agentleFS
alexagriffithalexagriffith/agent-skills/skills/visual-style-laws/SKILL.md

visual-style-laws

Skill2 starsChanged 4 months agoMITView on GitHub

Rules for visual output — HTML explainers, walkthroughs, slide decks, data-viz. Governs headings, wording, tables, boxes, and theming so pages read clean and consistent. Use when building or reviewing any visual page, deck, or chart.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    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:

    When you say what a feature does, use the verb its own docs use ("flow control

    line 34
  2. Distinctive:

    If a section is genuinely a two-sided comparison, name it ("Separation vs SLO").

    line 39
  3. Distinctive:

    No small uppercase kicker above a heading — no "SECTION · LABEL", no numbered "① STEP NAME", no per-slide tag.

    line 45
  4. Distinctive:

    Use the term a reader already knows.

    line 69
  5. Distinctive:

    Keep light and dark. A chart MATCHES the active theme — dark chart on a dark slide, light on light — it never floats as a white plate on a …

    line 75
  6. Distinctive:

    Every slide's title sits at the same position and size, like a PowerPoint master — a pinned title zone, content in a box below.

    line 88

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

Visual style laws

Rules for anything visual (walkthrough HTML, slides, interactive explainers, data-viz). Prose voice lives in the voice guide; this governs the visual layer.

Principles, not templates. Examples illustrate a rule, they are not the only right phrasing. Apply the principle in the author's voice — don't pattern-match the example.

1. Headings name the subject, plainly

A heading names what the section IS — the concept or the move — in the fewest plain words. It reads like a nav label. It does not moralize: no "don't", no "before you trust X", no problem framing that makes the reader parse a warning before they learn the topic.

  • Good: "Establish baseline", "Find the operating point", "Verify".
  • Wrong, same sections: "Check the cluster before trusting any prior number", "Sweep to the knee, don't guess it", "One run is not a result".
  • Test: could it be a tab label, and does it say the topic, not the pitfall?

2. Page-topic lines command; feature verbs come from the source

The hero line introducing a page is a command or a question, never a limp verb. "Measuring flow control under pressure" — not "Flow control keeps…", "…lets…". When you say what a feature does, use the verb its own docs use ("flow control enables intelligent request queuing"), not a softer invented one.

3. Name a real contrast; never the rhetorical not-Y

If a section is genuinely a two-sided comparison, name it ("Separation vs SLO"). Banned everywhere — headings, labels, sentences — is the not-Y flourish: "it's not just A, it's B", "what it is and what it is not". State the thing.

4. No eyebrow kickers

No small uppercase kicker above a heading — no "SECTION · LABEL", no numbered "① STEP NAME", no per-slide tag. The heading leads. A kicker is a second type level repeating the title; if you want one for orientation, fix the title.

5. Visuals carry the evidence, not paragraphs

A visual unit is title + one takeaway line (≤ 15 words) + the visual. The chart or number cards carry the evidence. Longer explanation goes to notes or a collapsible, never the face — a paragraph on a visual unit is a blog.

  • Prose is complete sentences; structure enumerates. A comma-strung verbless fragment ("Four scenarios, gate on and off, three repeats. 72 runs.") is slide-speak. The table or diagram enumerates the dimensions; the lead says the purpose or the conclusion — and never restates what the visual directly below already shows.
  • Technical detail goes in a table, not prose. Configs, sweep points, and per-tier numbers read better as a small table. Takeaway in one line, values in the table.
  • One "what we learned" callout, once. A walkthrough earns exactly one place to name where a measurement misled us and which metrics guard against it — a single callout framed as a checklist, not a confessional, not spread around.

6. Say the plain word, not the jargon

Use the term a reader already knows. Cut a fancier word that doesn't earn its place: "throughput", not "goodput". Label metrics plainly on the face; keep the exact metric id for the table or the docs.

7. Charts and pages honor the theme

Keep light and dark. A chart MATCHES the active theme — dark chart on a dark slide, light on light — it never floats as a white plate on a dark page (that reads as unfinished). Charts read the design system's light/dark token ramps; one theme flag flows through page, slides, and charts.

8. Tables: tight, consistent, centered

A key/value table is two columns, densely padded, with a center divider, both columns aligned the same way (left), and centered on the page. It should not sprawl. Use one compact table style across a page so every table matches.

9. Fixed title level on slides

Every slide's title sits at the same position and size, like a PowerPoint master — a pinned title zone, content in a box below. Titles never float or scale with content.

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.