privacy-review skill + end-of-session Stop hook (premium hides the body, not the fact)

[omars-lab]

Why

Direct outcome of a near-miss this session: a premium: true career doc titled "Targeting: Director, Deployed Engineering (LangChain)" was nearly published. Premium gating would have encrypted the body, but the title, URL, description, and teaser ship public in clear, broadcasting an active job search at a named company (and the current employer in the body). We caught it by hand and kept the doc off the repo. This encodes that check so it's never a lucky catch again.

What

• privacy-review skill — the mental model + review process:
  • A table of what premium encrypts (the rendered body only) vs. what ships public in clear (title, slug/URL, description, premium_teaser, tags, listing READMEs).
  • Two passes: a clear-fields leak audit + a third-party-lens read (as employer / competitor / recruiter / search engine).
  • Four dispositions, most-protective first: keep off the repo (+ git purge) · de-identify into a case study · draft: true · publish.
  • The purge procedure for getting a sensitive doc fully out of git (reflog expire + gc, verify never reached gh-pages).
• privacy-review-reminder.sh Stop hook — fires at session end when premium/career/personal content was touched and no review ran; silent otherwise (and once the skill has run). Ships a --selftest (3 assertions) per the every-hook-proves-it-bites tenet; registered in settings.json.
• manage-premium-content — a prominent "premium is NOT a privacy guarantee" cross-link.

Verification

• Hook --selftest: 3 assertions pass.
• make validate-hook-tests: green (all 33 validator hooks ship a passing selftest).
• 3 live-fire tests: fires on sensitive-content+no-review; silent on unrelated work; silent when the skill already ran.

🤖 Generated with Claude Code

https://claude.ai/code/session_01WRtyLha5c6gN3VJ6cEqRPu