mercedes-lab-updatelisted

# mercedes-lab-update Teaching content about a fast-moving tool rots quietly. A lesson that was correct when written can, months later, tell a learner to run a command that no longer exists, install a package that has been archived, or follow a file convention that has changed. The reader hits a wall and loses trust in the whole course. This skill keeps Claude Code teaching content **honest against the source of truth** — Anthropic's official documentation at `https://code.claude.com/docs` — and proposes the fixes. It is the companion to writing the content: write it once, re-run this whenever the product moves or before you re-share. > This is different from a docs-vs-code sync. This skill validates content *about Claude > Code* against *Anthropic's external, live docs*, not a repo against its own code. ## When to use it - Before publishing or re-sharing a Claude Code course, tutorial, or internal guide. - After a Claude Code release, when commands, conventions, or models may have changed. - Periodically (e.g. quarterly) on any "evergreen" Claude Code training material. ## What to check (against the live docs — never from memory) Read the content (markdown lessons, docs, slides-as-text), then verify every Claude Code claim by fetching the relevant page under `https://code.claude.com/docs`. Common drift points, with the facts to confirm each run: - **Slash commands** are real and current — e.g. `/clear`, `/compact`, `/context`, `/memory`, `/skills`, `/agents`, `/ini