The AI LinkedIn Post Writer skill for Claude
You ask for a LinkedIn post and the skill does not start writing but asking: what really happened in your work this week? From that one moment it builds two variants in your own tone, each with a chosen hook and the reason behind it, a line-break rhythm that stays readable on a phone and one clear close. No comment-bait, no invented success stories, no thirty posts lined up in advance. Not a limitation, by the way. That is the whole point.
the lines above come from the zip on this page · SKILL.md is 9,977 bytes
First the interview, then the post
reg. L.001This is the worked example that sits literally in the SKILL.md itself, shown here in shortened form. You see the most important thing straight away: the skill does not write on command. Three interview questions come first, because the post has to be about something. The user's answer is small and everyday, and that is exactly why it works.
What the AI LinkedIn Post Writer is
The AI LinkedIn Post Writer is a free skill from the skill library we maintain. A skill is an instruction file, SKILL.md, that gives an AI assistant a fixed way of working for one task. Not software you install, not a subscription, not a connection to your LinkedIn account: one text file of 1,536 words that tells Claude how a post should come together, in what order the steps go and what it may never do under any circumstances. You download the zip at the top of this page, put it in your Claude environment, and from that moment Claude writes LinkedIn posts following this logic.
You probably recognise the problem the skill solves in one of two forms. The first: you know LinkedIn is the channel in your trade where clients and prospects are reading along, but every time you stare at an empty text box for a quarter of an hour and close it again. The second is the trap on the other side: you ask an AI for "a post about entrepreneurship" and get exactly that back, a slick text that is about nobody and that anybody could have posted. Recognisable by its structure, its rule-of-three flourishes and its fridge-magnet wisdom, and readers scroll straight past it without fail.
This skill takes a third route: it forces the conversation that a good ghostwriter would also have with you. What really happened? What did you think at the time? What should the reader do with it? Only once those three questions are answered does anything go on paper, and then in your tone, because it asks for that too. The result is not a content machine but a writing aid that takes your working week seriously as a source.
It is written for entrepreneurs, freelancers and professionals who want to be visible to their own network without losing themselves in slick sales talk. It is part of the skill library NL that we make available for free from our AI and automation service, with no account and no sales email afterwards. One thing you should know in advance: the skill writes posts, it does not promise reach. Nobody can guarantee what an algorithm does with your post, and whoever does promise that is selling something other than writing.
Why most LinkedIn posts fall flat
Anyone who watches their posts die a quiet death usually looks for the cause in the algorithm. Almost always it sits somewhere else: in the text itself. The rules in the skill read like a list of what goes wrong in practice.
The post is not about anything. "Running a business is about falling and getting back up": nobody disagrees with it, and that is exactly why it does nothing for anybody. Made to Stick by the Heath brothers explains why: abstractions do not stick, concrete and unexpected stories do. That is why the skill demands one concrete work moment as its source. Not "client contact is important", but that one client who said last week that he never reads your reports.
The first line does no work. On LinkedIn only the start of your post is visible until the reader clicks see more. If your post opens with "Last week I had an interesting conversation", the thumb has no reason at all to stop. The hook therefore gets its own design step: four options from different families, each with the reason it might work and what the risk is. Section 03 shows that.
The text is a wall. A five-line paragraph reads fine on a laptop and like concrete on a phone. Because most LinkedIn use is mobile, the skill applies a line-break rhythm: paragraphs of one to three lines with air between them. Section 02 shows the anatomy, including the fold where the reader has to decide whether to keep reading.
The ending asks three things at once. Follow me, share this, and what do you think? Three requests are two too many. Exactly one thing goes underneath: a question that can genuinely be answered, or one invitation. The AIDA principle in its simplest form: the action comes at the end, and there is one of it.
But the biggest killer is this: it is not true. Invented client cases, inflated figures, somebody else's story told in the first person. It sometimes scores well, until somebody in your network recognises it. That whole family gets refused, and that is not a side note but the foundation: on a platform where you write under your own name, your credibility is the one capital you cannot buy back.
In diff form, with lines from the skill's example: what goes out and what comes back in its place.
Four layers and the air between them
reg. L.002Every post the skill produces has the same structure: hook, context, turn, close. But at least as important is what sits between them: the line breaks. On the left the skeleton as it appears on a phone, with the fold where "see more" cuts the post off; on the right why each layer is there.
