Claude LinkedIn Skill: Draft Posts in Your Voice, Then Post Them Yourself

A Claude LinkedIn skill is a small folder of instructions that teaches Claude how you write LinkedIn posts. Once it's installed, you ask for a post and Claude follows your rules for hooks, length, line breaks and tone, instead of starting from a blank prompt every time.

Three things to know before you go looking for one:

• There is no official Anthropic LinkedIn skill. The ones you'll find are community-built, published on GitHub as SKILL.md folders.
• The good ones draft. You post. A skill writes and edits the text. You read it, fix it, and publish it from your own account.
• Most of the value is your voice file. A generic "LinkedIn post generator" writes generic posts. A skill fed with how you write is the version worth keeping.

Below: how a skill works, two real LinkedIn post skills compared side by side, how to install one, how to build your own, and the LinkedIn rules that decide what a skill should never do.

What a Claude LinkedIn skill does (and doesn't)

You've probably written the same prompt a dozen times: "Write a LinkedIn post about this. Short paragraphs. Strong first line. Don't sound like a press release." A skill saves that prompt, plus examples and templates, so Claude applies it automatically.

In practice, a LinkedIn post skill helps with three jobs:

• Post writing. Turn a blog post, a shipped feature, a meeting note or a rough idea into a post with a clear hook and a readable structure.
• Profile rewriting. Rework a headline or About section against rules you set: who you help, what you've done, no buzzwords.
• Voice consistency. Keep every post sounding like the same person, even when you write on a bad day.

What it shouldn't do is act on LinkedIn for you. Scraping profiles, auto-connecting, auto-commenting and auto-liking are exactly the activities LinkedIn's rules prohibit. More on that in the last section.

How a skill works: SKILL.md in one minute

Anthropic's Agent Skills documentation describes a skill as a directory with a SKILL.md file at its root. The file starts with YAML frontmatter:

---
name: linkedin-voice
description: Drafts and edits LinkedIn posts in my voice. Use when I ask for a LinkedIn post, a hook, or a rewrite of my LinkedIn headline or About section.
---

Only two fields are required, and both have hard limits:

That reserved-word rule trips people up: a skill named claude-linkedin-skill breaks it. Call it linkedin-voice or linkedin-post instead.

The docs also explain why skills don't bloat your chats. Claude loads only each skill's name and description at startup, around 100 tokens per skill. The full SKILL.md body loads only when your request matches the description. Extra files, like a voice guide or post templates, load only when the instructions point to them.

So the description is the trigger. If it says "Use when I ask for a LinkedIn post," Claude reaches for it when you ask for a LinkedIn post. If it's vague, the skill sits unused. For a broader primer, see our guide to what Claude skills are.

Two public LinkedIn post skills, compared

You'll find plenty of "LinkedIn skills" listed around the web. These two are on GitHub with readable source and an MIT licence. Everything below comes from their own README and SKILL.md files.

linkedin-post-generator (in Varnan-Tech's OpenDirectory repo) turns a blog URL, pasted article, GitHub PR or a description of something you built into a LinkedIn post. The SKILL.md makes Claude define the audience, goal, core insight and proof first, then pick one of five formats (operational story, case study, contrarian opinion, framework post, build-in-public). It also bans a long list of filler words, em dashes, semicolons and hashtags. One rule stands out: "DO NOT INVENT SPECIFICS." Every number has to come from your source material.

linkedin-post (in Harry Munro's claude-content-skills repo) is built around your brand. You fill in a voice guide, a business context file and a content-pillars file, and the skill reads them before writing. It also runs a small Python script that picks rhetorical devices for the post, and ships three templates: technical tip, industry take and personal story.

Notice that the two disagree on hashtags. Neither is "right." That's the point of a skill: it encodes one set of opinions, and you should pick (or edit) the one that matches how you want to write.

The README and SKILL.md of linkedin-post-generator also disagree on details. The README says you get two extra hook lines, while the SKILL.md asks for three hook variants, a sharper alternate version and three first-comment ideas. Expect the SKILL.md, which is what Claude actually reads, to win.

