Test-Driven Development
TDD is the red → green loop. This skill is the reference that makes that loop produce tests worth keeping: what a good test is, where tests go, the anti-patterns, and the rules of the loop. Every section applies on every cycle — consult them before and during the loop, not after.
When exploring the codebase, read CONTEXT.md (if it exists) so test names and interface vocabulary match the project's domain language, and respect ADRs in the area you're touching.
What a good test is
Tests verify behavior through public interfaces, not implementation details. Code can change entirely; tests shouldn't. A good test reads like a specification — "user can checkout with valid cart" tells you exactly what capability exists — and survives refactors because it doesn't care about internal structure.
See tests.md for examples and mocking.md for mocking guidelines.
Seams — where tests go
A seam is the public boundary you test at: the interface where you observe behavior without reaching inside. Tests live at seams, never against internals.
Test only at pre-agreed seams. Before writing any test, write down the seams under test and confirm them with the user. No test is written at an unconfirmed seam. You can't test everything — agreeing the seams up front is how testing effort lands on the critical paths and complex logic instead of every edge case.
Ask: "What's the public interface, and which seams should we test?"
Anti-patterns
- Implementation-coupled — mocks internal collaborators, tests private methods, or verifies through a side channel (querying the database instead of using the interface). The tell: the test breaks when you refactor but behavior hasn't changed.
- Tautological — the assertion recomputes the expected value the way the code does (
expect(add(a, b)).toBe(a + b), a snapshot derived by hand the same way, a constant asserted equal to itself), so it passes by construction and can never disagree with the code. Expected values must come from an independent source of truth — a known-good literal, a worked example, the spec.
- Horizontal slicing — writing all tests first, then all implementation. Bulk tests verify imagined behavior: you test the shape of things rather than user-facing behavior, the tests go insensitive to real changes, and you commit to test structure before understanding the implementation. Work in vertical slices instead — one test → one implementation → repeat, each test a tracer bullet that responds to what the last cycle taught you.
Rules of the loop
- Red before green. Write the failing test first, then only enough code to pass it. Don't anticipate future tests or add speculative features.
- One slice at a time. One seam, one test, one minimal implementation per cycle.
- Refactoring is not part of the loop. It belongs to the review stage (see the
code-review skill), not the red → green implementation cycle.
AKILI-SPECS Integration
In AKILI projects this skill runs inside the Implementer's inner loop, assigned per task — it changes how a task is built, never what the spec approved.
| Upstream rule |
How it lands here |
Assignment is the Leader's call (binding: conditional) |
The /akili-execute Leader assigns this skill in the brief for logic-heavy tasks (algorithms, business rules, data transformations, contract implementations). Never blanket: TDD on a copy tweak or a styling task is pure cost. When assigned, the Implementer loads it before writing any code. |
| Expected values come from the spec, not from the user mid-loop |
The anti-tautological rule demands an independent source of truth — in AKILI that source already exists and is already approved: the Given/When/Then scenarios in requirements.md named by the task. Derive each red test's expected value from its scenario; if a scenario is too vague to yield one, that is a spec gap to report to the Leader, not a value to invent. |
Seams are pre-agreed in design.md, not asked live |
The upstream "confirm seams with the user" question is already answered: the public interfaces in the spec's design.md (and the TRD's module boundaries) are the agreed seams — the Leader's brief points at them. Only when the design genuinely defines no interface for the task's area does the seam question go to the Leader (as a design ambiguity, per /akili-execute's error handling — never straight to the user). |
| Bug Mode is already TDD |
/akili-specify Bug Mode mandates a regression test red before the fix, green after — this skill generalizes that same discipline to feature tasks. On a bugfix task, the two are one loop: the mandatory regression test is the first red. |
| Refactoring belongs to review |
The upstream rule maps 1:1 onto the AKILI harness: the Reviewer's 4R lenses and the simplify pass own refactoring. The Implementer's loop stays red → green; do not gold-plate inside it. |
No duplication with /akili-test |
TDD tests are the author's tracer bullets at unit seams; /akili-test's Testers remain the independent proof (author ≠ tester): negative constraints, integration, E2E. Testers cite existing TDD coverage in their per-scenario matrix instead of rewriting it, and add what the author's loop does not cover. |
| Verification evidence |
The Implementer's completion report cites the red → green history as verification evidence: the test file(s), the scenario each test proves, and the final green run. A test that was never seen red is not TDD evidence — say so rather than implying it. |
One sentence to remember: the spec supplies the expected values, the design supplies the seams, the Leader supplies the assignment — the loop supplies only the discipline.
Authorship
TDD skill by Matt Pocock — github.com/mattpocock/skills, MIT License. Adapted for AKILI-SPECS by Juan Carlos Cadavid — jcadavid.com.
