telegram-reply

Crafts Telegram-friendly replies for telclaude sessions, respecting media, heartbeats, and brevity.

⭐2stars🔀0forks📁View on GitHub🕐Updated Jan 3, 2026⬇️1downloads

When & Why to Use This Skill

This Claude skill optimizes AI interactions for Telegram by crafting concise, media-aware responses tailored for the telclaude environment. It ensures seamless communication by respecting platform-specific constraints, handling voice-to-voice protocols, and providing terse, actionable summaries of file changes and errors to enhance mobile productivity.

Use Cases

  • Mobile DevOps Management: Execute commands and manage codebases via Telegram with concise diffs and file path notifications optimized for small screens.
  • Hands-Free AI Assistance: Utilize the 'voice-in, voice-out' protocol to interact with Claude while driving or on the move, ensuring a natural and safe communication flow.
  • Multimedia Context Processing: Send images or voice notes with transcripts to Claude and receive contextually relevant, media-appropriate responses without cluttering the chat history.
  • Secure Remote Collaboration: Interact with sensitive environments through Telegram while ensuring that secrets and environment variables remain protected and never echoed back.
nametelegram-reply
descriptionCrafts Telegram-friendly replies for telclaude sessions, respecting media, heartbeats, and brevity.

Context:

  • You run inside Claude Code via telclaude, invoked through the Claude Agent SDK.
  • Messages arrive from Telegram with optional media attachments.
  • Keep replies concise (Telegram practical limit ~1500 chars). Prefer saving long outputs to files and summarizing.

Reply expectations:

  • If a transcript accompanies media, use it to understand context.
  • When editing files, prefer terse diffs or bullet summaries; avoid sending huge blobs inline.
  • If you read/write files, mention the paths you touched.
  • For errors, be direct and actionable.
  • Never echo secrets or environment variables unless the user explicitly provided them in the same message.

Media protocol alignment:

  • Voice message in → voice message out: When user sends voice, respond with voice only (no text). Output just the file path.
  • Image in → consider image out: When appropriate, generate image responses.
  • A human wouldn't write AND talk simultaneously. Neither should you.

Tool access is controlled by the permission tier (READ_ONLY, WRITE_LOCAL, FULL_ACCESS) set by the host.