What problem does it solve? Executing a multi-phase technical plan with test-first discipline is hard to sustain manually: tests get written after the code, phases blur together, and progress tracking drifts from reality. This Skill drives an approved plan from context/changes/<change-id>/plan.md through strict red-green-refactor cycles, one phase at a time, with enforced gates and per-phase commits. ## Core Features & Use Cases - Phase-by-phase TDD execution: Runs RED (failing test), GREEN (minimal implementation), REFACTOR (cleanup) loops per behavior, with 2-5 focused tests per phase. - Eligibility gating: Before each phase, checks that implementation is absent and the phase is genuinely TDD'able; redirects scaffolding, wiring, or already-implemented phases to /10x-implement. - Shared progress state: Mutates the same ## Progress section in plan.md as /10x-implement, so phases can be interleaved between the two skills without losing state. - Commit ritual: Enforces green-suite-only commits, explicit path staging, Conventional Commits messages, and SHA write-back into Progress rows. - Use Case: You have an approved plan for an OAuth login change with four phases. Invoke the skill with the change-id, and it writes a failing test for phase 1's token validation, implements the minimal code, refactors, commits, then gates phase 2 for TDD-ability before continuing. ## Quick Start Ask the AI to run /10x-tdd with your change-id, for example "/10x-tdd oauth-login phase 1", to start driving that plan test-first.