Skip to content

Open Source / MIT License

Telegram AI Bridge

Bring the full power of Claude Code and Codex CLI
into a Telegram chat window anyone can use.
Not at your desk? Dispatch AI agents from your phone.

Node.js ~900 LOC 3 Dependencies MIT

The Problem

Claude Code and Codex CLI are the most powerful AI coding assistants available today — but they live in the terminal. For PMs, designers, and most non-engineers, the terminal is an invisible wall.

They only help if people can actually reach them.

There's another scenario: engineers on the go — commuting, in meetings, away from their laptop. An urgent issue needs an AI to investigate. Open Telegram on your phone, type a message, done.

The Solution

Telegram AI Bridge is a lightweight open-source relay: a Node.js process runs on your server, receives Telegram messages, forwards them to Claude Code or Codex CLI, and sends the results back to your chat.

From the user's perspective, it's just chatting with AI in Telegram. No dev tools to install, no terminal commands to learn. Type text, send voice, share a screenshot — the AI handles it all.

How It Works

Telegram  -->  Long Polling  -->  bot.js  -->  executor.js  -->  claude -p / codex exec
   ^                                                                      |
   +-------------- response (split at 4000 chars) <-----------------------+

bot.js

Command routing
Session tracking

executor.js

Process management
Output buffering

auth.js

User allowlist
Access control

config.js

Environment vars
Cross-platform config

Core Features

Multi-Backend Switching

/backend claude|codex — switch instantly. Claude Code has conversation memory; Codex has powerful code generation. Pick the right tool for the moment.

Voice & Images

Send voice messages — auto-transcribed and forwarded to AI. Share screenshots for visual context. When you can't type, just talk.

Session Management

Claude backend auto-tracks sessions — conversations never break. /sessions to list history, /resume to switch back.

Shell Access

/sh runs terminal commands directly on the server. Not just AI chat — remote machine management too.

Not at Your Desk? Still in Control.

Thought of a bug while commuting? Need AI to run an analysis during a meeting?

Open Telegram on your phone, type a few words, and the AI agent starts working on your server.
Results come back right to your chat —

Your AI assistant, on call 24/7.

~900

Lines of Code

3

Dependencies

2

AI Backends

3

Platforms Supported