Server defines 4 tools with complete input schemas and basic descriptions. All tools are read-only, properly named with action verbs (search, get, creators, exhibitions). Schemas include type definitions and parameter descriptions. However, descriptions are terse (10-60 chars), lack context on WHEN to use each tool, and output schemas are undocumented. Parameter descriptions are minimal, e.g., 'query' lacks guidance on search syntax or expected result cardinality. No error handling guidance, no pagination documentation despite limit/skip parameters. Missing idempotency hints and chaining IDs in response documentation.

• Expand tool descriptions to 80 - 150 chars. Example: 'search: Search the Cleveland Museum's 63k artworks by title, artist, or keyword. Use filters (type, cc0, has_image) to narrow results. Returns paginated list of artwork summaries with IDs for get_artwork().'
• Document output schemas for each tool. Specify what fields are returned (e.g., search returns [{id, title, artist, image_url, ...}]). Include required chaining IDs (artwork IDs for get_artwork).
• Add parameter constraints to descriptions. Example: 'limit: 1 - 1000, default 25. Pagination: use skip to offset results. Returns up to limit items per call.'
• Enhance error handling with recovery hints. Example: if 404, return 'Artwork not found. Try search() with a partial title or artist name.' If 500, suggest retry.
• Add idempotency note: 'All tools are read-only and safe to retry. No state changes occur.'