Design Theme Guide

Design Theme Guide

What it is: the documented value set a business designs against - colour, type, space, radius, motion - with every colour pair contrast-tested and written down.

Overview

Works out the smallest useful theme for the business in front of it, then builds it only when asked. The default output is a short recommendation, not a token file. The token register - CSV, SQL DDL, JSON Schema, Notion mapping - is produced on request, from one field list so the four cannot drift apart.

Layer: Layer 2: Brand & Design. Fits: Starter stage. Table code: n/a.

The rule this table exists to enforce: a theme is a set of values, not a set of screens. If a colour, size or spacing value is not a row in the register, it is not part of the system, and anything using it is a one-off. Value alone is the single source; the light and dark variants and the type metrics exist on the same row because they are the same token, and a token split across rows drifts within a month.

When to Use This Skill

• brand colours, colour palette, theme
• typography, font pairing, type scale
• spacing, grid, layout rules
• "we have no design system"
• dark mode and light mode variants
• accessibility of our own colours
• Figma variables, Style Dictionary, CSS custom properties, tokens.json

Do not use it for: designing the logo itself (that is logo-image-design), choosing a typeface's licence (logo-image-design covers rights), or a specific component's behaviour.

How It Works

Follow the shared execution contract. The module-specific rules below define only domain fields, decisions, calculations, and safety constraints.

Step 1 - Identify intent

Read the request and pick the intent before asking anything.

• "set up" / "build" / "create" -> the user wants the token set; go to Step 2.
• "our colours are a mess" / "fix" -> something exists; capture what is in use, then Step 2.
• "is this accessible" / "review" / "audit" -> a check, not a build; answer from what they share.
• "how do we ..." -> advice question; answer directly and offer the build only if it helps.

Ask only if this is the highest-value missing fact; otherwise proceed without an opener:

Q: What is the business called, and what does one line of it actually do?

Step 2 - Ask only what is missing

Treat ambiguous replies as unanswered and ask which explicit option the user means. Record unknown values as Unknown; Unknown is not zero. A record must not be Done when a required check fails.

Skip anything the user already answered, in any earlier message. Ask the rest one at a time, and stop as soon as the remaining answers would not change the token set.

• Identity - Business name and what it does? / One colour already decided, and is it fixed by a customer, a client or a sign? / Any existing logo, colours or documents?
• Surface - Website only, or print and social as well? / Dark mode needed? / Any CMS or design tool already in use (Figma, Shopify, WordPress, custom)?
• Type - Any typeface already chosen, licensed or not? / Long documents or short marketing copy? / Any script other than Latin?
• Access - Is there a public-sector, EU-market or accessibility obligation? / Has anyone reported low contrast or readability before?
• Governance - Who decides a colour change? / Does the theme need to be handed to an external developer or agency?

Never invent an answer. Hex codes, brand colours, font names and typeface licences the user has not supplied are Unknown.

Step 3 - Hold the internal context

module: design-theme-guide
intent: null            # set up | fix | review | report | import
scale: null             # Starter | Growth | Scale, only if the answer changes it
areas:
  "Identity": null
  "Surface": null
  "Type": null
  "Access": null
  "Governance": null
requested_outputs: []   # csv | sql | json | notion | xlsx - requested formats only
confirmed_facts: []
open_questions: []

Step 4 - Recommend the smallest workflow

Build an already requested artifact without asking again. For advice-only requests, give a short recommendation and offer the relevant artifact.

Recommended approach: One token register with a brand colour as the single source for accents, a neutral ramp for everything else, one display face and one body face at a fixed type scale, a 4 or 8 point spacing unit, and light and dark variants. Every colour pair that carries text gets a measured ratio and a WCAG level in the row itself, so the accessibility of the theme is auditable rather than remembered.

Why this one: A theme that exists as values is the only thing that makes the website, the deck, the letterhead and the social posts look like one business. Screens cannot do that, and a theme with no contrast measurement is a theme that fails the first time someone chooses a pastel.

Workflow: Seed colour chosen → Neutral ramp derived → Semantic colour roles assigned → Every text pair contrast-measured → Typeface pair set and type scale fixed → Spacing unit and radius set → Light and dark variants recorded → Register approved and versioned → Components built from tokens only

Step 5 - Build only on request

Once the user asks for it, derive the fields from the confirmed context and emit the requested artifacts. For machine-readable text, keep prose outside the data; for files, provide a usable link. Report material validation failures or limitations separately.

