What problem does it solve? Manually driving a feature from story to landed branch requires constant human supervision at every step. This Skill runs the entire implementation loop unattended — decomposing a story into tasks, running each through test-design, implementation, refactoring, and review, then closing tasks only on executed evidence and committing them to a dedicated feature branch. ## Core Features & Use Cases - Autonomous task cycles: Decomposes a user story into tasks and runs each through test-case design, language-specific implementation, refactoring, and triaged code review via subagents. - Evidence-based closure: Replaces human approval gates with an independent verifier that re-runs the test command, reproduces reviewer findings, and maps every acceptance criterion to executed evidence before committing. - Plan adaptation and learnings: Detects when mid-run learnings invalidate the remaining plan, re-decomposes remaining tasks, and persists durable learnings across runs. - Use Case: Hand it a well-scoped story on a fresh feature branch, let it run unattended, and review the resulting commits as a branch or PR afterward. ## Quick Start Ask the agent to implement your user story end-to-end on a dedicated feature branch using the implement-flow skill, then review the resulting commits afterward.
