This one runs a structured post-epic review to pull out lessons and decide what worked. It's a facilitated retrospective with two parts: analyzing what just shipped and prepping for the next epic. The workflow auto-detects which epic you finished by checking sprint-status.yaml for completed stories, though you can override it. Notable feature is the "party mode" protocol where the AI agent uses character dialogue format and creates back-and-forth with you as an active participant, complete with disagreements and team dynamics. It's built around psychological safety (no blame, systems focus) and generates action items with clear ownership. You trigger it by saying "run a retrospective" or "lets retro the epic".
npx skills add https://github.com/bmad-code-org/bmad-method --skill bmad-retrospective