Muzzle

A firewall for your AI coding agents.

Muzzle stops Claude Code, Codex and Cursor from sending out your email, keys and secrets, and catches pages and files that try to take them over.

Download for Mac

Free · Apple Silicon · macOS 12 or later

Four checks, all on your Mac

Inside Claude Code and Codex

Commands and files are checked before they run or get written.

On every commit

A commit that would leak doesn't go through.

On what your scripts send

A local proxy checks the requests your scripts actually make.

Muzzle's Destinations table: leaks to webhook.site, an ngrok tunnel, pastebin.com and transfer.sh, with registry.npmjs.org, pypi.org and github.com trusted.

Across your projects

At setup, a scan of your code shows what already leaks.

Muzzle's Code view: 4,080 files scanned, an email found in a User-Agent and in a VAPID subject.

Your deploys keep working

Your own platforms are trusted

Pushing to GitHub, deploying to Vercel or publishing to npm goes through as usual.

  • github.com
  • vercel.com
  • api.cloudflare.com
  • registry.npmjs.org
  • netlify.com
  • gitlab.com
  • pypi.org
  • fly.io
  • supabase.co
  • docker.io

Keys reach their own service

A Stripe key going to Stripe passes. The same key going anywhere else is blocked.

sk_live_…api.stripe.comAllowed
sk_live_…webhook.siteBlocked

You decide how strict

Every level can be changed. For the first week Muzzle only watches, then shows you what it would have blocked.

EmailBlock
PhoneBlock
API keysBlock
Full nameWarn

Pro

A contact address for everything that asks for one

Crawlers, APIs and push keys often want a contact email. Muzzle gives you a personal address that forwards to you, and when it blocks your real email it tells the AI to use this one.

reach-7kq2m9@in.usemuzzle.com forwards to your inbox

The Muzzle Threat Feed. Every Mac makes the others safer.

Macs that contribute to the Threat Feed report what they block. It becomes a signed rules list that every Muzzle install downloads daily.

Seen on

3Macs

Flagged on all

Cleared on

10Macs

Warns instead of blocks

Trusted on

25Macs

Suggested to you

What does the Threat Feed collect?

Pricing

Free

$0

  • Claude Code protection
  • Git commit check
  • Project scan
Download for Mac

Pro

$8/month, billed $96 yearly

$19/month

Free when you contribute to the Threat Feed

  • Everything in Free
  • Codex and Cursor
  • Scripts through the local proxy
  • Injection alerts
  • Contact address
Get Pro Get Pro

Enterprise

Custom

  • Everything in Pro
  • Volume licenses
  • Invoice billing
  • Help with rollout
Get a quote

Questions

Which tools does it work with?

Claude Code and git on Free. Pro adds Codex, Cursor and any script or program that sends requests through the local proxy.

What does it look for?

The email, phone number and name you enter at setup, plus API keys and tokens from Stripe, AWS, GitHub, OpenAI, Anthropic, Slack, Google, Supabase and others, and private keys.

Does it slow anything down?

The checks run on your Mac, and traffic to your trusted platforms passes straight through.

What leaves my Mac?

The checks run on your Mac and the app needs no account. The contact address needs your real email to forward mail to, and the Threat Feed collects only what's listed below.

What does the Threat Feed collect?

Which sites got blocked requests (the domain only), which checks fired, which false alarms you cleared, and fingerprints of injection attempts. Never your values, files, paths or code.