What problem does it solve? Driving a feature from a raw prompt to a production-ready pull request requires coordinating product discovery, definition, design, specification, issue decomposition, and implementation while preserving human decisions and evidence at each step. This Skill orchestrates that entire A-Team pipeline over a target repository with a manifest as the single source of truth. ## Core Features & Use Cases - Phase orchestration with human gates: Runs discovery, definition, design, spec, issues, dev, and PR phases in order, with human-in-the-loop approval gates at definition, design, and final PR review. - Manifest-driven state management: All state transitions, approvals, retries, and milestones go through a validated CLI with atomic compare-and-swap writes and durable event IDs, making interrupted runs resumable. - Independent dev verification: The dev phase dispatches each issue to an isolated runner where an independent reviewer session approves the diff before it counts as done. - Use Case: Run /feature "Add saved searches" --repo /path/to/target to grill requirements with the human, produce a PRD and design, decompose the spec into Gherkin-acceptance issues, implement them serially in worktrees, and open a reviewed PR. ## Quick Start Ask the AI to run /feature with your feature prompt and target repository path to start the discovery grill and drive the change to a reviewed pull request.
