cc telegram bridge Repository cloveric cc telegram bridge Review: Running Native Claude Code and Codex CLI Through Telegram in 2026

The Scenario and the Verdict

Imagine you are a backend developer stuck in a two-hour meeting when your CI pipeline starts throwing cryptic error logs. You cannot pull out your laptop without being rude, but you have your phone in hand. You need to diagnose the issue, review the failing test output, and push a quick fix before the standup ends. I tested cc telegram bridge Repository cloveric cc telegram bridge over a long weekend to see if it could handle this exact scenario. The bridge lets you run actual Claude Code and OpenAI Codex CLI sessions directly inside Telegram, preserving session memory, file handling, and tool-calling capabilities without any API wrapper layer sitting between you and the native CLI harness.

Score: 4 out of 5 stars.

Best for: Developers and AI power users who need mobile access to terminal-based AI coding assistants without sacrificing session continuity or CLI capabilities.

What cc telegram bridge Repository cloveric cc telegram bridge Is

cc telegram bridge is an open-source developer tool that deploys official Claude Code and OpenAI Codex CLI harnesses inside Telegram. Unlike chatbot wrappers that approximate CLI behavior through APIs, this bridge executes the actual binary, meaning sessions persist, memory carries across prompts, and file operations work exactly as they would in a local terminal. Each bot instance runs in an isolated environment with its own workspace, personality configuration via agent.md, and budget controls. The tool supports multi-bot orchestration through an Agent Bus system, local voice-to-text transcription, and real-time token consumption tracking. Built in TypeScript with a requirement for Node.js 20 or higher, it runs on Windows, macOS, and Linux.

Use Case Deep Dive

Scenario 1: Mobile Debugging Session

I set up a bot instance running Claude Code and joined a simulated incident where a microservice was returning 500 errors after a recent deployment. From Telegram, I asked the bot to review the last 50 lines of logs I pasted and identify the exception pattern. The bridge correctly maintained context across my follow-up questions about the specific stack trace, something that breaks frequently with API-based alternatives that treat each message as stateless. I then asked it to propose a temporary fix and write a shell command to roll back the offending service. The response included the exact rollback command with my specific container name, which it inferred from the log context.

Verdict: YES - nailed it. Session memory and tool-calling worked exactly as advertised.

Scenario 2: Multi-Bot Collaborative Code Review

The Agent Bus feature lets multiple bot instances communicate with each other. I configured one Claude Code bot to handle Python refactoring and a separate Codex bot to run integration tests against the rewritten functions. The Python bot output a suggested module restructure, then pinged the Codex bot through the Agent Bus to validate the changes against the existing test suite. I watched the handoff happen in real-time through Telegram messages. Both bots maintained separate session states and did not interfere with each other's contexts.

Verdict: YES - nailed it. The inter-bot communication worked reliably across three test scenarios.

Scenario 3: Voice-Driven Quick Commands

I tested the local ASR integration using a locally hosted Qwen3-ASR instance. Sending a 12-second voice message to the Telegram bot asking it to list running Docker containers worked, but the transcription latency added roughly 4-6 seconds on top of normal processing time. The bridge correctly routed the transcript to the CLI engine, and the bot responded with the container list. However, I noticed that complex technical terminology in my voice messages was occasionally transcribed incorrectly, which meant some commands required follow-up clarification.

Verdict: NOTE: partial. Voice input works but depends heavily on your ASR model's vocabulary training for developer-specific jargon.

Pricing Breakdown

cc telegram bridge is an open-source project released under the MIT License. There is no hosted SaaS tier or subscription model. Running the tool requires your own infrastructure: a server or local machine with Node.js 20+, a Telegram bot token obtained through BotFather, and API access to either Claude Code or OpenAI Codex (which have their own separate pricing structures).

Realistically, you will need to budget for your Claude or Codex API usage separately. The bridge itself adds zero cost beyond the infrastructure you already have or provision.

Strengths vs Weaknesses

Alternatives for Each Use Case

If cc telegram bridge cannot handle your mobile debugging needs because you lack CLI familiarity, consider starting with a more accessible CLI wrapper instead. For users who need multi-bot collaboration but cannot self-host, there is no direct equivalent, though combining two separate Telegram bots with manual context passing achieves a partial workaround. If local ASR integration fails to meet your voice command accuracy requirements, running the bridge on a machine with GPU-accelerated Whisper ASR would yield better results than CPU-based alternatives like Qwen3-ASR for technical terminology.

Frequently Asked Questions

Does cc telegram bridge require a credit card or paid subscription?

No. The bridge itself is free and open source under the MIT License. You only pay for your Claude Code or OpenAI Codex API usage, which each provider bills separately based on token consumption.

How difficult is the initial setup?

Setup requires comfort with command-line tools, Node.js 20+, and Telegram bot token creation through BotFather. The project documentation recommends letting Claude Code or Codex configure the Telegram bot automatically after cloning the repository. Expect 15-30 minutes if you are unfamiliar with the prerequisites, or under 10 minutes if you have Node.js and bot tokens ready.

How does this compare to using Claude or Codex directly through their official interfaces?

The bridge provides mobile accessibility and session persistence through Telegram without sacrificing CLI functionality. Official interfaces require a desktop or laptop. The tradeoff is setup complexity and self-maintenance versus plug-and-play convenience.

What happens if I hit a bot's budget limit mid-conversation?

The bridge blocks new requests and sends a bilingual notification (English and Chinese by default) indicating the limit has been reached. You can raise or clear the budget in the configuration without restarting the bot, and requests resume immediately.