Claude Code Telegram
Remote conversations with a codebase
Claude Code Telegram is a community-maintained Telegram bot that connects to Claude Code. The project repository describes asking it to analyze, edit, or explain a codebase through messages, with session persistence for each project. It is a separate integration project, not an Anthropic-owned Telegram product or a replacement for the underlying coding service.
The documented default mode accepts conversational requests. A classic mode provides commands for navigating directories, managing sessions, running quick actions, and working with repositories. Verbosity controls change whether Telegram shows only a final response or additional tool activity. Those controls affect the presentation of the conversation, not the authority granted to the agent.
Setup and operating scope
The stated prerequisites are Python 3.11 or later, the Claude Code CLI, and a Telegram bot token. Configuration also requires the bot username, an approved project directory, and allowed Telegram user IDs. The repository recommends installing a tagged release rather than tracking the development branch.
The feature list includes file and screenshot uploads, optional voice transcription, session export, usage tracking, and SQLite persistence. Transcription choices include external providers and a local option; enabling a particular option has different processing and credential implications. This overview does not assume all message content stays on the host machine.
Webhook, scheduler, and notification features extend the bot beyond interactive chat. The documentation lists GitHub event handling and recurring jobs, with separate settings needed to enable the API server and scheduler. These are configurable capabilities rather than features that every installation automatically runs.
Permissions and review
The publisher documents allowed-user authentication, directory sandboxing, rate limits, tool allowlists, webhook authentication, and audit logging. Those are stated controls, not an independent security audit. Review the configured user list, directories, tools, and webhook secrets before exposing a bot or allowing it to change a repository.
A first evaluation can use a non-sensitive test project and a read-only question, followed by an inspection of the host logs and Telegram output. Keep review of proposed code changes in the normal development workflow. The repository is MIT-licensed, but that does not make underlying model usage or hosting free. Provider authentication and usage costs remain separate from the bot's open-source license.