Use this when someone hands you reviewer feedback on your analysis or paper and you're about to start making changes. It's built around a verify-first pattern: restate the concern, check it against your actual data and pre-registration, then decide if the suggestion is correct for your setup. The part about pre-registration guard is solid: it catches when a well-meaning "improve this" comment would silently convert your confirmatory analysis into an exploratory one. Also enforces pushing back with technical reasoning when feedback would introduce bias, like conditioning on a collider. The forbidden responses section bans performative agreement phrases, which honestly needed saying.
npx -y skills add k-dense-ai/science-superpowers --skill receiving-critical-review --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
github/awesome-copilot
microsoft/win-dev-skills
github/awesome-copilot
github/awesome-copilot