{"id":"24073dc9-44a8-4a4b-ba1d-1280caf54421","slug":"linkedin-post-writing-skill","kind":"skill","name":"linkedin-post-writing-skill","summary":"A complete system for writing LinkedIn posts that sound human, not AI. Built for Claude Projects. Includes anti-AI detection, voice calibration, 2026 algorithm strategy, and weekly workflow.","owner":"marian-kamenistak","repoUrl":"https://github.com/marian-kamenistak/linkedin-post-writing-skill","license":"MIT","licenseTier":"permissive","trust":"community","stars":47,"match":null,"fewStars":false,"format":"skill","createdAt":"2026-03-01T12:21:16.000Z","addedAt":"2026-10-14T05:55:18.924Z","logoUrl":"https://avatars.githubusercontent.com/u/10233850?u=20cc9460b120c8ea0a4bd15620ad0f58bd826828&v=4","linkOnly":false,"collection":null,"skillPath":null,"categories":["writing","design-media"],"tags":["ai-writing","claude-ai","engineering-leadership","linkedin","prompt-engineering","linkedin-skill"],"runtime":"binary","local":true,"cloudCapable":false,"cloudCapability":"unknown","cloudReadyOnPlatform":false,"needsApiKey":false,"reviewed":false,"syncedAt":"2026-10-17T07:25:24.804Z","sample":false,"url":"https://conduits.dev/tools/linkedin-post-writing-skill","language":"","transports":["stdio"],"install":{},"requirements":{"env":[],"upstreamMayCharge":false},"tools":null,"readme":"# LinkedIn Post Writing Skill for Claude AI\n\nA complete system for writing LinkedIn posts that sound like a real person, not AI. Built for Claude Projects but adaptable to any AI writing workflow.\n\n## What This Is\n\n4 files that turn Claude (or any LLM) into a LinkedIn ghostwriter that actually sounds human. The system includes voice calibration, anti-AI detection, content strategy, algorithm optimization, and a weekly publishing workflow.\n\nI use this system to write 3 LinkedIn posts per week for my engineering leadership coaching practice. Every post goes through an AI detection checklist before publishing.\n\n## The Files\n\n### Skill Files (upload these to your AI project)\n\n| File | What It Does |\n|---|---|\n| `files/LinkedIn_SKILL.md` | Core writing guidelines. Voice profile, banned words (40+), post types, 2026 algorithm strategy, pre-publish checklist. This is the main instruction file. |\n| `files/anti-ai-writing-guide.md` | How to NOT sound like AI. 8-part guide covering vocabulary, structure, tone, formatting, openings/closings, and content patterns. Includes side-by-side rewrites. |\n| `files/LinkedIn_POST_EXAMPLES.md` | Real post examples for voice calibration. Annotated with what makes each one work. |\n| `files/linkedin-weekly-system.md` | Weekly content workflow. Sunday planning, Tue/Wed/Thu writing sessions, Notion backlog integration, content mix rules. |\n\n### Setup Files (for configuring Claude Projects)\n\n| File | What It Does |\n|---|---|\n| `SETUP_GUIDE.md` | **Start here.** Step-by-step guide to build your own LinkedIn writing skill from scratch. Written for beginners. |\n| `LinkedIn PROJECT_INSTRUCTIONS to paste.md` | Project instructions to paste into your Claude Project's system prompt. Defines triggers, formatting rules, and behavior. |\n| `LinkedIn MEMORY_SEEDS to paste.md` | Suggested memory entries for tracking story numbers, high-performing formats, seasonal context, and voice corrections. |\n\n## How to Use It\n\n### With Claude Projects\n1. Create a new Claude Project\n2. Upload the 4 files from `files/` to the project's knowledge base\n3. Copy the contents of `LinkedIn PROJECT_INSTRUCTIONS to paste.md` into the project's custom instructions\n4. Optionally seed your memory with entries from `LinkedIn MEMORY_SEEDS to paste.md`\n5. Start writing posts\n\n### With Other LLMs\nThe files are plain markdown. You can paste them into any LLM's system prompt or context window. Start with `LinkedIn_SKILL.md` and `anti-ai-writing-guide.md` as the minimum setup.\n\n### Adapting to Your Voice\n\n> **⚠️ Important:** These files are calibrated to MY voice (direct, confrontational, coaching-oriented). If you use them as-is, your posts will sound like me, not you. You MUST customize the voice profile before publishing anything.