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
- Vishwajeetkumbhar379/claude-skillstheir own version, first seen Nov 1, 2026
- Samin12/skillstheir own version, first seen Aug 14, 2026
- Typeless-Git/skillstheir own version, first seen Mar 30, 2026
- karagos/ai-skills-by-caiotheir own version, first seen Mar 23, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- Samin12/skills★ 8same name
- karagos/ai-skills-by-caio★ 2same name
- Typeless-Git/skills★ 1same name
- Vishwajeetkumbhar379/claude-skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository9 other files it gives its agents.
- brand-voiceSkill
- document-extractionSkill
- image-genSkill
- nextjs-conventionsSkill
- og-generatorSkill
- pharosSkill
- seo-auditSkill
- skill-distributionSkill
- supabase-migrationsSkill
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
- Distinctive: line 44
No paragraph anywhere: If a slide has a paragraph, it wanted to be an article.
- Distinctive: line 77
Check at thumbnail size before you post: Every time.
“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.
Files
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
| Slide | Job |
| 1 | Hook. One claim, large. It has to work as a static thumbnail, because that is all most people see |
| 2 | Stakes. Why this costs the reader something today |
| 3-9 | One idea per slide. A heading and at most 25 words. If it needs 40, it is two slides |
| 10 | Recap. The three things worth remembering |
| 11 | CTA. 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
- 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.
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" } ] }Render:
python3 scripts/render.py deck.json -o carousel.pdfThe renderer produces 1080x1350 pages with embedded fonts and reports the page count and file size so you can check both before uploading.
- 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:
- A cost. "Deze fout kost je een kwartaal aan verkeerde cijfers."
- A contradiction. "Snellere agents maken je trager."
- A concrete number. "Zes van de tien facturen die ik zag hadden een verkeerde datum."
- 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.
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
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.

