taste-typography
Skill1 starsChanged 3 months agoMITView on GitHub
Typography craft for agent-built interfaces: pairing rules, fluid type scales, line lengths, baseline rhythm, weight discipline, legibility minimums, and the typography reasons most AI-generated pages read as cheap. Use whenever generating or reviewing any UI with text. Stack on the design-taste core skill.
In this repository23 other files it gives its agents.
- brutalist-industrialSkill
- cinematic-motionSkill
- corporate-trustSkill
- dark-techSkill
- design-tasteSkill
- editorial-minimalSkill
- japanese-minimalSkill
- neo-brutalismSkill
- playful-colorSkill
- quiet-luxurySkill
- retro-printSkill
- taste-accessibilitySkill
- taste-brand-identitySkill
- taste-code-craftSkill
- taste-code-reviewSkill
- taste-colorSkill
- taste-design-reviewSkill
- taste-design-systemSkill
- taste-gsapSkill
- taste-landing-pageSkill
- taste-motionSkill
- taste-ux-copySkill
- taste-web-performanceSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 30
If the brand mandates a family, the pairing is built around it — the second family comes from the same species (serif + sans from the same …
- Very specific: line 48
Vertical rhythm: all text blocks align to the 4px baseline grid.
- Distinctive: line 57
Ragged edges matter: no widows (single word on last line) in headings — fix with word breaks or sizing.
- Very specific: line 58
Faux styling: no fake italics (use the family's real italic face), no fake bold (never font-weight: 700 on a 400-only family).
≈ “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.
Files
taste-typography
Type is the first thing people register and the last thing they can name. This skill makes sure the type does the design's work.
1. PAIRING (Two Families Max, With Purpose)
Pick two families with structural contrast — never two similar ones:
| Display | Body | Works for |
| Serif with character (Fraunces, Newsreader) | Neutral sans or the serif itself | Editorial, luxury, calm |
| Heavy grotesk (Archivo, Bricolage) | Same family at 400 | Bold, playful, brutalist |
| Neutral grotesk (Space Grotesk, system) | Same family | Dark tech, developer tools |
| Mono (labels/metadata only) | Grotesk body | Technical, terminal |
- Three families = soup. One neutral sans = template.
- Mono is a label voice, not a body voice. Mono for body text is for data tables and terminals only.
- If the brand mandates a family, the pairing is built around it — the second family comes from the same species (serif + sans from the same designer is a safe pair).
2. THE SCALE (Two Scales, Applied Consistently)
Display scale (marketing, heroes): 64 / 48 / 32 / 24 / 18 / 16 — apply with clamp() for fluid size: clamp(2rem, 5vw, 4rem).
UI scale (products, dashboards): 13 / 14 / 15 / 16 / 20 / 24 — never below 12px, never above 24px inside UI surfaces.
Hierarchy rules:
- One display size per section. If two headings on a page are equally large, neither is a heading.
- Weights: one weight for display (the family's natural voice: 400 for serifs, 700–900 for grotesks), 400 for body, 500–600 for emphasis. Three weights per page max.
- Small type (≤ 12px) must be accompanied by
letter-spacing≥ 0.02em or it renders muddy — but never below 10px without a data-table justification.
3. MEASURE, LEADING, RHYTHM (The Numbers)
- Line length: 45–75 characters for body. Hero subheads ≤ 40 chars per line, headlines ≤ 24 chars per line (break early, not by luck).
- Leading: headlines 0.95–1.08 (serifs need the low end), body 1.55–1.7, labels 1.2–1.4.
- Letterspacing: uppercase labels 0.08–0.2em; display lowercase −0.02em; body 0. NEVER apply tracking to lowercase body text above 16px except as a deliberate editorial act.
- Vertical rhythm: all text blocks align to the 4px baseline grid. When in doubt,
line-heightin multiples of 4: 20/24/28/32px. - Paragraphs: max ~3 lines per paragraph in marketing copy. A 6-line paragraph in a card is a layout failure, not a copy failure.
4. TEXT PLACEMENT (The Details That Read as "Designed")
- Never set body text full-width — constrain to 32–40em columns, even (especially) on widescreen.
- Break headlines deliberately: control word breaks with manual line structure where it matters ("Know exactly / what your users do").
- Headline + sub + paragraph: each step should visibly shrink in size AND weight, not just one of them.
- Justified text is banned except for print-style editorial columns ≥ 60 chars with hyphenation. Left-aligned everywhere else.
- Ragged edges matter: no widows (single word on last line) in headings — fix with word breaks or sizing.
- Faux styling: no fake italics (use the family's real italic face), no fake bold (never
font-weight: 700on a 400-only family). - Fluid type:
clamp()with rem, not px, for all text. Nevervwunits without a rem floor.
5. ACCESSIBILITY FLOOR (Non-Negotiable)
- Body ≥ 16px on marketing pages, ≥ 14px inside UI; below that, only metadata with 4.5:1 contrast.
- Contrast ≥ 4.5:1 for all body, 3:1 for large display (≥ 24px / 18.66px bold).
text-rendering: optimizeLegibilityon display; no-webkit-text-stroke(breaks at small sizes).- Font loading:
font-display: swap; preload the display face; never render invisible text waiting for fonts.
6. CHECKLIST (Fail the Page If…)
- One neutral family doing every job.
- More than two families.
- A heading below 20px with body beside it at 16px (no hierarchy).
- Body text wider than 75 characters.
- Any text under 10px.
- More than three font weights in use.
- A 6-line paragraph inside a card or sidebar.
- Justified text without hyphenation support.
If the typography works, people will say the site "feels premium" and not know why. That's the goal.
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.

