Scans your installed skills to find principles that show up in multiple places, then suggests promoting them to rule files. The approach is solid: bash scripts collect the raw inventory, then an LLM cross-reads everything and produces verdicts (append, revise, new section, new file). Nothing happens automatically. You get a table of candidates with evidence from specific skills, you approve or skip each one, and it saves a timestamped results file. The batching by theme and cross-batch merge to catch patterns that span categories is thoughtful. Useful for periodic maintenance when your rules feel incomplete relative to what your skills actually assume, or after adding a bunch of new skills that share unstated conventions.
npx -y skills add affaan-m/everything-claude-code --skill rules-distill --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills