Claude Skills by Chris-Carruthers

This skill should be used when the user wants to check whether they actually understand something they built with an agent — "quiz me", "cognitive coverage", "do I understand this", "test my understanding", "what did we just build" — or after a substantial feature lands and nobody has verified the human can explain it. Quizzes against the real code, grades honestly, and records a dated assessment in the knowledge vault.

This skill should be used before writing or revising any vault document — a product spec, one-pager, PR-FAQ, six-pager or Shape Up pitch, decision note or RFC, handoff, runbook, postmortem, architecture overview, design system or component note, design critique, research findings, content style guide, project or client note, session note, or a strategy memo or vision document. Also use it when the user asks how a document should read, asks for a rewrite, says a draft "reads wrong", or asks wh...

This skill should be used when a repo's CLAUDE.md / AGENTS.md has an empty "This repo" section, or when the user asks to document how a repo is built, run or tested — "fill in the repo context", "repo-context", "document this repo for agents", "why is there no build command written down". Derives the build, run and test commands from the repo itself, verifies them rather than guessing, and emits the vault's testing-canon rows as a byproduct.