What problem does it solve? Feature work often drifts from requirements, ships without verification, or loses context between sessions. This Skill enforces a spec-driven workflow where requirements are written in testable EARS notation, tasks are atomic with Conventional Commits, and an independent Verifier confirms the implementation matches the spec before work is declared done. ## Core Features & Use Cases - Adaptive 4-Phase Workflow: Specify and Execute are always required; Design and Tasks auto-scale based on feature complexity, from one-liner specs to full architecture documents. - Deterministic Validation Gates: Python scripts validate spec closure, task granularity, Conventional Commits format, and completion state so structural rules are enforced by code rather than memory. - Independent Verification: A fresh Verifier sub-agent (author != verifier) runs a spec-anchored outcome check and a discrimination sensor that injects behavior-level faults to confirm tests actually catch regressions. - Project Memory & Lessons: A decision log (STATE.md), handoff snapshots for pause/resume, and a self-improving lessons layer that distills verification failures into reusable guidance. - Use Case: Ask the agent to specify and implement a new API endpoint; it writes EARS acceptance criteria, breaks work into atomic tasks with per-task gate checks, commits each task separately, and runs the Verifier before reporting completion. ## Quick Start Ask the agent to specify and implement a new feature, for example by saying "specify and implement a password reset endpoint with email verification".