What actually is in the SKILL.md
A skill is only as good as its instructions, so we simply describe them here. The file opens with frontmatter that states when Claude should pick the skill up. Not only for the obvious requests such as write a linkedin post or turn this into a post, but also for sighs of frustration: I never know what to post, I have been staring at an empty text box for an hour, my posts do nothing, I find myself sounding so boastful. Anyone who says something like that to Claude while the skill is loaded gets it automatically. The frontmatter also states what the skill is not for: advertising copy, press releases, and replies on somebody's behalf in an ongoing conflict.
Then comes the core: a fixed six-step writing logic. Step one is the work-moment interview with the three questions from FIG.01, including the instruction to keep asking if a general topic comes back instead of a moment. Step two fixes the tone: the skill asks for two recent posts or texts from the user, or otherwise a description in plain words, and then names in one line which tone traits it is going to keep, so you can adjust course before anything gets written.
Step three creates four hooks from different families, each with a reason and a risk. Step four builds the post with the line-break rhythm from Section 02. Step five delivers a second variant that is genuinely different: a different hook, a different angle or a different end point, not the same post in other words.
Step six is the self-check pass, and it is strict: does it contain nothing that did not really happen, is there no figure the user did not supply, is the close one question or one invitation, does the post sound like the user and not like AI text, and the finest test in the whole file: would the user dare to read this out loud to the person it is about.
The output format sits in the file as a literal template: per variant the hook with its reason, the full post with line breaks as it will be posted, below that the two dropped hooks with the reason they did not make it, and a notes block with what to check before you post. Then comes the refusals block with seven lines, which come up one by one below in Section 04, the shortened example you saw in FIG.01 and FIG.02, and a source list of six titles. More on that below.
Four openings, four reasons
reg. L.003For the same work moment the skill always makes four hooks from different families, and each hook comes with why it might work and what the risk is. Two are chosen, one per variant; two are dropped, with a reason. That way you are not choosing blind, and after a few posts you start recognising for yourself which family suits your story.
The theory the skill rests on
There is no science of the LinkedIn post, and the skill does not pretend otherwise. What does exist is a century of knowledge about writing and persuasion, and the choices in this file trace back to that. The SKILL.md names its sources, and that is deliberate: then you can read them yourself and decide whether you agree.
The work moment comes from the Heath brothers. Made to Stick (2007) shows why concrete, unexpected stories stick and abstractions do not. Hence the skill's strictest rule: no post without a real moment. Anyone who wants to apply that thinking more broadly to their writing can turn to the writing skills in the library, starting with the StoryBrand Writer skill, which rests on the same principle: the reader is the hero, you are the guide.
The structure comes from AIDA. The model attributed to Elias St. Elmo Lewis (1898) is older than any platform and still works: first attention, then interest, only then the action, and there is one of it. The hook is the attention, the context and the turn are the interest, the close is the action. The full copywriting version of that model has its own skill, the AIDA Copywriter skill.
The writing rules come from Zinsser and Handley. On Writing Well (1976) stands behind cutting and simplicity; Everybody Writes (2014) behind writing for readers in a hurry, and the LinkedIn reader is in more of a hurry than anyone. Short sentences, no self-importance, no jargon you would never say out loud yourself.
And the honesty comes from Cialdini himself. Influence (1984) is often cited to sell tricks, but the book says the opposite: persuasion principles only work sustainably on the basis of what is true. Use them on something that is not right and you pay it back in trust. That is why this skill's refusals block is so long, and why the psychology behind it has its own skill, the Cialdini Persuasion skill.
Even without installing the skill you can improve your own posts with these principles: start from one real moment, give the first line real work to do, and ask for exactly one thing at the end.
What the skill refuses
reg. L.004The SKILL.md contains a list of seven things the skill never does, and on LinkedIn that list is your protection. You write there under your own name, for people who know you and people who have yet to get to know you. In conversation the rules play out like this: each rule is a request you might make, with the response the skill gives according to its own instructions.
Installing in Claude Code, Claude.ai or Codex
The zip contains one folder with the SKILL.md inside it. Installing is a matter of putting the file in the right place, and that place differs per environment. SKILL.md has been an open standard since December 2025, so the same skill also works in Codex, Cursor and Gemini CLI. So you are not downloading a Claude file but a working instruction that any modern AI assistant can read.
- Unzip it into
~/.claude/skills/(or.claude/skills/in your project). - Claude then recognises the skill by itself as soon as you start talking about a LinkedIn post.
