map-design-language
Skill4 starsChanged 7 months agoMITView on GitHub
Research a project's established design language and write or refresh a lean meta/DESIGN.md that grounds mockups, critiques, and new UI. TRIGGER when: user asks to map, document, bootstrap, or refresh the design language; a skill needs design grounding and meta/DESIGN.md is missing, incomplete, or stale. SKIP when: meta/DESIGN.md is current (read it); the project has no established UI and needs a new aesthetic (offer frontend-design).
In this repository42 other files it gives its agents.
- AGENTS.mdAGENTS.md
- agent-soulSkill
- audit-transcripts-for-learningsSkill
- best-answerSkill
- check-coherenceSkill
- compress-fileSkill
- deep-modulesSkill
- design-critiqueSkill
- durable-docs-updateSkill
- execute-chatSkill
- execute-planSkill
- explain-deeplySkill
- explain-simplySkill
- find-gapsSkill
- fix-verify-loopSkill
- generate-mockupsSkill
- grill-meSkill
- handoffSkill
- interview-meSkill
- jtbdSkill
- memory-pruneSkill
- multi-agent-analysisSkill
- place-factSkill
- post-purchase-ui-extensionSkill
- product-interviewSkill
- prove-behaviorSkill
- refine-fileSkill
- second-opinionSkill
- sentry-analysisSkill
- shopify-dev-mcpSkill
- structure-proseSkill
- tech-designSkill
- test-completed-planSkill
- tighten-fileSkill
- tighten-instructionSkill
- tldrSkill
- triageSkill
- trim-specSkill
- two-pass-reviewSkill
- validate-answerSkill
- vet-factSkill
- write-planSkill
The file
We split this skill into the rules it gives an agent
12 rules. About 0 are common, 6 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 19
When no established UI exists, stop and offer to invoke frontend-design.
- Distinctive: line 34
If the repository layout is not clear after a light scan, dispatch one read-only subagent to return candidate surfaces and canonical …
- Distinctive: line 63
Keep each section to the project-used subset.
- Very specific: line 64
Write Overview from observable evidence.
- Distinctive: line 65
Keep universal quality rules such as contrast, keyboard access, focus visibility, and touch-target size in consuming skills.
- Very specific: line 69
Before writing, present this checkpoint and confirm via AskUserQuestion (Write it / Adjust):
- Distinctive: line 84
Read the complete result cold, fix any conflict or missing source, and report the path plus the sections changed.
- Distinctive: line 90
Use plain Markdown without YAML frontmatter.
- Very specific: line 159
Keep one global Overview.
- Very specific: line 160
Add ## Surfaces at a glance with Surface, Audience, Path, Toolkit, and Styling model columns.
- Very specific: line 161
Add one ## <Surface> block per surface and use Toolkit through Do's and Don'ts as ordered ### subsections.
- Distinctive: line 162
Keep a rule global when it applies to every surface.
≈ “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
Map Design Language
Research the stable visual rules already established in a project and record them in one meta/DESIGN.md. The file explains how the UI should look and behave; runtime styling sources keep ownership of exact values.
Protocol
Step 1 — Detect mode
Check for meta/DESIGN.md:
- Missing → bootstrap: research the established UI and create the file.
- Exists → read the whole file, then use the Step 2 source scan to decide whether it needs a refresh.
When no established UI exists, stop and offer to invoke frontend-design; this skill records a design language and does not invent one.
Step 2 — Map surfaces and canonical sources
Identify whether the repository has one design language or several distinct UI surfaces with different toolkits, token vocabularies, audiences, or render targets. Inspect the root package manifests, workspaces, UI directories, framework imports, and existing product or brand guidance.
- One design language → use one flat document without a surfaces table.
- Several surfaces → keep one
meta/DESIGN.md; add a surfaces table and surface-specific blocks. Never mix tokens or toolkit rules across surfaces.
Find the runtime sources that own exact styling values. Sources can include CSS files, design-token files, Tailwind or framework theme configuration, component-library theme objects, and platform-owned token documentation.
- Canonical source exists → record token names, roles, and source paths; do not copy the complete palette or scale into
DESIGN.md. - Values are distributed → record only verified values that repeat or materially constrain new UI, cite their source paths, and state that
DESIGN.mdis a maintained reference rather than runtime authority. - Library or platform owns appearance → record the allowed project-used tokens or props and link the authoritative documentation; do not mirror the full library.
If the repository layout is not clear after a light scan, dispatch one read-only subagent to return candidate surfaces and canonical styling sources with paths. Confirm a genuinely ambiguous surface boundary with the user.
For an existing meta/DESIGN.md, stop and report that no update is needed when the scan finds no conflict, material gap, or caller-named change. Otherwise, enter refresh mode for the affected sections.
Step 3 — Research the established language
Dispatch read-only subagents in proportion to the surface count:
- One design language → one subagent covers the complete document.
