01
It starts by itself
test fails on main. Runs in GitHub Actions or Claude Code routine.
- ↻ test fails on main
workflow_run.conclusion == failure on main
tw93 · 15K installsHunt skill
Runs by itself, test fails on main.
Find the cause before fixing.
Install the Hunt skill and make it a habit: follow https://gethabit.dev/skills/tw93/hunt.md
- 1Copy
- 2Paste into Claude Code, Codex or Cursor
- 3It runs by itself
15K installs7K starsMIT licenceSource on GitHub ↗
Hunt
by tw93 · 15K installs
The skill
The hunt skill finds the root cause of an error, crash or regression before any fix, then proves the fix with a repro and a regression test.
Only when you ask.
The habit
Find the cause before fixing.
↻ Runs by itselftest fails on main
How the habit works.
You don't have to.
When does it start? How does it check its work? What does it remember?
One run, step by step.
Then it stops.
Stops after one failure per run; after three failed hypotheses it hands off as blocked.
- 01
Collect symptoms
Your agent lists every observed symptom and builds a small repro, reliably for timing issues.
- 02
State the cause
It writes one sentence that names the file, function, line or condition, with the evidence.
- 03
Confirm or discard
It runs the one probe that would fail if the idea were wrong, and drops a disproven idea.
- 04
Bisect if needed
For something that used to work, it reads git diff first and bisects in a temporary worktree when the tree is dirty.
- 05
Fix in scope
Only when a fix was asked for, it changes the cause, never the symptom, and does not commit or push on its own.
- 06
Sweep siblings
It runs grep -rn for the same pattern and marks each match as same bug, safe, or unsure for you.
- 07
Guard and report
A regression test runs red then green, and the output block lists cause, fix, sweep and tests.
What it may do.
Nothing more.
It writes only in its own folder and asks you before anything goes out.
- Writes only insrc/**, tests/**, loops/hunt-root-cause-loop/**
- Talks tono outside service
- Sendsnothing
- Your OKIt asks you before anything goes out.
Is it for you?
Use it when it fits.
Read both lists. If the left one sounds like you, the Hunt skill helps.
Use it when
- ✓ Something errors, crashes or regressed
- ✓ A bug came back after a fix
- ✓ Something used to work after an update
Skip it when
- – You want a code review
- – You are building a new feature
Hunt skill, answered.
Does it fix code by itself?
Only when the request asks for a fix. Words like diagnose or investigate give a report only, and a fix never includes commit, push or publish.
What happens if it cannot find the cause?
After three failed hypotheses it stops and writes a handoff with what was checked, what was ruled out and what is still unknown.
Why the sibling sweep?
The skill says the same bug shape often sits in other places, so it greps for the pattern before it calls the bug done.
How do I install the Hunt skill?
Copy the prompt at the top into your agent. It installs the skill from tw93/Waza and sets up the habit. Or run npx skills add tw93/Waza --skill hunt.
Is the Hunt skill free?
Yes. It is published under MIT. Habit is free too.
Hunt. Now as a habit.
Install the Hunt skill and make it a habit: follow https://gethabit.dev/skills/tw93/hunt.md
All tw93 skills →More code skills →
Works in
- Claude Code
- Codex
- Cursor
- ChatGPT
- Claude app
Made by tw93, published under MIT. Habit links to the original and never changes it. Is this your skill?