- You can also call it directly, with
/ai-linkedin-post-schrijver.
- Go to Customize and then Skills.
- Upload the zip there as a skill.
- Or paste the contents of SKILL.md into the project instructions of a Project.
- Open
AGENTS.mdin your repo. - Paste the contents of SKILL.md into it, or put SKILL.md alongside it as a separate file and refer to it from
AGENTS.md. - Codex reads that along at every session.
After that, using it is simple: say you want a post and answer the three interview questions. Keep your example posts on hand for the tone step. New to skills, read what Claude skills actually are NL and how to install a skill NL first; want to build skills yourself after that, the method is in writing a SKILL.md yourself NL. You will find the broader explanation of working with AI in the knowledge base.
When to use it and when not
It is at its strongest when something has happened: a client conversation that stuck with you, a mistake you learned something from, an insight that changed your working week. A small moment works too, and often better than a big success, because small moments are credible and spark recognition. What does not work is warming up a moment from three months ago. It works for the sole trader who wants to post once a week and for the entrepreneur who finally wants something other than announcing partnerships.
There are also situations where you are better off leaving it be. Advertising copy and sponsored campaigns are one of them: that is a different trade with different rules, and for that you are better off looking at advertising. Press releases are a second: those have their own form and their own purpose. Then the most important limit: replies on your behalf in an ongoing conflict. An angry reply to an angry post is exactly the moment you do not want AI, but a night's sleep.
The limit that comes up most often is more subtle: this week simply nothing happened that is worth a post. Then the skill stops, and that is deliberate. The alternative, a slick post about nothing, costs you more than a week of silence. Anyone who structurally has nothing to say on LinkedIn does not have a writing problem but an attention problem: the moments are there, they simply are not being noted down. It actually helps with that too, with a list of moments instead of mass production.
And the most honest limit of all: the skill writes posts, it does not build a strategy. Which topics position you, how often you want to post, what you do with the replies: that is work that comes before the writing. The psychology of recurring attention has its own skill, the Hook Model Builder skill, and anyone who wants to use their client stories as evidence starts with the Case Study Generator skill.
Run it yourself or have it run
reg. L.005This skill is the free do-it-yourself version of work we also deliver as a service. It stays complete and with no catches, but be aware of what a skill is: it teaches your AI how to do something, while every new session starts empty. It is not the engine and not the memory. You supply your tone and your example posts again every time, you keep watch yourself over whether something got posted this week. Anyone who wants that differently has two next steps: hand over the engine, or sort out the memory.
where you are now The skill: you are the engine You run the AI LinkedIn Post Writer yourself in Claude, Codex or Cursor. Costs nothing, works today, and you keep it fully in your own hands: no trial period, no locked-off parts. The limit is your own time: it only happens when you do the interview and post the chosen variant yourself.
have it prepared The AI employee: it is ready without you prompting For most entrepreneurs LinkedIn is not a content channel but a sales channel, and that is exactly where the service version sits: the Sales Employee does prospect research and prepares commercial groundwork without you having to sit behind Claude for it. Control stays with you, because output stays a draft until a person approves it. We deliver this through Mansotti, the company TheSEO trades under, which alongside the Sales Employee also builds a Quote Employee and a Reporting Employee. Read what an AI employee is and does.
everything from one source Jarvis: all your AIs work from the same company knowledge The skill teaches the AI, the brain is where the memory lives. And with posts you notice that straight away: you do not want to supply your tone, your example posts and your brand rules again every session. That is Jarvis, the organisation brain. It connects ChatGPT, Claude, Codex and your people to the same projects, core knowledge and decisions, so your next AI session does not start from scratch. Client knowledge stays isolated and every step leaves a checkable trail. What that delivers in practice, from the plans to your first week, you can read at Jarvis itself.
What Jarvis actually delivers
reg. L.006Rung 3 deserves more than a paragraph, because a tone is exactly the kind of knowledge that evaporates between sessions. Today you explain that you write dry, without emoji, in the second person; tomorrow a stand-alone chat no longer knows that. Jarvis is the organisation brain: it remembers what your AIs need to know, divides up the work and keeps track of what happened. You notice it first with the next post.
We have been running on this system ourselves for months. Every agent session, every task and every decision is logged in it and can be read back. A new session therefore does not start blank: it first fetches the recorded decisions, the running projects and the latest changes, and carries on where the previous one stopped. So we are not describing a promise but the way we work every day.
