stillhunter

skill
Security Audit
Pass
Health Pass
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 43 GitHub stars
Code Pass
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

A Claude Code plugin that hunts bugs in a web app change: tests it in a real browser, proves each bug twice, and keeps it as a test.

README.md

Stillhunter

Stillhunter hunting a change to a watering app: the app splits into glass layers, a lens faults its requests with a 500, a slow reply and a malformed reply, then types -5 into the litres field. The total turns to -5 L, the bug is reproduced 2 of 2 times, confirmed as MEDIUM and kept as a test.

Find the bugs in your change before your users do. Stillhunter is a Claude Code plugin that tests your web app change in a real browser, the way a careful QA engineer would. It reports only bugs it has seen happen twice, and you can keep each one as a test.

Install

Inside Claude Code:

/plugin marketplace add stillhunter/stillhunter
/plugin install stillhunter@stillhunter

You also need Node.js 22.22 or later and Chrome, Edge or Chromium.

Use

  1. Start your app on your machine, as you do when you work on it.

  2. Tell Stillhunter where it runs (skip this for a Nuxt app; it is found for you). Create .stillhunter/config.json:

    { "target": { "url": "http://localhost:3000" } }
    
  3. Hunt. Make your change on a branch, then run:

    /stillhunter:hunt
    

    You can also name what to look at: /stillhunter:hunt the invite dialog. About 10 minutes later you get a list of confirmed bugs, each with a script that shows it.

  4. Keep a bug as a test, so you know when it is fixed:

    /stillhunter:keep 1
    /stillhunter:replay
    

Does your app need a sign-in? Set the test account in your shell before you start Claude Code, for example APP_EMAIL and APP_PASSWORD.

Use a local app with test data only: Stillhunter runs code the model writes against it.

Learn more

  • Guide: all commands and settings, how a hunt works, and how probes are contained
  • Benchmark: found the planted bug in 17 of 18 runs, with 0.41 false alarms per run
  • Changelog · Security · MIT licence

Reviews (0)

No results found