What problem does it solve? Coordinating multi-repo sprint execution by hand means copying sprint plans into each repo, wiring up session hooks, and manually opening and prompting a Claude session per repo. This Skill automates the entire Phase 2 fan-out: it writes a mission brief into each target repo and launches a self-starting dev-team session for it. ## Core Features & Use Cases - Mission brief generation: Reads each repo's sprint-plan.md and writes a brief.md with acceptance criteria, derived-artifact deliverables, and crash-recovery instructions into docs/sprints/<sprint>/. - Automated session launch: Opens one macOS Terminal tab per repo running claude with a kickoff prompt, pinned model, and a titled tab, then records the window ID for later monitoring. - Self-healing repo setup: Installs session-tracking and guard hooks, merges settings.json wiring, and pre-trusts MCP servers so launched sessions are tracked and governed. - Model verification: Reads each launched session's transcript to confirm which model it actually started on and warns when tabs inherit the wrong global default. - Use Case: After the CEO approves a sprint plan spanning three repos, run the handoff to drop briefs in all three and watch three Terminal tabs open, each dev team already executing its brief. ## Quick Start Ask the agent to hand off sprint-12 to all registered repos, optionally with a dry run first to inspect the generated briefs before any Terminal tabs open.
