agents know what data is available for downstream calls.
• Add error recovery guidance to each tool description. Example: 'If no results found, try broader search terms or list available categories with browse_category.'
• Clarify what 'page' and 'category' parameters accept. If they are Wikipedia-style page titles, document format (e.g., 'Page title as it appears in IMSLP URL'). If internal IDs, explain lookup process.
• Document tool chains in descriptions. Example: 'search_works returns work_id; pass to list_work_files to fetch editions or to get_work for full details.'
• Add example successful responses to documentation to show agents what fields to expect and extract.

Inferred effective spec: 2025-06-18+.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry