What problem does it solve? Coordinating multiple AI agents to autonomously implement a batch of GitHub issues requires manual setup of labels, task lists, review loops, and verification steps. This Skill provides the complete playbook for dispatching a Claude Code agent team (implementer, reviewer, verifier, researcher) that works through every open issue labeled AFK for a given PRD without human intervention. ## Core Features & Use Cases - Agent Team Orchestration: Spawns and coordinates a persistent roster of implementer, reviewer, and verifier teammates, plus on-demand researchers, with a shared task list ordered by issue dependencies. - Idempotent Pre-flight: Verifies Claude Code version, GitHub CLI auth, the experimental agent-teams flag, and creates the ten AFK workflow labels without overwriting existing ones. - Review and Verification Loops: Enforces staged-diff review (capped at 10 rounds), smoke-test verification with commit trailers, and resume support for runs paused by usage exhaustion. - Use Case: A maintainer tags five implementation issues under a PRD with the AFK label, invokes the dispatch with the PRD issue number, and the team autonomously implements, reviews, smoke-tests, and closes each issue in dependency order. ## Quick Start Run /afk-dispatch followed by your PRD issue number to spawn the agent team and process every open AFK-labeled issue.
