What problem does it solve?
Enforces disciplined test-driven development and per-task implementation workflows so production code is never introduced without a failing test first, while coordinating automated and human reviews to ensure correctness and safety.
Core Features & Use Cases
- Per-task TDD Orchestration: Spawn an implementer subagent for each task to write failing tests, implement minimal code to pass them, and perform self-review before committing.
- Reviewer Gatekeeping: Run spec, code-quality, silent-failure, and security reviewers (always) and optional thoroughness reviewers (deep mode) with clear execution order and gating rules.
- Quick Fix and Full Pipeline Modes: Support quick-fix single-task flows and full worktree-based feature flows with explicit artifact gating based on pipeline type.
- Artifact Tracking & Reporting: Produce per-task review artifacts and status codes (DONE, DONE_WITH_CONCERNS, NEEDS_CONTEXT, BLOCKED) and enforce review/fix loops with limits.
Quick Start
Orchestrate TDD for a dispatched task by running the implement skill to spawn an implementer subagent that writes a failing test, implements the minimal fix until tests pass, and dispatches configured reviewers.