MCP server for debugging Telegram bots via a user session (kurigram/pyrogram MTProto)
23 tools with consistent verb_noun naming (get_, send_, delete_, etc.) and complete input schemas. All tools have descriptions (avg 65 chars, baseline 194). However, descriptions are terse and lack WHEN/WHY context, dependency hints, or recovery guidance. Parameter descriptions are minimal (avg 35 chars, baseline 72). No output schemas documented. Error handling is absent, no guidance on retryability, user-fixable vs fatal errors, or recovery paths. Missing tool annotations (readOnlyHint, destructiveHint). Risk labels present but not integrated into descriptions.
Add emoji reaction to a message
Ban a user from a chat
Delete messages from a chat
Edit an existing message
Forward messages to another chat
Retrieve chat information by ID or username
Retrieve chat message history with filtering
No output schemas documented. LLMs cannot plan downstream tool calls or extract required fields (e.g., chat_id, message_id for chaining). Baseline: 100% of A+ tools document return types.
Descriptions lack WHEN/WHY context and recovery guidance. Avg 65 chars vs baseline 194. E.g., 'Get current logged-in user information' does not explain when to call whoami vs get_user, or what fields are returned. Baseline: 94% of A+ tools have complete docstrings.
Parameter descriptions are minimal (avg 35 chars). E.g., 'Chat ID (number or @username)' lacks format constraints, valid ranges, or examples. Baseline: 100% of A+ tool params have descriptions; avg 72 chars. LLMs cannot infer whether to pass 'john' or '@john' or '12345'.
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | 2026-07-28+ | v2 |
List members of a chat
List user's recent chats/dialogs
Download a file from Telegram
Retrieve messages from a chat
Get user information by ID or username
Leave a chat
Mark all messages in a chat as read
Test MTProto connection latency
Remove emoji reaction from a message
Search messages in a chat
Send typing/recording action to a chat
Send a message to a chat
Change chat description
Change chat title
Unban a user from a chat
Get current logged-in user information
No error handling guidance. Tools declare Risk labels (READ_ONLY, WRITE, DESTRUCTIVE) but descriptions do not explain retryability, user-fixable errors, or recovery paths. E.g., delete_messages should warn of irreversibility and suggest confirmation patterns.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Risk labels exist but are not exposed in tool metadata. LLMs cannot distinguish safe retry candidates from destructive operations without explicit annotations.