Server has 5 tools with visible schemas and descriptions, but quality is significantly below production baseline. Tool descriptions are present but lack LLM-optimization guidance (e.g., 'search_1688_products' description is 200+ chars of implementation detail rather than user intent). Parameters have types but many lack descriptions or constraints. Output schemas are not documented, tools return JSON strings without declared structure. Error handling is minimal (mostly 'ERROR_AUTH_REQUIRED' or generic exceptions). No tool annotations (readOnlyHint, destructiveHint). Naming is reasonable (verb_noun pattern) but descriptions fail to answer WHEN to use each tool or what downstream tools expect.

• Add explicit output schema documentation for each tool. E.g., search_1688_products returns {count: int, items: [{title: string, url: string, price: string, company: string}]}. Use JSON Schema in tool definition or docstring.
• Rewrite descriptions as user-intent statements (50-150 chars). E.g., 'Search 1688 products by keyword with optional location and factory filters. Returns top results with pricing and company info.' Avoid implementation details.
• Add parameter constraints and enums. E.g., location_filter: enum ['山东', '广东', '浙江', ...] or pattern '^[\u4e00-\u9fff]+$'. max_results: integer, min=1, max=50.
• Implement actionable error messages with recovery guidance. E.g., 'Authentication required (ERROR_AUTH_REQUIRED). Call update_auth_cookie() to log in, then retry.' Include the tool name to call next.
• Add tool annotations via FastMCP metadata. Mark read-only tools with readOnlyHint=true, update_auth_cookie with destructiveHint=true. This enables LLM reasoning about side effects and retry safety.