linkedin-ghostwriting

Installation
SKILL.md

Applies when ghostwriting a named executive's next LinkedIn posts. Produces three ranked, typed options per exec, each a few lines, each with an attachment, each cleared by an anti-repeat gate.

This skill exists to fix the two failure modes of exec ghostwriting: posts that repeat what the exec already said, and posts that read like marketing copy instead of the person. It runs in one pass and hands the exec ready-to-post options to approve by hand. It never posts anything.

Required inputs

Gather these before drafting. A thin input set produces generic posts.

  • A voice card per exec - their real diction, the arguments they own, what they will never say, and one or two of their best past posts. Build it from their actual posting history, not a guess. Format and starter questions are in references/voice-card-and-cadence.md.
  • A post ledger - every post that has gone out, with hook, core argument, attachment, and any performance note. This is what the anti-repeat gate reads.
  • A fresh signal feed - this week's news, competitor moves, earnings, and regulatory shifts on the exec's beat, plus repost candidates. Verify each peg's date at the primary source, not the commentary's date.

The play

Installs
3
GitHub Stars
148
First Seen
Aug 26, 2026
linkedin-ghostwriting — swan-gtm/gtm-skills