tlive
Self-hosted monitoring and remote approvals for Claude Code and Codex via Telegram, Feishu or a web terminal
Install
npm install -g tlive
tlive setup
tlive startThe wizard registers the plugin in Claude Code and Codex and configures the messenger.
This is third-party code. Review the repository files before installing.
What it does
tlive hooks into Claude Code and the Codex companion and pushes session status, finished turns and waiting prompts to Telegram or Feishu. By default it only watches and notifies; in full mode it holds tool calls so you can allow or deny them from your phone, while the local dialog stays live too. Launching the agent through tlive run serves a web terminal and a session dashboard from your machine that you can type into and upload files to from any device. For Claude Code it installs a plugin with hooks, a skill and the /tlive:url and /tlive:status commands.
Who it is for. For developers who run long Claude Code or Codex sessions and want to follow and answer them away from the computer.
Good fit when
- The agent runs for a long time and you want to know when it finishes or waits
- You need to approve risky agent commands from your phone
- You want to continue a session from a browser on another device
Not a fit when
- Session content and diffs must not reach a messenger
- You are always at the keyboard and do not need notifications
Example request
Help me configure tlive: I want Telegram notifications when a session finishesLimitations
Works only with Claude Code and Codex; some features, such as answering agent questions and failure alerts, exist only for Claude Code. Notifications need a Telegram or Feishu bot. Version 2.0 is incompatible with older configs. The all mode removes the local dialog for sub-agents until the wait window ends.
How to disable. Turn off all hooks temporarily with tlive mode off. Full removal: npm uninstall -g tlive, which removes the plugin via the agents' CLIs and keeps ~/.tlive.
Security check
- Lets you approve agent commands remotely
- Serves a web terminal with shell access
- Forwards session excerpts and diffs to a messenger
README in short
The README covers npm installation, the setup wizard and starting the daemon with a QR code for your phone. It explains two integration levels, hooks only and the tlive run wrapper with a web terminal, plus the off, notify, full and all modes, approvals, answering questions and sending files to the agent. It notes that installation uses the Claude Code and Codex plugin managers rather than editing their configs. MIT licensed.
FAQ
Can tlive hang and block the agent?
Not in the default notify mode: it never holds calls. Holding is enabled only in full and all modes, and an unanswered card denies nothing.
Does data go to third-party servers?
Sessions and the dashboard run on your machine; only messages to your chosen messenger leave it.
Related
Open-source personal AI assistant on your own machine: answers in Telegram, Slack, Discord and WhatsApp, extended with skills and plugins
A self-improving agent from Nous Research with a TUI, messaging gateway, cron jobs and skills it writes itself
An open source coding agent for the terminal and desktop with build and plan modes
Anthropic's coding agent for the terminal, IDE and GitHub: understands your codebase, runs tasks and handles git