13 tools with consistent naming (verb_noun pattern: telegram_*) and complete input schemas using Zod. Descriptions are present but brief (avg 80 chars, below 194-char baseline). All parameters have type definitions and descriptions. However, output schemas are not documented, LLMs cannot predict response structure. No error handling guidance, no recovery hints, and no parameter constraints (enums, ranges). Missing idempotency/destructiveness annotations despite destructive operations (delete_message). Tool composition is sound (single responsibility), but descriptions lack WHEN/WHY context needed for LLM selection.

• Document output schemas for all 13 tools. Include field names, types, and sample structure. E.g., telegram_get_me returns {id: number, is_bot: boolean, first_name: string, ...}. This enables LLM to chain tools and extract correct fields.
• Expand descriptions to 100 - 200 chars. Add WHEN/WHY context. E.g., 'Get chat details (name, members, permissions) by ID or @username. Use this to verify chat exists before sending. Prefer telegram_lookup_chat for username resolution only.'
• Add error handling guidance to write tools. E.g., telegram_send_message: 'If chat not found, try telegram_lookup_chat first. If rate-limited, retry after 30s. If permission denied, check bot admin status.'
• Annotate tools with readOnlyHint/destructiveHint. Mark telegram_get_* with readOnlyHint=true. Mark telegram_delete_message with destructiveHint=true. This signals safety to agents.
• Add parameter constraints. E.g., messageId: {type: 'number', minimum: 1, description: 'Message ID (must be > 0)'}. Add emoji validation for telegram_react_to_message: pattern or enum of valid emoji.