agentleFS
ASNNetworksASNNetworks/floh-skills/skills/linkedin-carousel/SKILL.md

linkedin-carousel

Skill0 starsChanged 4 months agoMITView on GitHub

Turn an article, a case or a set of notes into a LinkedIn carousel PDF — slide structure, per-slide copy, and a renderer that produces the exact aspect ratio LinkedIn accepts. Use when asked to make a carousel, a slide deck for social, or to repurpose a long piece into slides.

4 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No paragraph anywhere: If a slide has a paragraph, it wanted to be an article.

    line 44
  2. Distinctive:

    Check at thumbnail size before you post: Every time.

    line 77

“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

LinkedIn carousel

A carousel is a document upload, not an image post. That single fact decides the format, and getting it wrong is why most carousels look cropped.

When to use this

Repurposing an article, case study, lesson or set of notes into a LinkedIn carousel. Also for the hook-and-structure work before any slides exist.

The format, which is not negotiable

  • PDF, uploaded as a document. Not a set of images.
  • 1080 x 1350 (4:5 portrait). LinkedIn also accepts 1:1, but 4:5 takes the most vertical space in the feed, so it is the default.
  • 8 to 12 slides. Under 8 does not justify the swipe; over 12 and the drop-off is steeper than the reach gain.
  • Fonts embedded. A PDF with linked fonts renders as fallback boxes on LinkedIn's server-side thumbnailer, and you will not see it in your own preview.
  • Under 100 MB, which is never the binding constraint if you are not embedding photos.

The structure that works

SlideJob
1Hook. One claim, large. It has to work as a static thumbnail, because that is all most people see
2Stakes. Why this costs the reader something today
3-9One idea per slide. A heading and at most 25 words. If it needs 40, it is two slides
10Recap. The three things worth remembering
11CTA. One action, spelled out

Rules that hold across all of them:

  • One idea per slide. The most common failure is slide 4 carrying three points because the source paragraph had three sentences.
  • Slide 1 has no context. It cannot say "as I mentioned". It is a cold open.
  • Readable at thumbnail size. Headline at least 48pt at 1080 wide. Hold the slide at 20% zoom; if you cannot read the headline, nobody in the feed can.
  • No paragraph anywhere. If a slide has a paragraph, it wanted to be an article.

How to run it

  1. Write the outline first, as text. Slide number, headline, body. Get the structure approved before rendering anything. Regenerating a deck because slide 3 was wrong is ten seconds; rewriting after design is not.
  2. Put it in a JSON deck file:

    {
      "theme": { "accent": "#84CC16", "background": "#0B1020", "text": "#F8FAFC" },
      "slides": [
        { "kind": "hook",  "headline": "Je AI-agent leest je documenten verkeerd",
                           "sub": "En hij zegt het er niet bij" },
        { "kind": "point", "headline": "03/04/2026",
                           "body": "3 april of 4 maart. Zonder context is dat een gok, en een gok in je boekhouding." },
        { "kind": "cta",   "headline": "Wil je dit goed hebben?",
                           "body": "Stuur me een bericht.", "handle": "@alisina" }
      ]
    }
  3. Render:

    python3 scripts/render.py deck.json -o carousel.pdf

    The renderer produces 1080x1350 pages with embedded fonts and reports the page count and file size so you can check both before uploading.

  4. Check at thumbnail size before you post. Every time.

Writing the hook

The hook slide decides everything downstream. Four openings that work, in rough order of reliability:

  1. A cost. "Deze fout kost je een kwartaal aan verkeerde cijfers."
  2. A contradiction. "Snellere agents maken je trager."
  3. A concrete number. "Zes van de tien facturen die ik zag hadden een verkeerde datum."
  4. A named mistake. "De fout die iedereen maakt met SKILL.md."

Openings that do not work: a question the reader can answer with "no", a definition, and anything beginning "In het huidige digitale landschap".

The accompanying post

The carousel does not stand alone. The post text above it carries the hook again in plain words and ends with one question. Three to five lines. The first two lines are what shows before "see more", so the hook has to land there.

Files

  • scripts/render.py — deck JSON to a 1080x1350 PDF with embedded fonts. No design tool needed.
  • references/deck-schema.md — every slide kind and its fields, with layout notes.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.