See the four plans at jarvis/prijzen NL. Through the waitlist NL you only pass on your preferred plan, without obligation. That does not create an account, an order or a payment obligation yet. We discuss business bespoke work first.
The skills around it
reg. L.007A post never stands alone: there is structure behind it, psychology underneath it and evidence beside it. These skills from the same library each take on a different part of that chain.
The structure
structure and storyThe four layers of a post are a shortened form of older structures. These two give you the full version.
AIDA CopywriterThe full copywriting version from attention to action, for emails and landing pages too.SKILL StoryBrand WriterPuts the reader in the lead role of your whole story; the same rule as in every good post.SKILLThe psychology
why it worksWhy does a thumb stop at one post and not another? These two lay the mechanism bare.
Cialdini PersuasionThe six principles behind honest persuasion, and why tricks pay themselves back in distrust.SKILL Hook Model BuilderThe psychology of recurring attention, for anyone who wants more than one-off posts.SKILLThe evidence and the language
the finishingA client story is the strongest post there is, and nothing kills it faster than AI language. These two keep watch over that.
Case Study GeneratorTurns a finished project into the checkable client story that can carry a post.SKILL AI Text HumaniserStrips the recognisable AI patterns out of your text, so your post sounds like you.SKILLLooking further
the contextWhere this skill comes from and what else there is.
The whole skill libraryAll the free skills in one list, sorted by topic.HUB NL AI and automationThe service behind it: from separate skills to automation that works in your business.SRV AI trainingIf your team wants to learn to set up this kind of work themselves.SRV NL Knowledge baseArticles about SEO, AI and online visibility, searchable.DOCFrequently asked questions
What does the AI LinkedIn Post Writer skill cost?
Nothing. The skill is free, falls under the MIT licence and you do not have to create an account or leave an email address. You download a 4.4 KB zip containing a folder and a single file, SKILL.md, and that is the complete skill. There is no paid version and no sales email follows.
Does this skill also work in Codex, Cursor or Gemini CLI?
Yes. SKILL.md has been an open standard since December 2025, so the same file also works in Codex, Cursor, Gemini CLI and other tools that follow the standard. In Codex you unzip it into .agents/skills/ in your project, or into ~/.agents/skills/ for all your projects; Codex has supported SKILL.md directly since the open standard of December 2025. Pasting the contents of SKILL.md into your AGENTS.md still works too. The instructions are just plain readable text, so any AI assistant that accepts instruction files can handle it.
Does the skill invent stories or figures if I supply little?
No. The skill starts with three interview questions about a real work moment, and without a real moment it stops. Figures and results only go in if you supply them and confirm they are correct; if a figure is missing, the sentence is written without a figure or [FILL IN] is put there instead. According to the file itself, a post that is not true is not a good post, however well it would score.
Do the posts not just sound like AI?
That is what the tone step is for. The skill asks for two recent posts or texts of your own, or otherwise a description of your tone in plain words, and then names in one line which tone traits it is keeping, so you can adjust course before anything gets written. In the self-check pass it also checks whether the post sounds like you. Want to strip AI patterns out of existing text, there is the separate AI Text Humaniser skill.
Can the skill make a month of posts in advance?
No, and that is a deliberate refusal. Thirty variants of the same post are mass production, and that is exactly the kind of content readers scroll past. Every post starts from its own work moment. What the skill does do if you want a series: help you build a list of real moments and insights you can draw on over the coming weeks, and work through them one by one.
Why two variants and not just the best one?
Because the best one does not exist without your judgement. The two variants genuinely differ from each other, with a different hook and a different angle, and each hook comes with the reason it might work and what the risk is. Below that are also the two dropped hooks with the reason they did not make it. That way you are not just choosing a post, you also learn why one opening is stronger than the other.
Does the skill use tricks to beat the algorithm?
No. No comment bait, no artificial cliffhangers, no polarising statements you do not genuinely hold, and no promises about reach, because nobody can promise that. The skill writes for the reader, not for the algorithm: one real moment, clearly written down, with one close. That is not stubbornness on principle but the only approach that keeps your reputation intact in the long run.
Visible on LinkedIn, findable beyond it
A good post sets your network in motion, but whoever googles you after reading it lands on your website, and the story has to hold up there. The free SEO scan shows in a few seconds where your site stands. Would you rather hand the writing and publishing over entirely, that falls under our social media service. And if you want to talk further about what else AI can do for your business, from separate skills to full automation, we simply do that in a conversation.