Three tools with clear verb-noun naming (search, get_details, get_download_url) and documented schemas via schemars. All parameters have type definitions and descriptions. However, tool descriptions are brief (34 - 72 chars), below the 50 - 200 char LLM-optimized baseline. Output schemas are not documented, callers cannot predict response structure. Error handling returns generic messages without recovery guidance. No parameter constraints (enums, ranges) despite opportunities (e.g., page should have min/max). Missing idempotency and composition guidance.

• Expand tool descriptions to 80 - 150 chars. Example: 'search' → 'Search Anna's Archive for books, papers, and documents by title, author, or keyword. Returns paginated results with metadata. Use page parameter to navigate results.'
• Document output schemas for each tool. Specify fields returned (e.g., search returns [items: [{md5, title, author, year}], total_results, page]). Include field types and whether pagination is supported.
• Enhance error messages with recovery guidance. Example: 'Search failed: network timeout. Retry with a shorter query or check your internet connection.' or 'API key invalid. Ensure ANNAS_ARCHIVE_API_KEY is set correctly.'
• Add parameter constraints: page (min 1, default 1), path_index and domain_index (0-based integers, optional). Document what happens if indices are out of range.
• Add idempotency note to get_details and get_download_url: 'Safe to retry, no side effects.' This signals to agents that retries are safe.