LinkedIn Agent Skill: the link, the install and what each skill does LinkedInAgent Skill

The link

github.com/Jakeschincariol/linkedin-agent-skill (opens in a new tab)

This is the skill from the reel. It was built by a developer named Jake Schincariol, and it's free and open source (MIT license). It's an independent project, not Anthropic's and not LinkedIn's. I'm just walking you through it here.

How to install it

The easy way is Claude Code, the Claude app that works with files on your computer.

1. 1

   Step 1: Paste the link and ask Claude to install it

   Open a new session in Claude Code, paste the link above, and add this sentence, exactly as it appears on the skill's page:

   Install this skill, then confirm /li-post works.

   Before you approve any action, read what Claude is asking to do, as you would with any code that comes from the internet.

2. 2

   Step 2: Teach it to write like you

   All the skills read one file that describes your voice. Without that file, according to the creator, everything comes out sounding like everyone else.

   The quick way: paste three posts you wrote into Claude, and type: write my voice.md from these

3. 3

   Step 3: Start with one post

   Type /li-post followed by one idea in a single sentence. The skill gives you three openings to choose from and a full post, already cleaned up by the Humanizer.

   If the post needs a number you didn't give it, the skill leaves a marker in its place and asks you for it. It doesn't make up numbers in your name.

No Claude Code?

You can also add the skill as a plugin inside Claude Code, with two commands: /plugin marketplace add Jakeschincariol/linkedin-agent-skill and then /plugin install linkedin-agent.

And in a regular Claude chat, with no Claude Code at all, you can copy the contents of one SKILL.md file from the skills folder and paste it at the start of the chat. That way it works as a writing mode. According to the creator, you lose the Humanizer's two tools this way, but the rest works.

What each of the 11 skills does

According to the skill's page. You run each one with its own command:

/li-post

One idea becomes a post: three openings out of 21 formulas, and a full draft.

/li-comment

Comments on other people's posts, matched to the type of post. Never "Great post!".

/li-reply

The replies under your own posts: sorts every comment by type and writes back in order of priority.

/li-profile

A profile score out of 100, against 12 criteria, then a rewrite of whatever lost points.

/li-plan

The plan for the week: what to post, when, and which 10 people to engage with.

/li-human

The Humanizer: strips typographic tells, hidden characters and stock phrases out of the text, then runs it through five checks.

/li-carousel

A document post (carousel): the text for every slide, the first slide, and a PDF to upload.

/li-repurpose

A video, newsletter or transcript becomes a week of posts, each of which stands on its own.

/li-dm

The connection note, the first message and two follow-ups.

/li-inbox

Sorts your messages: leads, recruiters, colleagues, requests and spam.

/li-audit

An analysis of what you've already posted, by engagement rate rather than by impressions.

What to know

The five checks are not an AI detector. They're tests that run on your own computer and look for the signals detectors look for, like uniform sentence length, em dashes and stock phrases. They don't send your text to any service, and the creator himself writes that nobody can promise text that's "undetectable".

This is third-party code. It's free and open, and you can read every file before you install it. We don't review or maintain it, and updates come from the creator on GitHub.