agentleFS
NirDiamantNirDiamant/Agentic_Engineering/skills/tester/SKILL.md

tester

Skill34 starsChanged 52 days agoApache-2.0View on GitHub

Try to fail what just changed. The scoped, everyday complement to /testify. Reads the conversation or PR diff, picks the few behaviors most worth attacking (the art of covering enough without over-testing), writes tests that attempt to break them on the existing runner, summarizes each attempt with severity, and updates docs/TESTING.md. Trigger on "/tester", "test this", "add tests for this change", "cover this PR", "try to break this".

11 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 11 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Follow the runner and conventions from Infrastructure.

    line 34
  2. Distinctive:

    Test behavior through public entrypoints, not internals.

    line 35

“Common” is measured: kept by at least 6 in 10 of the 11 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Tester (test the change, not the world)

Where /testify stands up a project's whole suite, tester covers one change: the diff or the work discussed in this session. The few tests that matter, then stop.

Scope

What this conversation / PR changed. Not the whole repo, not existing untested code: that's /testify's job.

Infrastructure

docs/TESTING.md is the source of truth: read it first for the runner, conventions, and what's already covered. If it doesn't exist, detect the setup from the repo. No runner at all? Wire the minimal one for the stack (pytest / Vitest / Playwright, enough to run tests and nothing more), or hand off to /testify if the project deserves a real stand-up.

Select

Rank the changed behaviors by what would hurt most if silently broken, and test only the top of that list, typically 3–7 tests:

  • Core behavior: the thing the change exists to do, exercised through its real entrypoint.
  • Edge that plausibly breaks: boundaries, empty/invalid input, the bug's exact repro (a fix gets a test that fails without it).
  • Contract others rely on: response shape, return type, side effect.

Skip: trivial code (getters, config, framework glue), unreachable errors, permutations of the same logic, anything a type checker already guarantees.

Write

  • Follow the runner and conventions from Infrastructure; don't add frameworks beyond what that step wired.
  • Test behavior through public entrypoints, not internals. Mock only what's genuinely external (network, clock, paid APIs).
  • Attack, don't confirm: each test is an attempt to fail the code (hostile input, boundary, wrong order). Passing means the code survived the attempt.
  • One assertion story per test; name it after the behavior.

Finish

Run the tests, then hand-mutate the single most important behavior: break the code, watch a test go red, revert (a test that stays green on broken code is theater). Summarize results, one line per attempt: what you tried to fail, whether it broke, how it fails, and severity (critical / major / minor). A red test on a real bug is a finding: report it with its severity, don't quietly patch code the user didn't ask to change. Update docs/TESTING.md with what's now covered, creating it if absent, noting what you deliberately didn't test.

Self-edit rule

THIS SKILL SHOULD BE AS MINIMALISTIC AS POSSIBLE. Before writing any change or making any edit, ask: how can I do this with as few words as I possibly can?

Applies to every future edit. New instruction? Find the line it belongs in and rewrite that line. Do not append. Fixing a failure? Cut the wording that allowed it rather than adding a warning next to it. Skill grew? It regressed.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.