| name | grammar-token-designer |
| description | Translate a slide DESIGN.md or high-end web reference into renderer-ready visual grammar tokens for HTML-first academic or technical decks. |
Grammar Token Designer Skill
Use before editing renderer CSS or creating a new visual grammar.
Workflow
- Read the target
templates/<grammar>/DESIGN.md or a small set of design references.
- Extract renderer tokens, not vibes:
surface: base, dark, proof, and annotation backgrounds.
ink: headline, body, muted, and metadata colors.
accent: one primary and optional secondary accent.
type: headline, body, mono, and metric families.
composition: cover, proof, map/ledger, matrix, close.
cadence: dark-page ratio, proof-page ratio, and scale contrast.
evidence: image area, crop priority, caption style, callout style.
- Convert tokens into a CSS checklist for
render_html.py.
- Name anti-patterns the grammar must avoid.
Rules
- A grammar is not a palette. It must change layout rhythm and proof treatment.
- Tokens must be concrete enough to implement without another design discussion.
- Use neutral bases and deliberate contrast spikes instead of one-note color systems.
- Do not introduce animation-only ideas unless the static screenshot/PPTX export still works.
- If a reference has strong brand styling, borrow structure and cadence only.
Output
Return:
Tokens: 8-14 concrete values or constraints.
CSS hooks: selectors or variables to touch.
Slide cadence: how cover, evidence, matrix, and close should differ.
Do not copy: brand or decorative details to ignore.