What problem does it solve? Teams commit to plans, epics, and design decisions without systematically stress-testing what could go wrong, discovering risks only after failure. This Skill applies prospective hindsight to surface failure modes, hidden assumptions, and early warning signs before a decision becomes irreversible. ## Core Features & Use Cases - Five-phase protocol: Context Gathering with a fail-loud minimum bar, Frame Setting with a verbatim failure statement, Raw Premortem across five categories (Execution, External, People, Technical, Assumptions), Parallel Deep-Dives via up to 8 sub-agents, and a Synthesis phase. - Structured Risk Registry: Outputs a 9-column risk table classifying risks as Tigers, Paper Tigers, or Elephants, with urgency levels, calibration warnings, a revised plan, and a pre-launch checklist. - Opt-in and non-destructive: Triggered only by explicit invocation, never auto-applies changes to epics, user stories, or tasks, and logs metadata-only telemetry to an episodic memory file. - Use Case: Before promoting a high-impact epic from draft to review, run a premortem on it to receive a narrative of the most likely failure, the hidden assumption behind it, and a checklist of mitigations to discuss with stakeholders. ## Quick Start Ask the AI to run a premortem on a specific epic, user story, task, wiki page, or free-text plan description, optionally specifying a timeframe such as six months.
