Two tools with basic schemas and descriptions, but significant gaps in parameter documentation, output schema clarity, and error handling guidance. send_message and get_history both lack descriptions for their return values, making it unclear what fields the LLM should expect. Parameter descriptions are present but minimal (e.g., 'Bot username' lacks format guidance). No enum constraints on the 'bot' parameter despite it accepting a specific format. Error responses return JSON strings rather than structured objects, and lack actionable recovery guidance. Tool composition is reasonable (send vs retrieve), but output documentation is insufficient for reliable agent chaining.

• Document output schemas formally. For send_message, specify: {status: 'ok'|'timeout'|'error', bot: string, sent: string, reply?: string, timestamp: string, error?: string}. For get_history: {status: 'ok'|'error', bot: string, messages: [{id: int, from: string, text: string, date: string}], count: int, timestamp: string}.
• Add enum constraint to 'bot' parameter or validate format server-side with clear error: 'Invalid bot username. Must be @username or username (alphanumeric + underscore, 5-32 chars).'
• Enhance error messages with recovery guidance. E.g., 'Session expired. Run telegram-mcp-auth to re-authenticate, then retry.' or 'Bot not found. Check spelling or use search_bots() to discover available bots.'
• Clarify timeout behavior in send_message description: 'If bot does not reply within timeout seconds, returns status=timeout. Retry with longer timeout or check if bot is responsive.'
• Add idempotency key or deduplication: accept optional 'message_id' parameter to deduplicate retries, or document that repeated calls will send duplicate messages and agent should track sent message IDs.