Review product decision reasoning with AI
Examine the evidence, alternatives and reversibility behind a product choice
Review product decision reasoning with AI by asking what supported the choice when it was made. The pm-decision-quality-audit skill examines a decision memo or retrospective without treating a fortunate outcome as proof of sound reasoning. It looks for missing alternatives, unsupported confidence, sunk-cost arguments and a mismatch between deliberation and reversibility.
Findings include a severity, the relevant principle and a concrete fix, ending with three priorities. Review effort should match the cost of being wrong: a small reversible choice does not need the ceremony of an expensive commitment. This evaluates a documented rationale and process, not someone’s personality, and cannot validate the underlying evidence automatically.
What it does
- Separate reasoning from later outcomes
- Check alternatives and rejected options
- Assess the cost of reversing a choice
- Expose assumptions and confidence claims
- Challenge arguments based on sunk costs
- Prioritise changes that matter to the decision
Install and use
- Use the instruction file and licence in a compatible text assistant through the maker’s documented prompt route. Supply the rationale and evidence known at the time; the host processes those materials.
- For example: “Review the reasoning behind this tool purchase. Use only evidence available then, and identify what could change the choice.”
- State the stakes, reversibility and whether the decision is proposed or already made. Ask for text-specific findings and practical countermeasures rather than unsupported labels.
- Review the three priorities with the decision owner. Record assumptions and revisit triggers; other audits are separate optional lenses, not prerequisites for this direct review.
FAQ
Source
Made by Uxcel Lab. Original skill: pm-decision-quality-audit, published on GitHub under the MIT licence.
View on GitHub