dads
Skill0 starsChanged 3 months agoMITView on GitHub
Design tokens, components, and usage rules of the Digital Agency Design System (DADS, デジタル庁デザインシステム). Use when output should look DADS-compliant or markup is audited against DADS.
In this repository11 other files it gives its agents.
- build-justfileSkill
- c4-sequenceSkill
- create-issueSkill
- create-prSkill
- create-repoSkill
- handoverSkill
- mouse-doctorSkill
- oss-bus-factorSkill
- ponytail-reviewSkill
- tmuxSkill
- writing-documentsSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 59
Read references/components.md before touching any component.
≈ “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
DADS — Digital Agency Design System
Knowledge for making output look and behave like the Digital Agency Design System (デジタル庁デザインシステム, DADS) — Japan's government design system, MIT-licensed tokens and components published by digital-go-jp.
Scope: visual authority only
This skill answers one question: what does DADS-compliant look like, and how do I get there from primary sources? It owns colors, typography, component markup, usage rules, and the compliance policy below.
It deliberately does not own structure or medium. Compose it:
- New artifact →
artifact-design(and the Artifact tool's constraints) own the medium; this skill owns the look. Seereferences/artifact.mdfor where the two collide (CSP, dark theme). - Existing frontend restyle → the codebase owns structure; change only what the three-layer check below flags as deviating.
The three-layer model
DADS compliance has three layers. Applying only the first produces "なんちゃって DADS" — colors that look right on markup that an official page would never emit. Real deviations (a table wrapped in a rounded container, a bold chip label) are invisible at the token layer and only surface when compared against official component CSS.
| Layer | What it is | How to get it | What it buys |
| tokens | values — color hex, font, spacing, radius, elevation | pinned copy in references/tokens.css | plausible colors |
| components | shapes — markup, dimensions, dads-* class names | fetch per references/components.md | correct forms |
| guidelines | rules — when to use which | curated copies in references/guidelines/ | correct choices |
Work top-down through all three for anything user-facing. Token-only application is acceptable solely for throwaway internal sketches.
Key token values (design-tokens 2.0.1)
The full file is references/tokens.css (226 lines, pinned verbatim). The values that
make or break the DADS look:
- Key color:
--color-key-*= blue family; the anchor is blue-900#0017c1(the Digital Agency blue). - Semantic: error-1 = red-800
#ec0000/ warning-yellow-1 = yellow-700#b78f00/ success-1 = green-600#259d63. - Font:
'Noto Sans JP', ...— mono is'Noto Sans Mono'. - Body text: 16px, line-height 1.75, letter-spacing 0.02em — omit the letter-spacing and nothing will read as DADS no matter how right the colors are.
- Heading sizes: 64/57/45/36/32/28/24/20/18/16 (there is no 22).
- Radius scale: 4/6/8/12/16/24/32/full.
Components
Read references/components.md before touching any component. The one rule worth
stating here: port official dads-* class names and CSS verbatim — never transcribe
values into homemade classes, or the result can no longer be diffed against the official
source when it drifts.
Pick components from references/guidelines/components-index.md (the official catalog).
Per-component usage guidelines for the ones already battle-tested (notification-banner,
table, heading) plus color/typography foundations are in references/guidelines/.
Policy (always in force)
- No impersonation. Tokens and components are MIT; the Digital Agency's logo, illustration assets, and name are not part of that grant. Never produce something that reads as an actual Digital Agency page. DADS-styled ≠ Digital-Agency-branded.
- Label extensions as extensions. Official DADS is light-only (
whitehardcoded). A dark theme is a DADS-compliant extension, not official DADS — derive it perreferences/artifact.mdand say so in the page footer or equivalent. - Attribution for guideline excerpts follows the official notice:
https://design.digital.go.jp/dads/introduction/notices/ — each bundled guideline file
keeps its
source_urlfrontmatter for this reason.
Refreshing pinned sources
The pins keep this skill reproducible; refresh deliberately, not per-use.
# tokens — check the published version before overwriting the pin (currently 2.0.1)
curl -sL https://unpkg.com/@digital-go-jp/design-tokens@latest/package.json | jq -r .version
curl -sL https://unpkg.com/@digital-go-jp/design-tokens@latest/dist/tokens.css
# guidelines — the ZIP link is dated with no latest-URL; scrape it from the resources page
# (bundled copies are from dads-markdown-20260805.zip)
curl -sL https://design.digital.go.jp/dads/resources/ | grep -oE '/dads/dads-markdown-[0-9]+\.zip'Components are fetched live (see references/components.md) rather than pinned — the
set is large, and live fetch preserves diff-ability against the current official source.
Dead ends (do not revisit)
design.digital.go.jp/llms.txt— 404, does not exist.tokens.cssat the npm package root — the real file is underdist/; list withhttps://unpkg.com/@digital-go-jp/design-tokens@latest/?metafirst.digital-go-jp/tailwind-theme-plugin— a repackaging of the tokens, zero new information unless the target project already uses Tailwind.components/chip-labelguideline — official doc is a "準備中" stub; the porting notes inreferences/components.mdare the best available source.
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.