How to install a LinkedIn skill in Claude

Where the skill goes depends on which Claude you use. Anthropic's docs say custom skills don't sync across surfaces in general. One exception: recent versions of Claude Code, signed in with a claude.ai account, download the skills you've enabled on claude.ai. That only works one way, so a skill you add to Claude Code's folders doesn't appear in the Claude apps.

Claude Code

Per the Claude Code skills docs, a personal skill lives at ~/.claude/skills/<skill-name>/SKILL.md and works in all your projects. A project skill lives at .claude/skills/<skill-name>/SKILL.md inside a repo, and your team gets it when you commit it. Claude uses a skill when your request matches it, or you can call it directly with /skill-name.

linkedin-post-generator installs with the skills.sh command from its README:

# add --global to install into ~/.claude/skills/
npx skills add Varnan-Tech/opendirectory --skill linkedin-post-generator

linkedin-post needs more care, because it reads files outside its own folder. Its SKILL.md links to ../../config/VOICE_GUIDE.md and ../../config/CONTENT_PILLARS.md, and runs python shared/rhetoric_selector.py. The README's copy commands put config/ inside the skills folder, where those ../../config/ links don't reach, and the script path only works if Claude happens to be running from the repo root. This personal-install layout makes both resolve:

git clone https://github.com/harrymunro/claude-content-skills.git
cd claude-content-skills
mkdir -p ~/.claude/skills   # without this, cp would copy the folder *as* skills/
cp -r skills/linkedin-post ~/.claude/skills/
mkdir -p ~/.claude/config && cp -r config/. ~/.claude/config/   # where ../../config/ resolves
cp -r shared ~/.claude/skills/shared
pip install numpy   # the rhetoric selector needs NumPy

Then open ~/.claude/skills/linkedin-post/SKILL.md and change python shared/rhetoric_selector.py to python "${CLAUDE_SKILL_DIR}/../shared/rhetoric_selector.py". Claude Code replaces ${CLAUDE_SKILL_DIR} with the skill's own folder, so the script runs from any project. Fill in the files in ~/.claude/config/ with your voice and content pillars.

Claude apps (web and desktop)

Anthropic's help article on using Skills in Claude gives this path as of October 2026:

1. Turn on code execution. On Free, Pro and Max, that's Code execution and file creation (called Create and edit files in the newer layout) under Settings > Capabilities. On Team and Enterprise, an admin enables it under Organization settings > Plugins & skills.

2. Check the SKILL.md frontmatter. Per the Claude Code skills docs, claude.ai uploads accept only name, description, license, compatibility, metadata and allowed-tools, and any other top-level field makes the upload fail. linkedin-post-generator, for example, has top-level author and version fields, so move them under metadata: first:

   ---
   name: linkedin-post-generator
   description: (keep the original text)
   metadata:
     author: OpenDirectory
     version: 1.0.0
   ---
   

3. Zip the skill folder (the one that contains SKILL.md).

4. Go to Customize > Skills, click +, then Create skill, choose Upload a skill, and upload the zip.

linkedin-post doesn't suit an upload as it stands: its voice files and script live outside the skill folder, so a zip of that folder leaves them behind. Use it in Claude Code with the layout above.

One caveat: Anthropic's developer docs still describe an older path (Settings > Features) and an older plan list than the help article. Menus have moved between releases, so trust the help article for the current steps and plan availability.

Before you install anyone's skill

Anthropic's docs are blunt here: use skills only from trusted sources, meaning ones you wrote yourself or got from Anthropic, and audit every file in anything else, because a malicious skill "can direct Claude to call tools or run code" in ways that don't match its stated purpose. For a writing skill, that review is quick. Read the SKILL.md, read any scripts, and be suspicious of anything that asks for your LinkedIn password or session cookies.