- Several surfaces → one subagent per surface.
- Refresh → inspect only affected surfaces and sources, while preserving current sections and human edits elsewhere.
Each subagent returns concise facts with supporting paths:
- audience and the UI's established visual character, density, hierarchy, and dominant interaction approach;
- framework, component library, styling model, project-used components, official documentation, and inherent toolkit constraints;
- semantic color roles and themes; typography roles and hierarchy; layout, responsive behavior, and navigation model; shape and depth;
- motion rules when an established motion system exists; iconography and imagery when relevant;
- repeated component combinations and cross-component patterns, not individual page layouts;
- interface-copy rules for labels, errors, empty states, statuses, capitalization, and tone;
- project-specific do's and don'ts that prevent likely mistakes;
- canonical sources, ownership boundaries, related guidance, and refresh triggers.
Derive every fact from the current UI, code, or maintained product and brand documents. If the evidence does not establish a rule, report it as absent instead of inventing one.
Step 4 — Assemble and checkpoint
Assemble the DESIGN.md template with stable rules and source links:
- Keep each section to the project-used subset; exclude feature behavior, page-specific layouts, exhaustive library catalogs, and universal design advice.
- Write
Overviewfrom observable evidence. Do not substitute vague adjectives such as “modern,” “clean,” or “intuitive” for concrete rules. - Keep universal quality rules such as contrast, keyboard access, focus visibility, and touch-target size in consuming skills. Record a project-specific accessibility mode, constraint, or exception in the section it affects.
- List canonical sources once in
Maintenance. Add an inline path only for a non-obvious pattern or important constraint. - Omit
Motionwhen no deliberate motion system exists. OmitIconography and imageryonly when neither applies.
Before writing, present this checkpoint and confirm via AskUserQuestion (Write it / Adjust):
**Design language summary:**
- Surfaces: <single design language | surface — audience — toolkit, one per line>
- Canonical sources: <paths>
- Overview: <established character, density, hierarchy, and interaction approach>
- Stable rules: <most important toolkit, pattern, content, and guardrail facts>Step 5 — Write or refresh
- Bootstrap → create
meta/DESIGN.mdfrom the template. - Refresh → edit only affected sections; preserve current facts and human edits elsewhere.
Read the complete result cold, fix any conflict or missing source, and report the path plus the sections changed.
DESIGN.md template
Use plain Markdown without YAML frontmatter. The single-surface form is canonical:
# Design reference — <project>
<One sentence naming the UI surface and purpose of this reference.>
## Overview
<The audience, established visual character, density, hierarchy, and dominant interaction approach. Use concrete, observable rules.>
## Toolkit
- **Framework:** <framework and version>
- **Component library:** <library and version>
- **Styling model:** <CSS, utilities, theme props, platform-owned, or mixed>
- **Documentation:** <authoritative links>
- **Constraints:** <inherent toolkit limits only>
## Colors
- <Semantic roles, theme relationships, and colors reserved for specific meanings. Name tokens instead of copying code-owned values.>
## Typography
- <Font roles for UI, prose, headings, code, or data; hierarchy and density; special rules such as tabular numbers or reading measure.>
## Layout
- <Page width or grid model, main regions, responsive behavior, density, spacing approach, and navigation changes across viewports.>
## Shape and depth
- <Corner, border, divider, shadow, card, and surface-hierarchy rules. Keep this section even when the system is simple.>
## Motion
- <What moves, why, and the established timing or easing rules; patterns to avoid. Omit this section when no deliberate motion system exists.>
## Iconography and imagery
- <Approved icon set, icon treatment, photography or illustration style, asset sources, and restrictions. Omit when neither applies.>
## Components
- <Project-used components and stable usage rules.>
## Patterns
- <Repeated component combinations and cross-component behavior; no page-specific layouts.>
## Content
- <Interface-copy rules for labels, validation, empty and error states, statuses, capitalization, and tone. Exclude marketing strategy and page content.>
## Do's and Don'ts
- <Project-specific guardrails that prevent likely mistakes. Write `None established beyond the rules above.` when no additional guardrail is supported.>
## Maintenance
- **Canonical sources:** <paths that own exact values and implementation facts>
- **This file owns:** <design rationale and stable rules>
- **Refresh when:** <surface, toolkit, token-role, pattern, or content-rule changes that affect this reference>
- **Related guidance:** <brand, product, or other canonical documents; omit when none>For several surfaces:
- Keep one global
Overview. - Add
## Surfaces at a glancewithSurface,Audience,Path,Toolkit, andStyling modelcolumns. - Add one
## <Surface>block per surface and useToolkitthroughDo's and Don'tsas ordered###subsections. - Keep a rule global when it applies to every surface; otherwise put it only in the governing surface block.
- End with one global
Maintenancesection.
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.

