Two tools with complete input schemas and descriptions, but significant gaps in output documentation and error handling. search_housing has well-structured Pydantic models (SearchRequest, SearchResponse, Listing) with field descriptions in Chinese, but output schema is not explicitly documented in tool registration. get_listing_detail lacks output schema documentation entirely. Both tools accept reasonable parameters with types and defaults, but error recovery guidance is minimal. No tool annotations (readOnlyHint, destructiveHint) despite both being read-only operations.

• Document output schemas in tool definitions. Add 'outputSchema' field to each tool with JSON Schema for SearchResponse, DetailResponse, and Listing. Include field types and descriptions so LLMs know what to expect.
• Add error handling guidance to descriptions. E.g., 'May return warnings if captcha detected or access denied. Retry with different site_allow/site_block filters if blocked.'
• Add readOnlyHint: true to both tool schemas to signal safety to agents.
• Expand get_listing_detail description to 100-150 chars: 'Fetch detailed housing listing info from a URL. Extracts structured data from JSON-LD, Next.js data, and HTML meta tags. Use after search_housing to get full property details.'
• Add min/max constraints to search_housing limit parameter: 'limit (1-50, default 10)' in description and enforce in code.