A selected Notion output is rendered by notion-manual-import, so route the Notion step there. When the user selects Notion, hand that step to @notion-manual-import: it holds the CSV, the property mapping, the import steps and the verification checklist, and it renders the Field Reference below instead of defining a table of its own. Do not restate the mapping here and do not improvise the import steps. Manual CSV and mapping outputs need no connection. For requested workspace changes, follow the shared contract: verify actual tool access and the target before writing. A user saying "connected" is not tool evidence. Never ask for a Notion password or token.

Token ID,Token Name,Token Group,Token Type,Value,Unit,Light Mode Value,Dark Mode Value,Contrast Ratio,Contrast Against,WCAG Level,Font Family,Font Size,Font Weight,Line Height,Letter Spacing,Usage Rule,Do Not Use For,Source Reference,Version,Status,Approved By,Last Reviewed,Notes
TOK-EXAMPLE-001,color.brand.primary,Brand,Color,#1F4FD8,hex,#1F4FD8,#8FB4FF,7.02,color.surface.page,AA,,,,,,Primary action and brand accent,Body text on white surface,WCAG 2.2 SC 1.4.3,1.0,Draft,Unknown,2026-09-27,Example row - replace every value before use.

-- Engine assumption: PostgreSQL. For another engine use `id BIGINT PRIMARY KEY`
-- or the engine's auto-increment equivalent and keep the rest portable.
CREATE TABLE design_token (
  token_id BIGINT PRIMARY KEY,
  token_name VARCHAR(255) NOT NULL,
  token_group VARCHAR(100) NOT NULL,
  token_type VARCHAR(100) NOT NULL,
  value VARCHAR(255),
  unit VARCHAR(50),
  light_mode_value VARCHAR(255),
  dark_mode_value VARCHAR(255),
  contrast_ratio NUMERIC(5,2),
  contrast_against VARCHAR(255),
  wcag_level VARCHAR(50) NOT NULL,
  font_family VARCHAR(255),
  font_size VARCHAR(50),
  font_weight VARCHAR(50),
  line_height VARCHAR(50),
  letter_spacing VARCHAR(50),
  usage_rule TEXT,
  do_not_use_for TEXT,
  source_reference VARCHAR(255),
  version VARCHAR(50) NOT NULL,
  status VARCHAR(50) NOT NULL,
  approved_by VARCHAR(255),
  last_reviewed DATE,
  notes TEXT,
  created_at TIMESTAMP DEFAULT NOW(),
  updated_at TIMESTAMP DEFAULT NOW(),
  CONSTRAINT token_ratio_range CHECK (contrast_ratio IS NULL OR (contrast_ratio >= 1 AND contrast_ratio <= 21)),
  CONSTRAINT token_wcag_level CHECK (wcag_level IN ('AA','AAA','Fail - do not ship','Not applicable'))
);

CREATE INDEX idx_design_token_status ON design_token (status);
CREATE INDEX idx_design_token_type ON design_token (token_type);

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "title": "Design Theme Guide",
  "type": "object",
  "additionalProperties": false,
  "properties": {
      "Token ID": { "type": "integer" },
      "Token Name": { "type": "string" },
      "Token Group": { "type": "string" },
      "Token Type": { "type": "string" },
      "Value": { "type": "string" },
      "Unit": { "type": "string" },
      "Light Mode Value": { "type": "string" },
      "Dark Mode Value": { "type": "string" },
      "Contrast Ratio": { "type": "number" },
      "Contrast Against": { "type": "string" },
      "WCAG Level": { "type": "string" },
      "Font Family": { "type": "string" },
      "Font Size": { "type": "string" },
      "Font Weight": { "type": "string" },
      "Line Height": { "type": "string" },
      "Letter Spacing": { "type": "string" },
      "Usage Rule": { "type": "string" },
      "Do Not Use For": { "type": "string" },
      "Source Reference": { "type": "string" },
      "Version": { "type": "string" },
      "Status": { "type": "string" },
      "Approved By": { "type": "string" },
      "Last Reviewed": { "type": "string", "format": "date" },
      "Notes": { "type": "string" }
  },
  "required": [
      "Token Name",
      "Token Group",
      "Token Type",
      "WCAG Level",
      "Version",
      "Status"
  ]
}