1---2name: tdd3description: Trigger: test-driven development, TDD, red-green, test-first implementation of a logic-heavy task, 'red-green-refactor'. Red → green loop with tests worth keeping: seams, anti-patterns (implementation-coupled, tautological, horizontal slicing), vertical slices. In AKILI-SPECS the Leader assigns it per task; expected values come from the approved requirements.md scenarios and seams from design.md.4license: MIT5---67# Test-Driven Development89TDD is the red → green loop. This skill is the reference that makes that loop produce tests worth keeping: what a good test is, where tests go, the anti-patterns, and the rules of the loop. Every section applies on every cycle — consult them before and during the loop, not after.1011When exploring the codebase, read `CONTEXT.md` (if it exists) so test names and interface vocabulary match the project's domain language, and respect ADRs in the area you're touching.1213## What a good test is1415Tests verify behavior through public interfaces, not implementation details. Code can change entirely; tests shouldn't. A good test reads like a specification — "user can checkout with valid cart" tells you exactly what capability exists — and survives refactors because it doesn't care about internal structure.1617See [tests.md](tests.md) for examples and [mocking.md](mocking.md) for mocking guidelines.1819## Seams — where tests go2021A **seam** is the public boundary you test at: the interface where you observe behavior without reaching inside. Tests live at seams, never against internals.2223**Test only at pre-agreed seams.** Before writing any test, write down the seams under test and confirm them with the user. No test is written at an unconfirmed seam. You can't test everything — agreeing the seams up front is how testing effort lands on the critical paths and complex logic instead of every edge case.2425Ask: "What's the public interface, and which seams should we test?"2627## Anti-patterns2829- **Implementation-coupled** — mocks internal collaborators, tests private methods, or verifies through a side channel (querying the database instead of using the interface). The tell: the test breaks when you refactor but behavior hasn't changed.30- **Tautological** — the assertion recomputes the expected value the way the code does (`expect(add(a, b)).toBe(a + b)`, a snapshot derived by hand the same way, a constant asserted equal to itself), so it passes by construction and can never disagree with the code. Expected values must come from an independent source of truth — a known-good literal, a worked example, the spec.31- **Horizontal slicing** — writing all tests first, then all implementation. Bulk tests verify _imagined_ behavior: you test the _shape_ of things rather than user-facing behavior, the tests go insensitive to real changes, and you commit to test structure before understanding the implementation. Work in **vertical slices** instead — one test → one implementation → repeat, each test a **tracer bullet** that responds to what the last cycle taught you.3233## Rules of the loop3435- **Red before green.** Write the failing test first, then only enough code to pass it. Don't anticipate future tests or add speculative features.36- **One slice at a time.** One seam, one test, one minimal implementation per cycle.37- **Refactoring is not part of the loop.** It belongs to the review stage (see the `code-review` skill), not the red → green implementation cycle.3839## AKILI-SPECS Integration4041In AKILI projects this skill runs **inside the Implementer's inner loop**, assigned per task — it changes *how* a task is built, never *what* the spec approved.4243| Upstream rule | How it lands here |44|---|---|45| **Assignment is the Leader's call (binding: `conditional`)** | The `/akili-execute` Leader assigns this skill in the brief for **logic-heavy tasks** (algorithms, business rules, data transformations, contract implementations). Never blanket: TDD on a copy tweak or a styling task is pure cost. When assigned, the Implementer loads it before writing any code. |46| **Expected values come from the spec, not from the user mid-loop** | The anti-tautological rule demands an independent source of truth — in AKILI that source already exists and is already approved: the **Given/When/Then scenarios in `requirements.md`** named by the task. Derive each red test's expected value from its scenario; if a scenario is too vague to yield one, that is a spec gap to report to the Leader, not a value to invent. |47| **Seams are pre-agreed in `design.md`, not asked live** | The upstream "confirm seams with the user" question is already answered: the public interfaces in the spec's `design.md` (and the TRD's module boundaries) **are** the agreed seams — the Leader's brief points at them. Only when the design genuinely defines no interface for the task's area does the seam question go to the Leader (as a design ambiguity, per `/akili-execute`'s error handling — never straight to the user). |48| **Bug Mode is already TDD** | `/akili-specify` Bug Mode mandates a regression test **red before the fix, green after** — this skill generalizes that same discipline to feature tasks. On a bugfix task, the two are one loop: the mandatory regression test is the first red. |49| **Refactoring belongs to review** | The upstream rule maps 1:1 onto the AKILI harness: the Reviewer's 4R lenses and the `simplify` pass own refactoring. The Implementer's loop stays red → green; do not gold-plate inside it. |50| **No duplication with `/akili-test`** | TDD tests are the *author's* tracer bullets at unit seams; `/akili-test`'s Testers remain the **independent** proof (author ≠ tester): negative constraints, integration, E2E. Testers cite existing TDD coverage in their per-scenario matrix instead of rewriting it, and add what the author's loop does not cover. |51| **Verification evidence** | The Implementer's completion report cites the red → green history as verification evidence: the test file(s), the scenario each test proves, and the final green run. A test that was never seen red is not TDD evidence — say so rather than implying it. |5253One sentence to remember: *the spec supplies the expected values, the design supplies the seams, the Leader supplies the assignment — the loop supplies only the discipline.*5455## Authorship5657TDD skill by **Matt Pocock** — [github.com/mattpocock/skills](https://github.com/mattpocock/skills), MIT License. Adapted for AKILI-SPECS by **Juan Carlos Cadavid** — [jcadavid.com](https://jcadavid.com).
JuankCadavid