Both tools have descriptions and basic input schemas, but critical gaps undermine quality. publish_content has a reasonable description (127 chars) and complete schema with 4 parameters, each with type and description. crawl_web has a shorter description (108 chars) and 2 parameters with types and descriptions. However, neither tool includes output schemas, error handling guidance, or recovery instructions. Parameter descriptions lack constraints (e.g., max image count, URL format validation). No enum constraints for categorical inputs. Missing idempotency guidance for the destructive publish_content tool. Descriptions do not explain when to use each tool or what happens on failure. No mention of prerequisites or dependencies between tools.

• Add explicit output schemas for both tools. publish_content should document what fields are returned (e.g., post_id, publish_url, timestamp). crawl_web should document the structure of extracted content and images.
• Enhance publish_content description to include error scenarios: 'Publishes content to Xiaohongshu. Returns post_id on success. Fails if title exceeds 20 chars, content exceeds 1000 chars, or image count exceeds 9. On failure, returns error code and guidance for retry.'
• Add enum constraints for categorical parameters. If tags have a predefined set, use enum. If imagePaths must be URLs or local paths, document the format explicitly.
• Document idempotency for publish_content. State whether calling it twice with the same input produces one or two posts. If idempotent, include a request_id parameter for deduplication.
• Add recovery guidance to crawl_web: 'If URL is unreachable, returns error with HTTP status. Retry with a different URL or use search_content instead. extractImages=true returns image URLs; extractImages=false returns text only.'