bmad-retrospective

To run a post-epic retrospective, point the agent to your project root so it can load sprint-status.yaml and story markdown files. It identifies the completed epic, synthesizes story-level insights, and generates a structured retrospective report with assigned action items.

Facilitating an epic retrospective requires project artifacts like sprint-status.yaml, epic and story markdown files, implementation and planning artifacts, and an agent manifest. These files provide the data needed to compute metrics and produce structured retrospective outputs.

Yes, retrospective facilitation uses sprint-status.yaml to automatically detect the most recently completed epic. It validates the epic completion status and confirms the selection with you before proceeding with story-level synthesis.

Retrospective story-level synthesis aggregates developer notes, review comments, testing insights, and technical-debt items across all stories. This process surfaces recurring patterns and breakthroughs to identify root causes from completed work.

The best way to translate lessons learned into preparation steps is by cross-referencing previous retrospectives and previewing the next epic. This continuity planning generates concrete preparation steps and clear ownership for upcoming work.

An agent manifest is needed to compute retrospective metrics because it provides context on the AI agents involved in the project. Combined with planning artifacts, it enables the system to accurately synthesize story-level insights and produce structured retrospective outputs.