Hands the model a plan and asks it to write the post-mortem for a failure that hasn't happened yet. Surfaces the risks your team is too invested to name out loud.
You are running a pre-mortem. It is {{HORIZON}} from now and the plan below has failed — not spectacularly, but in the quiet, ordinary way real plans fail. PLAN: {{PLAN}} CONTEXT: {{CONTEXT}} Write the post-mortem for that failure. Follow this structure exactly: 1. **The headline.** One sentence a colleague would say in the hallway. 2. **What actually happened.** A short narrative, in past tense, of how it went wrong. Be specific about sequence and timing. 3. **The five root causes.** For each: name it, say which assumption in the plan it violates, and rate how visible it was at the start (obvious / knowable / genuinely unforeseeable). 4. **The tell.** For each root cause, the earliest observable signal that would have caught it — something measurable, not a feeling. 5. **What we'd change on day one.** Three concrete edits to the plan as written. Rules: - Do not hedge. Commit to a specific failure story rather than listing generic risks. - At least two root causes must be organisational or human, not technical. - Name the thing everyone would have been too polite to say in the kickoff meeting. - If the plan is missing information you'd need, state the assumption you made inline rather than asking me.
A prompt is a skill with the serial numbers filed off. Install it once and your agent reaches for it on its own.
Paste this into ChatGPT, Claude Code, Codex, Cursor, or whatever you use:
Install the pre-mortem-for-any-plan skill globally from https://promptsbuddy.com/prompts/pre-mortem-for-any-plan/skill.md.claude/skills/pre-mortem-for-any-plan/SKILL.md
---
name: pre-mortem-for-any-plan
description: "Hands the model a plan and asks it to write the post-mortem for a failure that hasn't happened yet. Surfaces the risks your team is too invested to name out loud. Use when the user asks for help with strategy tasks like planning, risk, product."
license: CC-BY-4.0
metadata:
source: https://promptsbuddy.com/prompts/pre-mortem-for-any-plan
author: "Rony"
version: "1.3"
---
# The pre-mortem: find how your plan fails before it does
Hands the model a plan and asks it to write the post-mortem for a failure that hasn't happened yet. Surfaces the risks your team is too invested to name out loud.
## Inputs to collect first
Ask the user for anything below that they have not already given you. Do not
invent values for these.
- `PLAN` — Paste the plan, PRD, or launch doc. Rough notes are fine. (e.g. Ship self-serve onboarding by Q4 so sales stops doing demos for sub-$5k accounts.)
- `HORIZON` — How far out the failure is imagined. (e.g. six months)
- `CONTEXT` — Team size, constraints, anything the model can't infer. (e.g. Four engineers, one designer, no dedicated PM. Enterprise contracts renew in Q1.)
## Instructions
You are running a pre-mortem. It is {{HORIZON}} from now and the plan below has
failed — not spectacularly, but in the quiet, ordinary way real plans fail.
PLAN:
{{PLAN}}
CONTEXT:
{{CONTEXT}}
Write the post-mortem for that failure. Follow this structure exactly:
1. **The headline.** One sentence a colleague would say in the hallway.
2. **What actually happened.** A short narrative, in past tense, of how it went
wrong. Be specific about sequence and timing.
3. **The five root causes.** For each: name it, say which assumption in the plan
it violates, and rate how visible it was at the start (obvious / knowable /
genuinely unforeseeable).
4. **The tell.** For each root cause, the earliest observable signal that would
have caught it — something measurable, not a feeling.
5. **What we'd change on day one.** Three concrete edits to the plan as written.
Rules:
- Do not hedge. Commit to a specific failure story rather than listing generic risks.
- At least two root causes must be organisational or human, not technical.
- Name the thing everyone would have been too polite to say in the kickoff meeting.
- If the plan is missing information you'd need, state the assumption you made
inline rather than asking me.
## Notes from people who use this
- Run it twice with different HORIZON values. Six-month failures are execution problems; two-year failures are strategy problems.
- The 'tell' section is the real output — those become your dashboard metrics.
- Paste the model's five root causes back into your planning doc as open questions, not as answers.
---
The pre-mortem: find how your plan fails before it does · by Rony · v1.3
From PromptsBuddy — https://promptsbuddy.com/prompts/pre-mortem-for-any-plan
Licensed CC BY 4.0.
Every prompt is also available at /prompts/pre-mortem-for-any-plan/skill.md — see all install options.
Most risk-listing prompts produce mush — "the timeline may slip", "stakeholders may not align" — because you asked an open question and got an open answer. This prompt closes the question. By fixing the outcome (it failed) and the date, the model has to construct a causal chain, and causal chains are checkable.
The instruction to include organisational causes matters more than it looks. Left alone, models write technical failure stories, because technical failures are the ones that get written up publicly. Real plans usually die of turnover, a reorganisation, or a quiet loss of executive attention.
Add a line to the prompt naming a specific stakeholder — "Write it as our CFO would tell it" — and the failure story reorganises around money instead of scope. Same plan, entirely different set of tells.
Separates what was decided from what was merely discussed, attributes owners, and flags the open questions everyone left the room assuming someone else would answer.
Lena FischerTurns a week of scattered notes into an honest account of where the time went, what moved, and the one thing worth protecting next week.
Rony