ghostwrite-client-postlisted

# /ghostwrite-client-post — Ghostwritten LinkedIn Post (Client Account) > **Purpose.** Write a single LinkedIn post on behalf of a CLIENT. The post is ghostwritten under the client's name, lands in the client's audience's feed, and represents the client's voice — not the agency's. Output is publish-ready, has passed voice-drift cosine-similarity check against the client's last 20 validated posts, has passed the Blind Output Test calibrated to the client's persona, and lands in the operator/account-manager review queue per `INVARIANTS.md` A-6. > > **Bar.** A reader inside the client's audience cannot tell the post was ghostwritten. The client themselves reads the post and ships it with <10% rewrite. If either fails, the encoding is wrong (voice profile thin, hook library thin, content sourcing protocol leaky) — fix the encoding, never lower the bar. > > **The single most common churn driver in LinkedIn ghostwriting is "this doesn't sound like me."** This skill is the line of defense. ## Scope — recursive context, locked This skill is **client-side only**. Every profile load is recursively scoped to a single `client_slug`: - `brand_voice.client_voices[{client_slug}]` — never `agency_voice`, never another client - `ideal_customer_profile.client_icps[{client_slug}]` — never `agency_icp` - `content_strategy.client_content[{client_slug}]` — never `agency_content` - `case_studies` — only entries flagged for this `client_slug` AND `use_in_content: true` **Cross-contamination is