What problem does it solve? AI coding agents often jump straight into code without clear requirements, producing scope creep, untraceable changes, and lost context between sessions. This Skill structures project and feature work into an adaptive pipeline that applies full planning rigor to complex features while skipping ceremony for small fixes. ## Core Features & Use Cases - Adaptive 4-Phase Pipeline: Specify, Design, Tasks, and Execute phases that auto-size based on scope, from quick fixes (≤3 files) to complex multi-component features. - Requirement Traceability: Captures testable acceptance criteria with unique IDs (e.g., AUTH-01) and tracks them from spec through design, tasks, and atomic Conventional Commits. - Persistent Memory & Handoff: Maintains STATE.md for decisions, blockers, and deferred ideas, plus pause/resume handoffs so work continues seamlessly across sessions. - Use Case: Ask the agent to map an existing codebase into 7 brownfield documents (stack, architecture, conventions, concerns), then specify a new authentication feature, break it into atomic tasks, and implement each with verification and one commit per task. ## Quick Start Tell the agent "Initialize project" to define your vision and roadmap, or say "Specify feature user-authentication" to begin planning a feature with traceable requirements.
