New · Subscriptionfor vibecoders

Your AI writes the bugs. Guard catches them.

LeakRank Guard installs into your coding agent with one command and scans every change for leaked secrets and insecure config, then hands your agent the exact fix. No dashboard. No context-switch.

Install
1 · Copy this command into your terminal
npx leakrank-guard init
— or, if you would rather not touch the terminal —
2 · Paste this straight into Claude Code
Run `npx leakrank-guard init` to install LeakRank Guard, then scan this project for leaked secrets.

The installer detects Claude Code, registers Guard's scanner as an MCP tool, and drops in a rules file so your agent scans every change on its own. You paste your API key once; it is never stored in your project.

How it works
01

Pick your agent, copy one line

Claude Code, Cursor, Codex, or Windsurf. One command wires Guard in as an MCP tool and drops a rules file so scanning is automatic.

02

Your agent scans every change

After each edit it runs Guard on the diff. A fast static pass flags candidates; our servers run the AI triage, never your agent's tokens.

03

Get the fix, not just the flag

Guard returns the exact file and line, a fix your agent applies directly, and precisely which secrets you need to rotate.

Pricing

One plan. Priced for the AI it runs.

Most scanners charge flat per seat and eat the AI cost. Guard is honest about it: the subscription covers real model inference, so the analysis stays genuinely smart instead of a regex that pretends. Bought a Fix Pack in the last 30 days? What you paid is credited when you upgrade.

LeakRank Guard
$180/ year

$15/mo equivalent · 2 months free vs monthly

  • Continuous scanning inside your AI agent, no dashboard to babysit
  • Leaked secrets, committed .env files, wildcard CORS, exposed source maps
  • AI-written fixes your agent applies directly, plus what to rotate
  • 500 scan credits / month (a change scan is 1 credit)
  • Works with Claude Code, Cursor, Codex, and Windsurf

30-day money-back guarantee

FAQ

How is this different from the one-time scan?

The Fix Pack scans your live site once. Guard lives in your repo and scans continuously as you code, catching leaks the moment they are written, before they ship.

Does my code leave my machine?

Only the changed files (or a diff) are sent to LeakRank's scanner to analyze, over an authenticated API. We store masked evidence of findings, never raw secret values.

Who runs the AI analysis?

We do, on our own models and our own bill. Guard does not spend your Claude or Cursor usage. That is why it is a subscription: the price covers real inference cost.

What are credits?

Each change scan costs 1 credit; a full-repo baseline costs 10. Every plan includes 500 credits a month, which comfortably covers a normal solo-dev commit cadence.

Can I cancel?

Anytime, from the Paddle receipt email. Your key keeps working until the end of the paid period.