Three tools with reasonable naming and descriptions, but schemas lack depth and parameter documentation is inconsistent. lanhu_design has the most comprehensive schema with enums and array support; lanhu_page and lanhu_resolve_invite are simpler. All tools are READ_ONLY (good security posture). Descriptions are adequate (100-200 chars) but lack actionable context for LLM selection. Parameter descriptions exist but are sparse, e.g., 'Lanhu project URL' doesn't explain format expectations or when to use stage vs detailDetach. No output schemas documented. Error handling and recovery guidance absent.

• Document output schemas for all three tools. Example: lanhu_design analyze mode returns {html: string, tokens: {colors: [...], fonts: [...]}, layout: {...}}. This enables LLMs to extract and chain results.
• Expand parameter descriptions with format constraints. E.g., 'Lanhu project URL in format https://lanhuapp.com/web/#/item/project/stage?pid=<PROJECT_ID>&tid=<TEAM_ID> or detailDetach format with image_id query param.'
• Add error handling guidance to tool descriptions. E.g., 'Returns error if URL is invalid, project is private, or auth token expired. If auth fails, check LANHU_COOKIE environment variable.'
• Clarify lanhu_design 'design_names' parameter: 'Can be design name (exact match), design ID, numeric index from list mode, or string "all". If multiple names match, returns first match.'
• Document lanhu_design 'include' array behavior: 'Specify which content types to extract. Default: [html, tokens]. Combining image+html returns base64-encoded images; omit image to reduce response size.'