nvk

nvk / agentnoise

Public

Chat with local coding agents through White Noise. agentnoise is a native desktop helper for using a phone running White Noise as the control surface for local Codex and Claude sessions. It is intentionally Rust-first and keeps Node/npm/bun out of the trusted bridge path.

agentnoise.org agent brew chat im mls
11
2
89% credibility
Found May 18, 2026 at 12 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

agentnoise is a desktop helper that lets you control local coding assistants entirely from your phone. You install it on your computer, pair your phone using a quick secure handshake, and then send commands through a chat interface. When you ask it to work on code, it launches a local agent on your machine, keeps you updated as it goes, and sends the results back to your phone chat. It was built because existing tools were too complicated or too slow for this simple workflow.

How It Works

1
📱 Download and install the app

You install agentnoise on your Mac or Linux computer using a simple package manager command.

2
🔐 Create your secure identity

The app creates a private identity just for your desktop helper and stores it safely in your computer's password vault.

3
📲 Pair your phone

You scan a quick code shown on your screen with your phone's White Noise app, then type a short code to prove it's really you.

4
💬 Start chatting from your phone

You open a chat with your desktop helper on your phone and send commands like '/codex fix this button'.

5
🤖 Your coding assistant springs to life

The desktop launches a coding agent that works on your project, sending you little updates as it goes.

6
📊 Watch the progress roll in

While the agent works, you see regular updates on your phone showing what it's doing and how far along it is.

✨ Results arrive on your phone

When the agent finishes, your completed work appears right in your phone chat, ready for you to review.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 12 to 11 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 agentnoise?

agentnoise is a Rust desktop helper that lets you control local AI coding agents through your phone using White Noise as the messaging layer. You send commands like `/codex fix this bug` or `/claude explain that function` from a White Noise chat, and agentnoise launches the appropriate agent locally, posts progress updates back to your phone, and delivers the final result. It keeps Node and npm out of the trusted execution path by design, using bondage as the recommended local policy boundary for sandboxing agent behavior.

Why is it gaining traction?

The hook is simple: your phone becomes a remote control for local coding agents without trusting a web interface or cloud service. The Rust-first architecture means the bridge between your phone and local agents is lean and auditable. Security-conscious developers notice the pairing PIN flow, OS keychain integration for secrets, and the explicit policy boundary that bondage provides. The fake-phone testing harness suggests the maintainer takes reliability seriously despite the early stage.

Who should use this?

This is for developers who want phone-controlled AI coding assistance with a local security boundary. If you run Codex or Claude Code locally and want to trigger jobs from your phone while keeping the execution sandboxed, this fills that niche. Security teams evaluating local agent isolation might find the bondage integration worth exploring. Most developers will prefer direct terminal usage, but if your workflow involves mobile-triggered coding tasks, this is one of the few tools built for exactly that.

Verdict

agentnoise solves a specific problem well, but the credibility score of 0.9% and 11 stars reflect a very early, niche project. The code quality and test coverage are solid for an alpha, and the security model is thoughtfully designed. If you need phone-to-local-agent control with a clean trust boundary, it is worth evaluating. For general use cases, wait for more maturity and community adoption.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.