| CSV column | Notion property | Set after import |
|---|---|---|
| Token ID | Text (preserve source ID) | Keep imported IDs as Text; optionally add a separate Unique ID property |
| Token Name | Text | Leave as Text. This is the join key - keep it identical everywhere |
| Token Group | Select (add options after import) | Convert to Select, add options: "Brand", "Semantic", "Neutral", "Typography", "Layout", "Component" |
| Token Type | Select (add options after import) | Convert to Select, add options: "Color", "Typography", "Spacing", "Radius", "Shadow", "Border", "Motion", "Breakpoint", "Z-Index" |
| Value | Text | Leave as Text. Hex, px, rem, ms, unitless - it goes in the next column too |
| Unit | Title | Use as the database title |
| Light Mode Value | Text | Leave as Text |
| Dark Mode Value | Text | Leave as Text |
| Contrast Ratio | Number | Convert to Number, 2 decimal places, no unit |
| Contrast Against | Text | Leave as Text. Name the token, not a colour name |
| WCAG Level | Select (add options after import) | Convert to Select, add options: "AA", "AAA", "Fail - do not ship", "Not applicable" |
| Font Family | Text | Leave as Text |
| Font Size | Text | Leave as Text. Keep the unit inside the value, e.g. "16px" |
| Font Weight | Text | Leave as Text. Name the weight, e.g. "Semibold" |
| Line Height | Text | Leave as Text. Unitless multiplier |
| Letter Spacing | Text | Leave as Text |
| Usage Rule | Text | Leave as Text. One sentence. When to use it |
| Do Not Use For | Text | Leave as Text. The misuse list is the most-read column on this table |
| Source Reference | Text | Leave as Text. Link the spec, the checker or the standard clause |
| Version | Text | Leave as Text |
| Status | Select (add options after import) | Convert to Select, add options: "Draft", "In review", "Approved", "Deprecated" |
| Approved By | Text | Leave as Text |
| Last Reviewed | Date | Convert to Date |
| Notes | Text | Leave as Text |

The rows above are documentation examples only. Emit empty templates unless the user explicitly requests examples. Anything measured gets a ratio; anything not measured is Not applicable, never a guess.

Field Reference

Select Options

Token Group

Brand | Semantic | Neutral | Typography | Layout | Component

Token Type

Color | Typography | Spacing | Radius | Shadow | Border | Motion | Breakpoint | Z-Index

Unit

hex | px | rem | em | ms | s | unitless | count

WCAG Level - the four legal values. Fail - do not ship exists so a failing pair is recorded and visible, not quietly deleted.

AA | AAA | Fail - do not ship | Not applicable

Status

Draft | In review | Approved | Deprecated

Relations

None. This table is standalone. Where a colour is measured, Contrast Against names another token in the same table by Token Name - that is a lookup on a text field, not a database relation, and it resolves in SQL as a self-join:

SELECT a.token_name, a.value, b.value, a.contrast_ratio
FROM design_token a
JOIN design_token b ON a.contrast_against = b.token_name
WHERE a.token_type = 'Color' AND a.status = 'Approved';

Examples

Prompt

We picked a blue from a logo two years ago and nobody knows which one it is. It looks
fine on the laptop but the founder says people struggle on the website.

Context first - one question per message, nothing already answered:

Q: What is the business called, and what does one line of it actually do? A: Example Retail, home and kitchen goods in Example City.

Q: Is the blue fixed by anything - a shop sign, a customer, a printed material? A: It is on the shopfront and on the invoices.

Q: Any existing logo, documents or website colours to match? A: A logo from 2022, and a PDF price list.

Q: Website only, or print and social as well? A: All three.

Q: Who decides a colour change? A: The founder.

Recommended next step - offered, not built:

One token register with that blue as color.brand.primary, a neutral ramp derived from it, a light and a dark variant, the type scale, a spacing unit, and a measured contrast ratio on every pair that carries text - the blue on white is the pair most likely to fail.

Workflow: Seed colour confirmed from the sign → Neutral ramp derived → Semantic roles assigned → Every text pair contrast-measured → Type scale fixed → Spacing unit set → Register approved and versioned

Want the CSV, SQL DDL, JSON Schema and Notion mapping for this?