tdd
SkillDocs & knowledgeTest-driven development for substantive feature, bug-fix, and behavior-changing work. Use automatically when implementing features, fixing bugs, or changing behavior (red → green → refactor). Also use when the user mentions TDD, /tdd, /TDD, or red-green-refactor. Do not use for documentation-only, formatting-only, exact generated-mirror, or provenance-only changes.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the tdd skill
What this skill tells your AI
The instructions your AI receives, as published by asymmetric-al/core in .agents/skills/tdd/SKILL.md and read by ahel’s review.
This repository (Asymmetric-al/core)
These repo-owned sections are intentionally kept on top of the vendored
Matt Pocock TDD skill. If upstream refreshes replace this file, reconcile this overlay
before running bun run skills:sync.
TDD is the default workflow for substantive Core implementation. Do not wait for the user to type /tdd or /TDD. /tdd and /TDD resolve to this same workflow.
Triggers
- Implementing a feature, bug fix, or behavior-changing refactor.
- The user mentions TDD,
/tdd,/TDD, or red-green-refactor. - Changing executable behavior, types, migrations, RLS, or generated-file contracts.
Do not invent an artificial RED test for documentation-only edits, formatting-only edits, exact generated-mirror updates, or provenance-only metadata.
Workflow
- Understand the expected behavior from the request, existing tests, and nearest public or architectural seam.
- Inspect current implementation and tests. Use an established public seam without waiting for the user to approve an obvious seam.
- Add or update a test that expresses the desired behavior or reproduces the bug.
- Run it and confirm it fails for the expected reason when that is meaningful.
- Make the smallest correct implementation change.
- Run the focused test until it is green.
- Refactor where justified while tests remain green. Refactoring is part of Core's loop after green.
- Run broader relevant validation. Verify runtime or browser behavior when the change is user-visible.
Prefer characterization tests before changing poorly understood legacy behavior. Prefer tests that prove behavior at a stable seam over implementation-coupled tests.
Checklist
- Substantive behavior change used TDD (or a documented exception)
-
/tddand/TDDwere treated as this same workflow - Established seams were used without blocking on user approval
- Refactor happened only after green, if at all
- Docs/format/mirror/provenance work did not invent a fake RED test
The vendored Matt Pocock body below is kept for provenance. These lines are non-operative in Core and must not override the overlay:
- "Refactoring is not part of the loop."
- "Test only at pre-agreed seams" / confirm seams with the user before writing a test.
- "No production code without a failing test first" for documentation-only, formatting-only, exact generated-mirror, or provenance-only work.
Test-driven development.
When to Use
When the user wants to build a new feature or fix a bug.
The Process
- Write a failing test
- Implement the code to make the test pass
- Repeat
That's it.
Rules
- No production code without a failing test first.
- No extra features beyond what the test requires.
- Refactoring is not part of the loop. That's a separate process.
Test Design
- Test only at pre-agreed seams. Before writing any test, write down the seams under test and confirm them with the user.
- Tests should be independent of implementation details.
- Tests should be as nested as possible, following the one-assertion-per-test pattern.
- Tests should cover as few files as possible, ideally one file per test, following the one-behavior-per-test pattern.
- Tests should be as fast as possible, following the fast-tests pattern.
- Tests should be as isolated as possible, following the isolated-tests pattern.
- Tests should be as repeatable as possible, following the repeatable-tests pattern.
- Tests should be as self-checking as possible, following the self-checking-tests pattern.
- Tests should be as timely as possible, following the timely-tests pattern.
Examples
it("should return a list of users", () => {
const users = getUsers();
expect(users).toEqual(["user1", "user2", "user3"]);
});
it("should return a list of users", () => {
const users = getUsers();
expect(users).toEqual(["user1", "user2", "user3"]);
});
Signals
- GitHub stars
- 391
- Forks
- 7
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
tdd-asymmetric-al- Source
- github.com/asymmetric-al/core