What problem does it solve? It turns an approved Technical Requirements Document (TRD) into working, tested code without skipping rigor: it orchestrates backend-developer and frontend-developer agents to implement each vertical slice of a feature using TDD on a GitHub Flow branch, with an approved plan before any code is written. ## Core Features & Use Cases - Slice-by-slice implementation: Reads the TRD's task decomposition table, proposes the next pending vertical slice, and confirms scope with the user before creating a branch and PR. - Parallel full-stack orchestration: When a feature spans backend and frontend, it builds a combined plan, gets explicit approval, then invokes both developer agents in parallel with git worktree isolation. - Review-fix resumption: When code review, QA, security, or SRE reject a slice, it resumes the same implementing agent with the findings instead of restarting from scratch, while keeping independent re-review mandatory. - Use Case: After your TRD for a new checkout feature is approved, run this step to have the backend endpoint and frontend client implemented in parallel via TDD, producing a single PR with full test coverage and a timing log entry. ## Quick Start Ask the AI to run the implement step for the approved TRD at specs/checkout/trd.md and implement the next pending slice.
