tuchg

tuchg / Lucarne

Public

Stop babysitting local AI agents. 0-Intrusion mobile control via Telegram/WeChat. No hooks, no skills, no MCP. Just sync, approve, and resume your Codex,Pi, or Claude sessions anywhere.

agent-im agent-management agent-sessions ai-agents claude-code
15
0
89% credibility
Found May 20, 2026 at 15 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

Lucarne is a lightweight background service that bridges local AI coding assistants (Claude, Codex, Copilot, Gemini, Cursor, and Pi) to your existing messaging apps (Telegram and WeChat). Instead of babysitting your AI agent at the computer, you receive real-time notifications on your phone whenever the agent needs permission, asks a question, or completes important work. You can approve or deny actions directly from your phone, and even continue conversations by quoting messages in your messaging app. The daemon runs locally on your Mac, requires no project changes or hooks, and uses minimal memory when idle.

How It Works

1
💻 Install the background helper

You install Lucarne on your Mac using Homebrew - it runs quietly in the background on your computer.

2
📱 Connect your messaging apps

You optionally connect Telegram or scan a WeChat QR code so your phone can receive updates from your AI assistants.

3
🤖 Start your AI coding assistant

You run Claude, Codex, Copilot, or another supported AI agent on your computer as usual - Lucarne watches everything automatically.

4
🔔 Get real-time updates on your phone

When your AI assistant wants to run a command or needs your approval, you instantly see a notification on Telegram or WeChat.

5
Choose how to respond
✅
Approve or deny

Tap approve or deny buttons to let the AI continue or stop its current task

💬
Continue talking

Quote a notification and reply to keep the conversation going with full context restored

🎉 Work freely while staying in control

You can step away from your computer knowing you'll be notified of anything important - your AI assistant keeps working while you stay informed.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 15 to 15 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is Lucarne?

Lucarne is a Rust daemon that bridges your local AI coding agents to your phone. Instead of sitting at your terminal watching Claude, Codex, or Copilot work, you get real-time updates and can approve or deny commands through Telegram or WeChat. No new mobile app required. Setup is a QR code scan and a single init command. Once running, agents notify you when they need permission, and you can reply to continue context without touching your keyboard.

Why is it gaining traction?

The pitch is simple: stop babysitting your terminal. Developers running long agent sessions often feel tethered to their desk, afraid to step away while the agent executes potentially destructive commands. Lucarne solves this by routing permission requests and progress updates through chat apps developers already live in. The quote-to-reply feature in WeChat is particularly clever—you can reference a specific notification to resume that exact conversation thread. Telegram users get a full console with workspace management and branching session support.

Who should use this?

Backend engineers running Claude Code or Codex on complex refactors will appreciate mobile oversight without terminal switching. Teams with security-conscious workflows can require human approval for file writes or shell commands without forcing developers to stay glued to their machines. Anyone tired of context-switching between an agent session and their phone will find the workflow natural.

Verdict

Lucarne solves a real pain point with a clean architecture and smart channel choices. The credibility score of 0.8999999761581421% reflects an extremely early project—15 stars and macOS-only support mean production adoption carries real risk. Test coverage exists but is not comprehensive, and Linux/Windows support is still on the roadmap. Worth watching, but wait for 1.0 unless you are comfortable being an early adopter.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.