This logs your performance investigation work to structured markdown files, keeping exact user quotes, phase summaries, decisions, and evidence pointers in one place. It's built for the kind of debugging where you need an audit trail: what the user said verbatim, what metrics or commands you ran, and why you chose one fix over another. The format is rigid by design, with timestamped phases and specific sections for quotes, evidence, and rationale. It writes to a state directory and follows a canonical contract defined in perf-requirements.md. Good for when you're deep in a gnarly performance issue and need to document your path through it without losing context or misquoting the person reporting the problem.
npx -y skills add composiohq/awesome-claude-plugins --skill perf-investigation-logger --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