Telegram MCP has critical definition gaps. Tool descriptions are present but minimal (10-60 chars, well below the 194-char baseline). Parameter descriptions are sparse or missing entirely. Input schemas lack proper type definitions and descriptions for most parameters. No output schemas documented. Error handling is absent, tools make raw HTTP calls without validation or recovery guidance. The server exposes internal HTTP endpoint details and lacks idempotency markers. No tool annotations (readOnlyHint/destructiveHint). send_message is a destructive operation but has no confirmation or dry-run pattern.

• Expand tool descriptions to 100-200 chars. Example: 'get_unread_entities: Retrieve all chats/channels with unread messages. Returns entity ID, title, and unread count. Use this to discover which conversations need attention before calling get_unread_messages for details.'
• Add full parameter descriptions. For 'id' in get_unread_messages: 'Entity ID (integer or string username). If omitted, returns unread messages from all entities.' For 'content' in send_message: 'Message text to send. Supports HTML markup (bold, italic, links). Required unless reply_to_message_id is set.'
• Document output schemas for each tool. Example for get_entities: 'Returns dict[int, dict] where each value has keys: id (int), title (str), unread_count (int), is_pinned (bool), type (str), last_message_id (int|null).'
• Add error handling with actionable messages. Wrap HTTP calls in try-except; return structured errors like '{"error": "Entity not found", "suggestion": "Try get_entities() to list available chats"}' instead of raw exceptions.
• Add tool annotations. Mark get_* tools with readOnlyHint=true. Mark send_message with destructiveHint=true and idempotentHint=false. Add a dry_run parameter to send_message to support confirmation pattern.