What problem does it solve? Coordinating work across many GitHub issues manually requires constant supervision, context switching, and repeated status checks. This Skill runs an autonomous orchestration loop that dispatches parallel worker agents, monitors their progress, merges passing PRs, and continues until every issue in scope is complete—without user intervention. ## Core Features & Use Cases - Parallel Worker Dispatch: Spawns up to 5 background worker agents concurrently using the Task tool and monitors them with TaskOutput. - GitHub-Centric State Management: Uses the GitHub Project Board as the source of truth, with issue comments as activity logs and MCP Memory markers to survive context compaction. - PR Resolution Bootstrap: Resolves all existing open PRs (checking CI status and review artifacts) before spawning new workers, and merges green PRs immediately each loop. - Failure Recovery: Runs up to 3 research cycles per failed issue before marking it blocked, with SLEEP/WAKE handling when only CI waits remain. - Use Case: Point it at a milestone or epic (e.g., "work through all issues in milestone v1.0.0") and it will autonomously assign workers, review PRs, merge passing changes, and report status to a tracking issue until the milestone is done. ## Quick Start Ask the AI to autonomously work through all open issues in a specific milestone or epic using the autonomous-orchestration skill.