The server has 8 tools with basic schemas and descriptions, but significant gaps prevent a higher score. All tools start with action verbs (search, get, check, suggest) which is good. Descriptions exist but are generic and lack contextual guidance (e.g., 'Search products by query, fiber type, color, technique, skill level' doesn't explain WHEN to use this vs other product tools). Most parameters have type definitions and brief descriptions, but lack constraints (enums, ranges, formats) that LLMs need for valid input generation. Output schemas are completely undocumented, no return types visible in the source. Error handling is absent, no guidance on recovery or retryability. The tool composition is reasonable (each does one thing), but chaining is not evident from documentation.

• Add documented output schemas for all tools. For each tool, publish a JSON object showing return fields (product_id, name, price, fiber_type, availability, etc.) with types and descriptions. This is critical for LLM planning.
• Expand tool descriptions to 100-150 characters with WHEN-to-use guidance. Example: 'search_products: Find products by keyword, fiber, color, or technique. Use this for open-ended searches. For project-specific suggestions, use suggest_for_project instead.'
• Convert freeform string parameters to enums with valid values. Add to search_products: fiber_type enum [cotton, wool, silk, linen, synthetic, blend]; skill_level enum [beginner, intermediate, advanced, expert]; color enum or pattern constraint.
• Add error handling descriptions to each tool. Example for get_product_details: 'Returns 404 if product_id does not exist. Call search_products to find valid IDs. Returns 400 if product_id is not an integer.'
• Clarify get_agent_guidance purpose and parameters. Rename to something more specific if this is a meta-guidance tool (e.g., get_search_guidance). Convert context_type to enum with examples: no_results, too_many_results, ambiguous_query, skill_mismatch.
• Add pagination parameters to search tools. search_products and search_support should accept limit (default 20, max 100) and offset (default 0). Return total_count so agent knows if more results exist.
• Document parameter constraints in descriptions. For product_id: 'Integer > 0'. For price_range: 'Format: "10-50" (in currency units) or open-ended like "50+"'. For difficulty_level: 'One of: beginner, intermediate, advanced'.