Build your own LinkedIn post skill

Public skills are a good starting point, but your own voice is the part nobody else can ship. A minimal personal skill is one file:

---
name: linkedin-voice
description: Drafts and edits my LinkedIn posts, headline, and About section in my voice. Use whenever I ask for a LinkedIn post, a hook, a comment reply draft, or a profile rewrite.
---

# My LinkedIn voice

## Who I write for
Ops leads at 20-200 person companies. They skim on mobile.

## Rules
- First line works alone, under 12 words, never starts with "I"
- 1-3 line paragraphs, blank line between each
- One idea per post. One question OR one call to action at the end, not both
- Never invent numbers, clients, or results. Ask me if a claim needs proof
- Words I never use: "excited to announce", "game-changer", "thought leader"

## Examples of posts that sound like me
(paste 3-5 of your best real posts here)

## Profile rewrites
Headline: who I help + what result + one proof point. No emoji.

The examples section matters most. Five real posts you're proud of teach Claude more about your rhythm than a page of adjectives. Keep the "never invent" rule too. A LinkedIn post with a made-up metric is worse than no post.

If you'd rather have Claude write the skill with you, Anthropic's skills repository includes a skill-creator skill built for creating a new skill from scratch and tuning its description so it triggers when it should. For other writing jobs, like editing and plain-language rewrites, our roundup of Claude skills for writing sorts the options by task.

Stay inside LinkedIn's rules

This is where a lot of "LinkedIn skill" listings go wrong. LinkedIn's help page on prohibited software and extensions says it doesn't permit third-party software, including crawlers, bots, browser plug-ins or extensions, that scrape, modify the appearance of, or automate activity on its website. It points to Section 8.2 of the LinkedIn User Agreement, which prohibits, among other things:

• Scraping or copying profiles and other data from the service
• Using bots or other unauthorized automated methods to add contacts, send messages, or create, comment on, like or share posts
• Driving inauthentic engagement or trying to manipulate LinkedIn's algorithms

The same help page says accounts using prohibited tools can be restricted or shut down.

Here's the practical line:

A writing skill lives entirely on your side: text in, text out. You stay the person who decides what goes live, which is also how your posts keep sounding like you.

FAQ

Is there an official Claude LinkedIn skill?

No. Anthropic publishes document skills (Word, Excel, PowerPoint, PDF) and example skills in its public repository, but no LinkedIn skill. Every LinkedIn skill you'll find is community-built, so read it before installing it.

What's the best LinkedIn post skill for Claude?

It depends on your input. If you mostly repurpose blog posts, release notes or case studies, linkedin-post-generator's source-first workflow fits. If you post often in a defined brand voice, a voice-file skill like linkedin-post, or your own, will sound more like you. Our list of the best Claude skills covers picks beyond LinkedIn.

Can a Claude skill post to LinkedIn for me?

Treat it as a drafting tool, not a posting tool. LinkedIn's User Agreement bans bots and unauthorized automated methods that create, comment on, like or share posts, so the safe setup is text in, text out: Claude drafts, you review and publish from your own account. Never use anything that automates likes, comments, connections or scraping.

Can I use a LinkedIn skill in the free Claude app?

Anthropic's help center lists Skills for Free, Pro, Max, Team and Enterprise plans, with code execution turned on. Its developer docs list fewer plans, so check the help article for your account.

Does a skill learn my voice automatically?

No. A skill knows only what's in its files. Paste real examples of your posts into it, and update them when your style changes.

Key takeaways

• A Claude LinkedIn skill is a SKILL.md folder: a name, a description that triggers it, and your writing rules.
• There's no official one. linkedin-post-generator and linkedin-post are two MIT-licensed community skills with real, readable source, and they disagree on things like hashtags.
• Install in ~/.claude/skills/ for Claude Code, or upload a zip under Customize > Skills in the Claude apps after checking the frontmatter uses only the fields uploads allow.
• The most valuable part is your own voice file with real example posts.
• Use skills to draft and edit. Scraping and automated activity on LinkedIn break its User Agreement.

Not every useful skill comes with a clean install path. Taku mirrors working AI setups, skills included, into a desktop workspace you can run and remix, and its marketplace is where you browse what's already set up. Taku is in Beta, and the Mac app is available now.