\n\nThe `anti-ai-writing-guide.md` is universal. Keep it. The banned words, structure rules, and detection checklist work for everyone.\n\nEverything else needs to become yours. Here's how:\n\n**Quick start: Use this prompt to build your own voice profile**\n\nPaste this into Claude (or any LLM) along with 5-10 of your best LinkedIn posts:\n\n```\nI'm building an AI writing skill for LinkedIn. Below are my real posts that performed well.\n\nAnalyze them and create a voice profile for me. Include:\n\n1. Tone description (how do I sound? formal/casual, serious/humorous, etc.)\n2. Sentence patterns (short fragments? long narrative? mixed?)\n3. Words and phrases I use often\n4. Words and phrases I NEVER use (that would sound wrong coming from me)\n5. How I open posts (what patterns do I repeat?)\n6. How I close posts\n7. Formatting habits (bullets, emojis, white space, line length)\n8. My typical post structure\n9. Topics and themes I care about (these become my content pillars)\n10. What makes my voice different from generic LinkedIn advice\n\nBe specific. Don't say \"casual tone.\" Say exactly what makes it casual.\n\n[PASTE YOUR 5-10 BEST POSTS HERE]\n```\n\nTake the output and use it to replace the voice profile section in `LinkedIn_SKILL.md`.\n\n**Then customize these sections:**\n\n1. **Voice Profile** in `LinkedIn_SKILL.md` → replace with your own tone, phrases, and banned words\n2. **Post Examples** in `LinkedIn_POST_EXAMPLES.md` → replace with your actual best-performing posts\n3. **Content Pillars** → swap my 3 pillars for your expertise areas\n4. **Proof Points** → replace my numbers and credentials with yours\n5. **Project Instructions** → adjust triggers and behavior to match your workflow\n\n**After your first 5 posts, refine further.** Tell Claude:\n\n```\nHere are 5 posts you wrote for me. I'm marking what felt right and what felt off.\n\n[Paste posts with your notes like: \"this sentence sounds like me\", \"I would never say this\", \"too formal here\", \"good energy in this paragraph\"]\n\nUpdate my voice profile based on this feedback.\n```\n\nRepeat every few weeks. The skill gets sharper over time.\n\nFor the full step-by-step walkthrough, see `SETUP_GUIDE.md`.\n\n## What Makes This Different\n\nMost AI writing prompts focus on what to write. This system focuses on what NOT to write.\n\nThe anti-AI detection layer catches 40+ vocabulary tells, 10+ structural patterns, and 6 tone problems that make AI-generated content obvious to readers. Every post runs through a 6-step quality gate before delivery.\n\nThe system also includes 2026 LinkedIn algorithm signals (360 Brew), content funnel strategy, and framework branding guidelines.\n\n## The Anti-AI Philosophy\n\nYour readers have developed an internal AI detector. They can't always name what feels off, but they feel it. The text is too smooth. Too balanced. Too careful.\n\nThis system eliminates that \"AI smell\" by:\n- Banning 40+ words that AI overuses (leverage, navigate, unlock, empower, delve...)\n- Breaking AI's default structure (intro → balanced body → summary → optimistic close)\n- Forcing specific details over vague generalizations\n- Requiring a clear stance instead of \"on the other hand\" hedging\n- Checking every post against a read-aloud test\n\nFull breakdown in `anti-ai-writing-guide.md`.\n\n## About\n\nI'm [Marian Kamenistak](https://www.kamenistak.com), an Engineering Leadership Coach based in Prague. I run the [Engineering Leaders Community](https://www.engineeringleaders.io/) (1,700+ members), mentor 300+ engineering leaders per year, and organize the annual ELC Conference.\n\nThese files are the actual system I use to write my LinkedIn content. Open-sourced because I think more people should know how to make AI-assisted writing sound human.\n\n## License\n\nMIT. Use it, adapt it, share it. If you build something cool with it, I'd love to hear about it.\n\n## Connect\n\n- LinkedIn: [mariankamenistak](https://www.linkedin.com/in/mariankamenistak/)\n- Website: [kamenistak.com](https://www.kamenistak.com)\n- Community: [engineeringleaders.io](https://www.engineeringleaders.io/)\n"}