What problem does it solve? Coordinating multiple user stories across a monorepo manually is slow and error-prone: stories get picked that are still blocked, parallel changes collide in the same files, and completed work is accepted without verifying PRs, CI, or cleanup. This Skill automates the full fleet lifecycle from candidate selection through adversarial audit. ## Core Features & Use Cases - Shovel-Ready Candidate Discovery: Runs a Python script against the GitHub CLI to find open issues with no blockers, no assignees, and no in-progress/blocked labels, ranked by priority tier and delivery phase. - Parallel Worker Dispatch: Spawns worker subagents in isolated wt worktrees based on origin/staging, with model and thinking-level selection matched to story complexity, and conflict analysis to guarantee disjoint workspaces. - Chief Judge Adversarial Audit: Verifies each completed story across 7 criteria including PR-to-issue linking, passing CI checks, ephemeral preview deployments, worktree teardown, and monorepo typecheck integrity. - Use Case: A maintainer says "orchestrate the next batch of stories" and the Skill selects the top 3-5 unblocked GitHub issues, dispatches parallel workers to implement them, audits each PR, and delivers a consolidated fleet report with PR URLs and unblocked next steps. ## Quick Start Ask the agent to orchestrate the next batch of shovel-ready stories so it discovers unblocked GitHub issues, dispatches parallel workers in isolated worktrees, and audits their pull requests.
