agentleFS
owainlewisowainlewis/vertica/.claude/skills/carousel/SKILL.md

carousel

Skill18 starsChanged 32 days agoMITView on GitHub

Write a Vertica carousel from a transcript, article or idea. Produces JSON using Cover, Body 1, Body 2 and CTA with a clear promise, useful sequence and readable copy.

9 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

10 rules. 0 are common, 6 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use supplied or verified results only.

    line 24
  2. Distinctive:

    Use Body 1 for most teaching.

    line 37
  3. Distinctive:

    Prefer a 4-10 word cover and about 15-35 words on teaching slides.

    line 50
  4. Distinctive:

    Keep the display headlines on covers.

    line 52
  5. Very specific:

    Use *italic* or bold sparingly.

    line 61
  6. Very specific:

    Use a diagram when it explains a mechanism better than text.

    line 68
  7. Very specific:

    Use viewBox="0 0 800 500", with no root width or height.

    line 69
  8. Very specific:

    Use parseCarouselConfig in app/carousel.ts to validate JSON.

    line 86
  9. Distinctive:

    Create from text → JSON config.

    line 87
  10. Distinctive:

    Read the sequence as plain text, then inspect the actual slides.

    line 89

“Common” is measured: kept by at least 6 in 10 of the 6 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

Write a useful carousel

Choose the reader, the problem and the decision the post helps them make. Write the argument before choosing images. The reader should get useful progress with each swipe. A consistent text layout is enough when the idea is clear.

Framework

The cover makes a specific, credible promise. Slide two starts paying it off with an example, consequence or useful distinction. Later slides explain the method, show evidence or address an objection. The final slide gives one relevant action.

For a change story, establish the problem and its cost, introduce another approach, show the method and evidence, explain what improves, and lead to a decision. These are possible beats, not a mandatory seven-slide formula. Combine or cut beats that do not help. A comparison should apply both choices to the same task and end with a usable decision rule. A playbook should leave a method the reader can repeat.

Use supplied or verified results only. Label hypothetical examples. A reference creator's reach claims do not establish what caused that reach. Offer a download or course only when the user has it and wants to promote it.

Four layouts

Editor nameJSON layoutJob
CovercoverA clear headline and one short subtitle in body
Body 1contentA bold lead in title followed by short paragraphs in body
Body 2noteOne statement or a visual example; title only
CTAclosingOne next action in title and a supporting line in body

Use Body 1 for most teaching. Repeated text roles share a fixed type scale; large cover headlines, weight and space supply the hierarchy. Body 2 gives a concise statement room, or holds a diagram or pictures. Do not add layout changes to meet a quota.

Body 2 may set visual: "photos" with an images list, or visual: "diagram" with a diagram SVG. Omit visual for text only. Body 2 does not draw body. Captions sit below the visual by default, or above with position: "top". Multiple pictures use a contained grid; images are shown in full. Photos and videos may also sit behind any layout using the background controls. Cinematic imagery is optional and should contribute to the idea or requested tone.

Copy and type

  • Prefer a 4-10 word cover and about 15-35 words on teaching slides.
  • Give each slide one useful point. Use concrete actions and consequences.
  • Keep the display headlines on covers. CTA headlines and standalone statements use fixed smaller sizes. Body 1 leads, paragraphs and visual captions use 18px at a 390px feed width. Consistency means repeated roles match, not that every text element has the same size. Never shrink type to fit.
  • Editorial covers and CTAs centre by default; teaching copy aligns left. Keep the theme's composition and quiet grounds. Avoid decorative colour changes. Reuse the text layout when the explanation does not need a visual.
  • Separate paragraphs with a blank line. Let headlines wrap naturally. Use | only when a deliberate break improves the meaning.
  • Use *italic* or **bold** sparingly. No hype or invented first-person claims.
  • title allows 120 characters and body 280. These are storage limits, not a guarantee the copy fits. Shorten any slide that triggers the overflow warning.
  • Preserve the user's voice and use no em dashes.

Diagrams

Use a diagram when it explains a mechanism better than text. At most six boxes. Use viewBox="0 0 800 500", with no root width or height. Use currentColor for ink, fill="none" on shapes, 2-unit strokes, and 8-unit corner radii. Use font-family="inherit" for theme labels. Use one 48-unit size for labels and notes; inspect actual phone-size rendering and reduce complexity if labels crowd. Arrows use lines and polygon heads. No scripts or external references.

A cover may carry a diagram as its artwork: set visual: "diagram" on the cover. Set diagramStyle: "drawn" to redraw any diagram as pen and pencil work with architect's hand lettering. Write the SVG the same way. Shade one or two key shapes with class="hatch" or class="crosshatch"; a tinted fill is hatched too, while fill="currentColor" stays solid ink for arrowheads. Labels stay clear of hatching.

Output and verification

Return the requested format. For app-ready JSON, include version: 1, title, author: "aiengineer.co", a short mark, optional theme: "ai-engineer", and a slides array using the four layout values above. Each slide needs title. Use parseCarouselConfig in app/carousel.ts to validate JSON. Load it through Create from text → JSON config. Writing a deck does not authorise social posting.

Read the sequence as plain text, then inspect the actual slides. Check the cover promise, slide-two payoff, relevant examples, fair claims, phone-size text, line breaks, visual labels and one useful final action. Avoid decorative changes that hide a weak argument.

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.