The WeChat MCP server has 7 tools with basic definitions but significant quality gaps. Most tools have descriptions and schemas, but parameter descriptions are incomplete or generic. Naming is reasonably clear (verb_noun style), but parameter documentation is sparse. Output schemas are not documented. Error handling is minimal, no recovery guidance or validation feedback visible in code. The server exposes raw database/API complexity to the agent without abstraction.

• Add output schema documentation to all 7 tools. Specify fields, types, and include example response structures. For get_wechat_message, document the list structure, user info mapping, and pagination metadata.
• Add input validation to sql_wechat_message: sanitize SQL, reject dangerous keywords (DROP, DELETE, ALTER), or switch to a parameterized query builder. Return clear error: 'SQL injection risk detected. Only SELECT queries allowed on message table.'
• Add pagination parameters to get_wechat_message, get_wechat_contacts, and get_wechat_groups. Default limit=20, max=100. Document the returned total_count or next_cursor.
• Expand parameter descriptions with format, constraints, and resolution hints. Example: 'name (string, required): Name of the contact or group to fetch messages from. If unsure of exact spelling, call get_wechat_contacts() first. Must match exactly.'
• Add per-tool error handling stanzas in code. Return structured errors: {'error': 'contact_not_found', 'message': 'Contact "Alice" not found. Did you mean: Alice_Work, Alice_Family?', 'suggestion': 'Try get_wechat_contacts() to list available contacts.'}
• Add dry-run parameter to send_wechat_message and send_wechat_file: 'dry_run' (boolean, optional, default=false). When true, return {'preview': 'Would send to Alice: Hello!', 'status': 'ready_to_send'}. When false, send and return send confirmation with timestamp.
• Document whether contact 'name' must be exact match or supports fuzzy lookup. If exact, update description. If fuzzy, clarify the matching algorithm and confidence threshold.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry