When your analysis is done, this walks you through closing the loop: verify it reproduces from scratch, then choose whether to merge locally, write it up for sharing, keep the branch, or discard. It detects whether you're in a worktree or detached HEAD and adjusts the menu accordingly. The discipline here is requiring reproducibility before you report anything, and forcing you to separate confirmatory results from exploratory ones in whatever you write. It also handles the cleanup choreography so you don't orphan worktrees or delete branches that haven't been merged yet. Opinionated about scientific rigor but the mechanics work for any analysis you want to archive properly.
npx -y skills add k-dense-ai/science-superpowers --skill reporting-and-archiving-findings --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