本文へ移動
cccskills
無料GitHub で公開

tdd

Use test-driven development for behavior-changing feature or fix work, and whenever the user mentions TDD, test-first, red-green-refactor, tracer bullets, integration tests, or public-interface behavior tests. Skip for docs-only, path-only rename, formatting-only, or purely mechanical chores unless explicitly requested.

インストール方法を見る

含まれるファイル(6)

  • SKILL.md3.0 KB
  • references/deep-modules.md619 B
  • references/interface-design.md991 B
  • references/mocking.md1.2 KB
  • references/refactoring.md545 B
  • references/tests.md1.6 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Test-Driven Development

Core Rule

Work in vertical red-green-refactor slices:

  1. RED: write one test for one observable behavior.
  2. GREEN: implement the smallest production change that makes that test pass.
  3. REFACTOR: clean up only while tests are green.

Do not write a batch of imagined tests before implementation. Each new test should respond to what the last cycle taught you about the actual behavior and interface.

Plaited Defaults

  • Use bun test for targeted tests and bun --bun tsc --noEmit for the type gate.
  • Use test, not it, and organize with describe.
  • Prefer real dependencies and real runtime boundaries over mocks.
  • Mock only external boundaries such as remote APIs, time, randomness, or rare filesystem cases.
  • Avoid conditional assertions: assert the branch first, then assert branch-specific values.
  • Test both branches for conditionals, fallbacks, and error paths.
  • Exercise public interfaces and runtime contracts rather than private helpers.
  • Follow applicable repo skills and AGENTS.md rules for the touched area.

When Starting Work

Explore the codebase first. If the public interface, expected behavior, or risk boundary is discoverable from code, tests, or docs, use that evidence instead of asking the user.

Ask one concise question only when the decision cannot be discovered and a reasonable assumption would be risky. Include your recommended answer.

Before writing production code, identify:

  • the public interface or runtime boundary under test
  • the first observable behavior to prove
  • the targeted test command for the cycle
  • the broader validation needed before handoff

Cycle Discipline

For each behavior:

  1. Add or update exactly one focused test.
  2. Run the targeted test and confirm it fails for the intended reason.
  3. Implement only enough code to pass.
  4. Run the same targeted test and confirm it passes.
  5. Refactor if useful, then rerun the affected tests.

If the test passes before production code changes, it is not a valid RED signal. Tighten the test, choose a different behavior, or explain why existing coverage already proves the behavior.

Reference Files

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Analyze graded `trial_result` JSONL outputs from the eval pipeline, including reliability and custom slicing.

日本語の概要は準備中です。原文の説明を表示しています。

behavioral-sh/agent-eval-harness62026年6月18日 更新

grill-me

無料

Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".

日本語の概要は準備中です。原文の説明を表示しています。

behavioral-sh/agent-eval-harness62026年6月18日 更新

Build command-only adapters for `eval` run mode using strict stdin/stdout JSON contracts.

日本語の概要は準備中です。原文の説明を表示しています。

behavioral-sh/agent-eval-harness62026年6月18日 更新

End-to-end eval-suite orchestration with the `eval` command: run -> grade -> compare -> calibrate, using strict JSON contracts and schema discovery.

日本語の概要は準備中です。原文の説明を表示しています。

behavioral-sh/agent-eval-harness62026年6月18日 更新

behavioral-sh のスキルをすべて見る

このスキルの問題を報告する