1. Habit
  2. Skills
  3. Test Harness skill
zernie · 2K installs

Test Harness skill
Runs by itself, harness files changed.

Hooks and skills always tested.

Install the Test Harness skill and make it a habit: follow https://gethabit.dev/skills/zernie/test-harness.md

  1. 1Copy
  2. 2Paste into Claude Code, Codex or Cursor
  3. 3It runs by itself

2K installs15 starsMIT licenceSource on GitHub ↗

Test Harness

by zernie · 2K installs

The skill

The test harness skill helps Claude Code users test hooks, skills, agents, settings and CLAUDE.md with vigiles, picking unit, deterministic or eval tiers and keeping eval results honest in CI.

Only when you ask.

The habit

Hooks and skills always tested.

↻ Runs by itselfharness files changed

How the habit works.
You don't have to.

Three questionsWhen does it start? How does it check its work? What does it remember?

01

It starts by itself

harness files changed. Runs in GitHub Actions.

  • ↻ harness files changed
  • push touches .claude/, hooks/, skills/ or CLAUDE.md

02

It checks its work

harness tests and locks pass. If a check fails, it tells you.

  • ✓npx vigiles test --no-skip passes
  • ✓vigiles eval --check finds no stale lock
  • ✓each changed hook has at least one runHook test

03

It remembers

eval lock files. The next run reads it first: recorded eval results and the inputs they were hashed from.

.vigiles/eval-locks/

One run, step by step.
Then it stops.

Stop ruleStops after one harness surface per run; real-model evals run only locally, never in CI.

  1. 01

    Pick the tier

    Choose unit, deterministic or eval from the tier table, cheapest first.

  2. 02

    Install vigiles

    Add vigiles as a dev dependency if it is missing.

  3. 03

    Find the surface

    Check .claude/settings.json, .claude-plugin/plugin.json, hooks/hooks.json and skills/.

  4. 04

    Write the test

    Follow references/writing-tests.md for the chosen tier.

  5. 05

    Run in CI

    Run npx vigiles test --no-skip so skipped tiers fail.

  6. 06

    Keep locks fresh

    After a local eval, run vigiles eval --update and commit the lock; CI runs --check.

What it may do.
Nothing more.

Safe by defaultIt writes only in its own folder and asks you before anything goes out.

Is it for you?
Use it when it fits.

Quick checkRead both lists. If the left one sounds like you, the Test Harness skill helps.

Use it when

  • ✓ You check that a hook blocks what it should
  • ✓ You test that a skill triggers
  • ✓ You confirm injected context reaches the model

Skip it when

  • – You test application code
  • – You do not use Claude Code

Questions

Test Harness skill, answered.

Why does every prompt show 0% triggering?

The skill says to rule out wiring mistakes first: bare ids, pluginDir versus skillsDir, and a missing fixture.

Do unit tests need the claude CLI?

No. runHook tests run without claude or an API key.

How do I install the Test Harness skill?

Copy the prompt at the top into your agent. It installs the skill from zernie/vigiles and sets up the habit. Or run npx skills add zernie/vigiles --skill test-harness.

Is the Test Harness skill free?

Yes. It is published under MIT. Habit is free too.

Test Harness. Now as a habit.

Install the Test Harness skill and make it a habit: follow https://gethabit.dev/skills/zernie/test-harness.md

  • Free
  • No account
  • Undo in one line

All zernie skills →More code skills →

Works in

  • Claude Code
  • Codex
  • Cursor
  • ChatGPT
  • Claude app

Made by zernie, published under MIT. Habit links to the original and never changes it. Is this your skill?

Copying didn't work here. Select the line and copy it by hand.