xplisted

# XP — Extreme Programming with AI Agents XP adapted for AI-human pairs: continuous code review (live pairing), relentless testing (TDD), constant design improvement (refactoring), frequent releases (small increments). See [philosophy.md](references/philosophy.md) for foundational values. ## When to Invoke - **Pair dev** — User wants structured iteration with tests + review (no big-bang implementation) - **Incremental features** — Break work into 5-min cycles; each cycle: plan → test → code → refactor - **Roles unclear** — Need to signal who drives (direction) vs. navigates (reviews); see [roles.md](references/roles.md) - **Refactor needed** — Existing code quality degradation; XP treats refactoring as continuous, not deferrable - **Not a fit** — Stop if: user wants a quick script (no test overhead), one-off task (no iteration), or user is not engaged for review ## Workflow (One Cycle) ### 1. Plan — Pick ONE Small Task Define a single, deliverable piece of work. Confirm with human: - **What** (acceptance criteria)? - **Why** (business value)? - **How** (constraints, conventions)? **Done when:** Human approves the task before coding starts. ### 2. Test — Write One Failing Test Write a test that describes behavior, not implementation. It must fail first. See `/tdd` for full red-green-refactor discipline. **Done when:** Test runs, fails predictably, human reviews & approves the test. ### 3. Implement — Minimal Code to Pass Write the simplest code that makes the test