Model or dataset
sergebulaev/linkedin-skills avatar
sergebulaev/linkedin-skills

linkedin-skills is twelve prompt bundles that refuse to publish without your approval

Claude skills for LinkedIn. 11 Claude Code and Codex skills that write human-sounding LinkedIn posts, craft comments that get noticed, analyze your feed, and build a publishing cadence, all from your terminal. Content engineering by Creative Content Crafts. MIT.

5,937 stars961 forksPythonMIT

At a glance

What is it?
sergebulaev/linkedin-skills ships a directory of skill files that coding agents read on demand, and the one invariant across all twelve is that nothing is posted without a human pressing yes. The interesting engineering is not the writing advice but the installation matrix, where seven install routes each get a different set of steps for the same directory of plain text.
Who is it for?
linkedin-skills suits someone who writes on LinkedIn from a terminal and wants a drafting assistant rather than an automation tool. It does not suit someone who wants hands-off scheduling, since the approval step is the point.
Can I use it commercially?
Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
Is it still maintained?
Yes. The repository last received commits 5 days ago.
What is it written in?
Mainly Python, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.

From daily snapshots

Star history

  • sergebulaev/linkedin-skills5,937 ★ · +4,121
Star history
Repository2026-09-112026-11-15
sergebulaev/linkedin-skills1.8k5.9k

4,121 Stars in the last 65 days, measured from the daily GitHub snapshot. Compare with other projects

Editorial analysis

Everything works in draft mode with nothing configured

The example environment file opens with a comment that answers the question most people arrive with. Every variable in it is optional, the skills work in draft-only mode with nothing set, and a single publishing key is what turns on auto-posting once you approve something. So the architecture is a drafting tool with an optional actuator, and the drafting half has no dependency on any external service. Two further variables are commented at length: a connection identifier in a named format for the publishing service, and a token for a scraping service that lets the skills fetch post bodies, comment threads and a user's recent comments without cookies. Without that token the documented fallback is to paste the post text or a comment URL by hand, which is the honest description of how the tool behaves on a machine with no credentials.

bash
npx skills add sergebulaev/linkedin-skills

Seven install routes for one directory of text files

The install section is long because each agent expects a different convention. One command-line tool takes a marketplace identifier and then a plugin identifier. The web client takes seven steps through a customize menu, a plugins tab and a repository field. The desktop client takes eight, because it adds a confirmation tab at the end and carries a note that the tab switch is what trips people, since syncing a marketplace puts a plugin in the catalogue and not in your installed list. One agent wants the repository cloned into its working directory and told about it in a system prompt that names three helper scripts. Another takes two slash commands. A sixth loads the files directly from a clone and asks you to call a skill by name. The seventh is a single package-runner command the readme says works across several of them at once.

A symlink mirror exists so one agent can find all twelve

One install route has a note attached that explains a piece of repository layout. The repository ships a mirror of symlinks, and the reason given is that the command-line agent looks for skills in a specific place without being told where, so the mirror makes discovery automatic with no plugin install at all. That is the route the readme recommends wherever the plugin command is unavailable. It also explains something about the tree: there are several directories at the root whose names correspond to different agents' package formats, so the repository is deliberately shaped to be read by machines rather than only browsed by people.

The approval step is stated as an invariant

The usage section gives seven example prompts, and each one stops before the action rather than after it. Write a post about a topic and make it viral. Comment on a post from its URL. Audit a pasted draft for tells and algorithm issues. Work out what hook formula a given post uses. Plan a week of content for a named audience. Optimise a profile URL. Humanise a pasted draft. Then one sentence states the rule: every skill shows you a draft first and waits for your approval, and nothing is posted without it. That sentence is the difference between this and an automation tool, and it is repeated in the description at the top of the readme rather than buried in a section.

The humanizer names the detectors it cannot beat

One skill, the Humanizer, is described as removing the AI tells that human readers and LinkedIn's slop filter react to, and the usage examples include a prompt to humanize a pasted AI-generated draft.

Three paid services, each optional, each commented at length

The example configuration names three external services and comments on each in prose rather than in a table. One is a publishing service whose free allowance is stated in posts per month across two networks, and whose dashboard is where you find the connection identifier. One is a scraping service whose free credit is stated in currency and in results, with a per-thousand figure to make it concrete. One generates illustrations and returns a hosted address that flows straight into the publishing service's media handling, with a reference code in its own sign-up link. Two libraries are declared for the code itself, a requests library and a dotenv loader, and that is the whole runtime requirement list.

Seven sibling bundles share one voice engine

A callout near the top lists six other repositories from the same team, one per platform, described as matching marketing skill bundles, and the sentence underneath claims two shared things: the same voice engine and the same approve-before-publish flow. That is an architectural claim rather than a marketing one, because it means the parts that decide how text sounds and the parts that decide when to send it are the same code across seven platforms, with only the posting surface differing. The repository also carries evaluations, tests, references and an agents directory at the root, so these skills are treated as something measured rather than only something shipped; the three most recent releases, v1.1.26 to v1.1.28, were all published on 2026-10-27.

Editorial conclusion

linkedin-skills suits someone who writes on LinkedIn from a terminal and wants a drafting assistant rather than an automation tool. It does not suit someone who wants hands-off scheduling, since the approval step is the point. Before installing, note that a paid plan is required for the web and desktop surfaces, that three of the seven install routes expect a specific folder convention, and that drafting works with no credentials at all while publishing needs a third-party service.

Frequently asked questions

What are LinkedIn skills?

In this repository a skill is a directory containing a skill file that an agent reads on demand. Twelve are bundled, including post drafting, comment drafting, replies, auditing a draft and removing writing tells, and the whole bundle installs into a coding agent rather than into LinkedIn itself.

how to use linkedin skills section

Nothing in this repository concerns a section of the LinkedIn website. The skills are installed into a coding agent, and once installed you ask that agent for help with LinkedIn in ordinary language and the relevant skill loads itself.

what linkedin skills to add

It names twelve and describes what each does, and it gives seven example prompts spanning writing a post, commenting on one, auditing a draft, analysing a post's hook, planning a week, optimising a profile and removing tells from a draft.

how to add linkedin skills

Seven routes are documented, one per agent family. Some take a plugin marketplace command, one is a package-runner command the readme says spans several agents, and one simply clones the repository into the working directory, where a mirror of symlinks makes the skills discoverable with no install step at all.

What are the 7 soft skills?

That is not what this repository is about. It contains agent skills, meaning bundles of instructions a coding agent reads, and there are twelve of them, all about drafting social media content rather than about a person's professional profile.

Official sources

  1. License: MIT
  2. Project website
  3. README
  4. Releases
  5. sergebulaev/linkedin-skills on GitHub
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/sergebulaev-linkedin-skills.svg)](https://hysenlabs.com/projects/sergebulaev-linkedin-skills)

Star history chart for your README

markdown
[![Star history](https://hysenlabs.com/chart/sergebulaev-linkedin-skills.svg)](https://hysenlabs.com/projects/sergebulaev-linkedin-skills)

Was this useful?