pre-mortem

Overview of pre-mortem skill

The pre-mortem skill helps you run a structured failure-first review of a PRD, launch plan, or product proposal before it ships. It is best for product managers, founders, strategists, and AI-assisted decision makers who need more than generic brainstorming: they need a practical way to separate real launch risks from noise and turn them into an action list.

What makes the pre-mortem skill useful is its decision structure. It does not just list concerns; it sorts them into Tigers (credible problems), Paper Tigers (overblown or low-probability worries), and Elephants (unspoken issues the team may be avoiding), then helps prioritize them by launch impact. That makes it especially valuable for pre-launch review, roadmap risk checks, and pre-mortem for Decision Support when you need to know what could block release.

Because this is a pre-mortem skill, the main job-to-be-done is clarity under uncertainty: identify the most likely failure modes early enough to change the plan, not after launch when the cost is higher.

What pre-mortem actually does

The skill reads your product context, imagines the launch has failed, and works backward to the causes. The output is meant to be operational: risks, rationale, and how urgently each issue needs attention.

Who should use it

Use it when you already have a real proposal to pressure-test: a PRD, launch brief, feature rollout, or go-to-market plan. It is a good fit if you want a pre-mortem guide that helps a model reason like an experienced product reviewer rather than a generic ideation tool.

When it is the wrong tool

If you only need lightweight brainstorming, a simple prompt may be enough. If you have no concrete plan, the skill will have too little signal to classify risks well. It is strongest when the input includes assumptions, audience, timing, and success criteria.

How to Use pre-mortem skill

Install and locate the skill

Use the repository install flow shown in the project instructions, then open pm-execution/skills/pre-mortem/SKILL.md first. In this repo, SKILL.md is the only source file, so there are no support folders to browse for extra rules, scripts, or references.

Give it a real launch artifact

The pre-mortem install is only useful if you feed it a specific plan. Strong inputs look like:

• a PRD with target user, value prop, and non-goals
• a launch plan with dates, channels, dependencies, and owners
• a feature brief with known risks, constraints, and success metrics

Weak inputs look like: “analyze this startup idea.” That is too vague for a useful pre-mortem because the model cannot tell what failure would actually mean.

Turn a rough ask into a useful prompt

Instead of asking for “risks,” ask for a failure review with context and output shape. For example:
“Run a pre-mortem on this launch plan. Assume the launch happens in 14 days and fails. Identify Tigers, Paper Tigers, and Elephants, then mark each as launch-blocking, fast-follow, or track. Focus on adoption, messaging, product readiness, and operational dependencies.”

That wording improves pre-mortem usage because it tells the model what to optimize for, what time horizon to assume, and how to classify findings.

Review the output for actionability

The best output should give you a short list of high-confidence blockers, not a long dump of risks. Look for:

• missing assumptions the team has not validated
• launch dependencies that could slip
• customer objections that would kill adoption
• issues that sound serious but do not change launch readiness

If the answer is too broad, add the missing plan details and rerun the pre-mortem.

pre-mortem skill FAQ

Is this better than a normal prompt?

Usually yes, if you care about decision quality. A normal prompt can generate risks, but the pre-mortem skill gives you a repeatable structure for ranking them and surfacing blind spots. That is especially useful for pre-mortem for Decision Support.

Do I need to be a product manager to use it?

No. The pre-mortem skill is beginner-friendly if you can provide a clear plan and explain what “failure” would mean. The quality depends more on input specificity than on job title.

Can I use it for anything besides product launches?

Yes, if the task has real stakes and a concrete plan: internal tool rollouts, pricing changes, experiments, or process changes. It is less useful for open-ended ideation or purely creative work.

What is the main limitation?

The skill can only be as sharp as the context you give it. If the PRD is thin, the output may overindex on obvious risks and miss the real blockers. The pre-mortem guide works best when the source material already contains assumptions worth stress-testing.

How to Improve pre-mortem skill

Feed it sharper inputs

The biggest quality gain comes from adding specifics before you ask for analysis. Include the launch date, target customer, success metric, distribution plan, dependencies, and any known weak points. The pre-mortem skill is much more useful when it can compare risk against a concrete path to launch.

Ask for classification, not just ideas

Do not stop at “what could go wrong.” Ask the model to separate Tigers from Paper Tigers and call out Elephants explicitly. That structure reduces vague output and makes the result more useful for planning, staffing, and escalation.

Include constraints and tradeoffs

If you have budget caps, engineering limits, legal review, or a hard launch date, say so upfront. Constraints change which risks are real. A pre-mortem that ignores them may sound smart but fail to improve the plan.

Iterate after the first pass

Use the first output to tighten the next prompt. If the model missed a likely failure mode, point to the gap and ask for a second pre-mortem focused on that area, such as adoption, implementation, or launch operations. The best pre-mortem usage is iterative: first broad, then narrower, then action-oriented.