Critique product plans with AI
Pressure-test a product plan by exposing its riskiest assumption, strongest objection and realistic alternatives, with findings anchored to the document.
Critique product plans with AI before a costly decision becomes difficult to change. The strategic-critique skill reads a plan, PRD or decision in full, identifies the assumption that could undermine it, and tests the reasoning against the available evidence.
Findings point to specific passages rather than broad advice. Its sceptical stakeholder is simulated, never presented as real customer feedback. The output offers genuine alternatives and the evidence that could change the conclusion, while leaving revisions and the eventual decision to the author.
What it does
- Expose the plan’s riskiest assumption
- Formulate a strong, simulated objection
- Check unsupported claims and evidence labels
- Compare genuinely different alternatives
- Find a small test for a critical uncertainty
- Collect required fixes before implementation
Install and use
- Install the complete Draftt plugin in Claude Code and make the selected references and critique template accessible. Supply the artifact and the decision it supports, omitting unnecessary internal information.
- Ask: “Critique this annual-subscription plan. Find its central customer assumption and a way to test it, without rewriting the plan.”
- Check each finding against the cited section. Unsourced numbers, assumptions and synthetic examples must remain distinguishable from real evidence.
- Review alternatives and the proposed test, then separately choose revisions. When available, the native critic reviews an artifact written in the same session; otherwise the workflow provides a direct-review fallback.
FAQ
Source
Made by Tom Drugan. Original skill: strategic-critique, published on GitHub under the MIT licence.
View on GitHub