linkedin-skills: 12 LinkedIn Marketing Skills That Draft First and Publish Only on Your OK

AI-written LinkedIn posts read like AI, and LinkedIn's own filters are tuned to notice. linkedin-skills — a 3k-star MIT-licensed bundle for Claude Code and Codex — attacks that from both ends: 12 skills that draft in your voice, audit drafts against AI-detection patterns, and refuse to publish anything until you approve it.

Why This Skill Matters

Most social-media AI tools optimize for volume. This bundle optimizes for not getting filtered: every skill shows you a draft first, the built-in Humanizer targets the tells human readers and LinkedIn's slop filter react to, and the default flow ends at copy-paste — nothing is posted without your approval. The bundle is also honest about its boundary: auto-publishing exists, but only as an optional layer you connect separately.

With 12 skills covering writing, auditing, planning, and engagement, it replaces a calendar of manual LinkedIn chores rather than one tool.

Installation

One command covers Claude Code, Codex, Cursor, and any other agent that reads SKILL.md files:

npx skills add sergebulaev/linkedin-skills

Inside Claude Code you can alternatively use the plugin flow:

/plugin marketplace add sergebulaev/linkedin-skills
/plugin install linkedin-skills@linkedin-skills

On claude.ai or Claude Desktop, add sergebulaev/linkedin-skills as a marketplace from the Customize → Plugins panel. Note the README's requirement: skills and plugins on claude.ai need a paid Claude plan (Pro, Max, Team, or Enterprise) with code execution enabled.

Real Workflow: Draft, Audit, Then Plan the Week

Once installed, describe the task and the matching skill activates. A first post from scratch:

Write me a LinkedIn post about why AI agencies are replacing traditional ones. Make it viral.

The Post Writer drafts it, drawing on what the README describes as 20 proven 2026 hook formulas plus a founders-edition angle library. Before publishing, run the check that matters:

Audit this post draft for AI tells and algorithm issues: [paste your text]

Post Audit checks your draft against 2026 algorithm rules and AI-detection patterns, and the Humanizer rewrites the tells it finds — the README says it caps em dashes rather than banning them, and it does not promise to beat detectors. For ongoing presence, hand the week to the planner:

Create a 7-day LinkedIn content plan. I'm a B2B SaaS founder targeting VPs of Marketing.

The Content Planner returns daily topics, formats, hooks, posting times, and comment targets. Eight more skills round out the 12: Comment Drafter, Reply Handler, Hook Extractor, Engagement Monitor, Profile Optimizer, Employee Advocacy, Repurposer, and Interviewer.

Tips

• The Interviewer skill is the starting point if you have never posted — it builds a Story Bank from your career that every other skill reads, so drafts stop asking you for numbers mid-request.
• Reading-side skills (Comment Drafter, Reply Handler, Hook Extractor, Engagement Monitor) work without any key by asking you to paste text; add an Apify token and they fetch posts, comments, and engagers automatically.
• Auto-publishing is optional via Publora, whose free tier allows 15 posts per month — until you connect it, the approval card is the whole publishing pipeline.
• The voice rules are worth stealing even outside this bundle: specific numbers beat adjectives, 200–350 characters for comments, 900–1,300 for posts.

When Not to Use This

If you want fully hands-off posting, the default approval flow will feel slow — and that is the point of the design. The outreach use case is not covered by the core 12: a community skill adds connection-request drafts, but LinkedIn has no invite or DM API, so you paste and send them yourself. And if your platform is X, Instagram, YouTube, TikTok, Threads, or Facebook, the same team ships platform-specific bundles instead of stretching this one.

See the leaderboard for more skills.