Three tools with basic schemas and Chinese descriptions. All tools have input schemas with type definitions, but descriptions lack LLM-optimization guidance (no WHEN to use, no prerequisites, no error recovery hints). Parameters are minimally described. No output schemas documented. Error handling is absent, no guidance on retryability, user-fixable vs fatal errors, or recovery paths. Naming is verb-based (fetch_, set_, check_) but descriptions don't explain distinctions or dependencies between tools. No tool annotations (readOnlyHint/destructiveHint). set_wecom_cookie modifies state but description doesn't warn of side effects or idempotency.

• Expand fetch_wecom_doc description to 100-150 chars: 'Fetch WeChat Work document content and convert to Markdown. Supports docs, sheets, slides, mind maps. Requires valid Cookie (set via set_wecom_cookie first). Returns Markdown-formatted text with title.'
• Add output schema documentation: fetch_wecom_doc returns {type: 'string', description: 'Markdown-formatted document with title as H1 heading'}, check_wecom_auth returns {type: 'boolean', description: 'true if Cookie is valid, false if expired or invalid'}
• Add tool annotations: mark fetch_wecom_doc and check_wecom_auth with readOnlyHint=true; mark set_wecom_cookie with destructiveHint=true (modifies persistent config)
• Enhance parameter descriptions: url should include 'Must be a valid WeChat Work doc URL (https://doc.weixin.qq.com/doc/*, /sheet/*, /slide/*, /mind/*, etc.)'; cookie should specify 'Complete Cookie string from browser Network tab, typically 100-500 chars'
• Add error recovery guidance to descriptions: 'If auth fails (401/403), Cookie may be expired, call set_wecom_cookie with a fresh Cookie from browser. If doc not found, verify URL is accessible in browser first.'