Open-source AI skill set turns LinkedIn posting into a pipeline
linkedin-skills is an open-source, ten-skill toolkit for Claude Code and Codex that automates LinkedIn writing, publishing, and analytics safely.
A developer broke down LinkedIn posting into eight distinct tasks — planning, hooks, de-AI-ing, publishing, replying, and analytics — and rebuilt the whole loop as linkedin-skills, a ten-skill open-source bundle for Claude Code and Codex. The project grew out of analyzing 400 posts that beat their own authors' baselines across ten verticals, distilling 16 recurring hook formulas and the insight that writers should pick what a post should earn (comments, reposts, saves) before choosing a format or topic.
Every skill follows a draft-then-approve model with no autopilot publishing. The write layer runs on the Publora API, the read layer on Apify, sidestepping LinkedIn's messy API quirks like three different post URN types and a broken INSIGHTFUL reaction endpoint. A three-tier humanizer strips AI-sounding text, cross-checks drafts against five detection tools, and an audit mode scores posts against a 2026 algorithm checklist.
For engineers, the real lesson is that the hard part of LinkedIn automation isn't prompting a model — it's making an inconsistent, undocumented API reliable enough to trust. The MIT-licensed project installs in one line via npx or a Claude Code/Codex plugin marketplace command.