pre-mortem

name	pre-mortem
description	Runs a pre-mortem: imagine the project/plan has failed 6–12 months from now, then work backward to find what caused the failure. Use when the user says "let's pre-mortem this", "imagine this failed", "what's the worst case 6 months out", "before we commit, let's stress test", or wants to surface failure modes before launch. Calls the Decision Duck `pre_mortem` MCP tool.
metadata	{"author":"Troy Taylor","version":"1.1","pattern":"tool-call"}

Pre-Mortem

What This Skill Does

Flips a forward-looking plan into a backward-looking failure analysis: "It's 6 months from now and this plan failed. Why?" The framing surfaces risks that a normal risk register tends to miss — political, narrative, team-dynamic, slow-burn risks — because the question shifts from "what might go wrong" to "what already did go wrong".

Implemented by calling the dedicated pre_mortem tool, which has the pre-mortem framing and system prompt built in.

When to Activate

• "Let's pre-mortem this" / "run a pre-mortem"
• "Imagine this failed in 6 months — why?"
• "What's the worst case 6 months / 12 months out?"
• "Before we commit, stress test the plan"
• User is about to launch something and wants a final failure check

When NOT to Activate

• User wants a forward-looking risk register → use risk-analysis
• User wants to imagine adversaries attacking the plan → use red-team

Workflow

1. Capture plan and horizon. Confirm:

   • Plan: the thing being committed to, one paragraph
   • Horizon: when "failure" is measured (default: 6 months)
   • What "failure" means: explicit success criteria they'd fail to hit

2. Call the tool.

   pre_mortem(
     plan: "<the plan being committed to>",
     horizon: "<time horizon, e.g. '6 months' or '1 year'>",
     failure_criteria: "<what failure looks like>"
   )
   

3. Present the output. The tool returns failure modes, root causes, missed warning signs, and preventive actions in bullet-point format. Relabel for the user:

   • Lead items as "Failure modes" not "Risks"
   • Highlight the single most likely root cause
   • End with the one highest-leverage change to make this week