# Tdd

> Red-green-refactor loop for new behaviour, bugs, or regressions.

- Skill: `darknesskiller/tdd` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add darknesskiller/tdd`
- Raw SKILL.md: https://api.skillmd.com/api/skills/darknesskiller/tdd/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: darknesskiller (https://skillmd.com/u/darknesskiller)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/darknesskiller/tdd

---


# TDD

## Loop

1. **Red** — write one failing check for the next behaviour slice.
2. **Green** — make the smallest implementation pass.
3. **Refactor** — simplify while the check stays green.
4. Repeat until the contract is covered.

## Output

Name every phase: `Red`, `Green`, `Refactor`, `Repeat`. Finish with: `Passing check`, `Normal command`, `Untestable` (none or remaining edge).

Completion: every requested behavior has a passing check, the check ran in the repo's normal command, and any untestable edge is reported.

## Test choice

Prefer the highest-level test that proves the behaviour without brittle setup. HTTP paths: request/response tests. Service/repository: integration tests when practical. Go: table-driven with `testify/require` if project uses it. Keep mocks at boundaries. Do not mock what you can run cheaply. One `assert`-style self-check suffices for tiny cases.