agentleFS
Gaurav890Gaurav890/everything-agentic-engineering/.claude/skills/design-intake/SKILL.md

design-intake

Skill1 starsChanged 2 months agoMITView on GitHub

Collect the minimum product-specific visual constraints, generate comparable design directions, and obtain explicit approval before design-system or token changes. Use for design-critical greenfield products, major redesigns, inherited brands, theme creation, or substantial UI work where palette, typography, density, geometry, motion, accessibility, platform, or brand authority is unclear.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 4 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not ask the user to design the token system.

    line 8
  2. Very specific:

    Read .agentic/project.json and ask only about active platforms.

    line 24
  3. Distinctive:

    Ask one to three short groups at a time.

    line 30
  4. Distinctive:

    If the user has no exact color preference, do not force a hex choice.

    line 49
  5. Very specific:

    When color roles or typography remain genuinely open, run ./agentic design resources --need palette --need typography.

    line 52
  6. Very specific:

    Write facts, constraints, preferences, assumptions, open questions, and approval authority to docs/20-design/DESIGN_BRIEF.md.

    line 62
  7. Very specific:

    Before invoking design-system or editing canonical tokens, record:

    line 91
  8. Distinctive:

    Do not overwrite a coherent existing system with starter defaults.

    line 110

≈ “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

Design intake

Do not ask the user to design the token system. Gather intent and constraints, then translate them into reviewable directions.

Route the intake

  • Existing brand: inventory exact assets, licensed fonts, colors, Figma sources, restrictions, and the authority that may approve changes. Preserve exact brand inputs; do not approximate them from screenshots.
  • Greenfield design-critical product: collect preferences, research references, and produce comparable materially different directions. Two or three per round is a starting point, not a catalog limit.
  • Redesign: inspect the running product and current tokens first. Separate what must remain compatible from what may change, and capture baseline evidence.
  • Design is not a competitive advantage: use one restrained accessible baseline; skip expressive questions and unnecessary alternatives.

Read .agentic/project.json and ask only about active platforms. Do not ask mobile questions for a web-only product or require a Figma workflow when Figma is not part of the project.

Interview adaptively

Ask one to three short groups at a time. Do not make every question mandatory. Resolve, in this order:

  1. Authority and mode — existing brand, greenfield, or redesign; approver; fixed assets and non-negotiable constraints.
  2. Experience intent — audience, task, context, desired feelings, density, and qualities the product must avoid.
  3. Color constraints — exact existing colors or preferred/forbidden hue families, neutral temperature, restrained/expressive use, data visualization, and required light/dark/high-contrast modes.
  4. Type and composition — licensed fonts, reading/data needs, typography personality, spacing density, geometry, elevation, and content width.
  5. Interaction and motion — input methods, motion intensity, reduced motion, signature behavior, and agentic states when applicable.
  6. Accessibility and platform — WCAG target, localization, color-vision, touch/keyboard, and active web/native surfaces.
  7. References — examples the user values and the exact quality to learn from; record what must not be copied.

If the user has no exact color preference, do not force a hex choice. Generate coherent proposals from product intent and show them in context.

When color roles or typography remain genuinely open, run ./agentic design resources --need palette --need typography. If Realtime Colors is recommended, invite the owner to use it manually as a generic live comparison aid. Bring back the chosen/rejected roles, type candidates, share URL or screenshot, and contrast readings. Do not copy its website source, accept five generic colors as the final semantic system, or skip product-specific previews and independent WCAG verification.

Create durable intake

Write facts, constraints, preferences, assumptions, open questions, and approval authority to docs/20-design/DESIGN_BRIEF.md. Never present an assumption as an approved decision.

Generate comparable directions

For design-critical work, create a reviewable set of coherent directions in docs/20-design/DESIGN_DIRECTIONS.md. Each direction must include:

  • a name and product-specific thesis;
  • palette roles for all requested modes, not swatches alone;
  • typography, density, geometry, elevation, icon, and motion choices;
  • representative normal, loading, empty, error, focus, and agent states;
  • realistic rendered previews at relevant breakpoints;
  • WCAG contrast evidence for required text and UI pairs;
  • reference provenance, tradeoffs, risks, and implementation cost.

Alternatives must differ in meaningful system decisions, not only accent color. References are inputs to synthesis; do not clone a product or library aesthetic.

Generated projects use project-onboarding and an open-ended candidate catalog. Register actual local preview routes through ./agentic design propose; do not force the bundled reference styles on custom or existing-brand projects. List preview code and its shared appearance-affecting dependencies/assets so approval fingerprints detect drift. Record complete intake and reviewed screenshot paths before approval; see docs/60-tooling/PROJECT_ONBOARDING.md.

Approval gate

Before invoking design-system or editing canonical tokens, record:

  • selected direction or explicitly approved synthesis;
  • approver and date;
  • accepted compromises;
  • decisions still open;
  • evidence reviewed.

Status must remain Draft or Needs approval until a human approves the direction. Silence, an agent recommendation, or an attractive screenshot is not approval. After approval, invoke design-system, then design-tokens.

Redesign contract

For redesigns, preserve baseline screenshots and audit existing token usage. Classify each proposed change as primitive, semantic, component, or structural. Token changes may retheme an interface; they do not substitute for changes to information architecture, interaction, content, or component behavior.

Do not overwrite a coherent existing system with starter defaults. Use token deprecation and a migration plan when consumers cannot change atomically.

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.