Inspect, Build, and Sync Figma Design Systems1 skill
A single skill that audits design system files, creates or extends components and tokens, and optionally exports them to code.
Product & Design Blueprint
Safety checkedA single skill that audits design system files, creates or extends components and tokens, and optionally exports them to code.
Copy this prompt and paste it into your AI agent to install.
Inspects and builds Figma design systems end-to-end in two modes. Inspect mode audits components for token compliance, interactive states, WCAG accessibility, detached instances, naming quality, and description completeness, then pauses for your decision before proceeding. Build mode creates tokens with proper variable collections, constructs components with variable bindings and slots, writes structured descriptions, validates, and optionally generates tokens.css, audit scripts, and AI rules for your codebase. The skill bridges the design-to-code gap by keeping Figma structure and code exports in sync.
You invoke the skill with a request like 'audit my design system' or 'create a design system for fintech'. The skill detects mode automatically: inspect runs read-only audits with weighted scoring and always pauses before any changes; build proceeds through six sequential phases (discovery, foundations, file structure, components with slots, validation, optional code export). Within each phase, the skill pauses for your review and confirmation. Once you approve, it applies changes and validates the result before moving forward.
A single skill that audits design system files, creates or extends components and tokens, and optionally exports them to code.
Design systems in Figma stay consistent with code because the skill exports tokens, audit rules, and AI instructions that reflect the same structure the agent reads from Figma, eliminating fabrication and drift between the design tool and the IDE.
Requires the figma-use skill and Figma MCP server (remote recommended). LineHeight variables must store pixel values, not percentages, because Figma variables are unitless and interpreted as pixels when bound. Every variable must have explicit scopes (not ALL_SCOPES) and codeSyntax.WEB set, so agents receive CSS token names instead of Figma internal names. The skill treats component descriptions as plain text (no markdown) to match MCP delivery format.
Phase 6 (code export) requires file write access to your project available in Claude Code, Cursor, Codex, and similar MCP clients. When running in Claude.ai web or mobile, Phase 6 outputs files as text for you to copy manually. Inspect mode is read-only but Phase 6 can overwrite existing tokens.css if regenerated; ask the skill to append or skip if you have custom edits. The skill migration from v1.x to v2.0 changed the repo URL, so update any hardcoded links.
Inspect mode (read-only)
Audits components with six weighted modules: token compliance with severity tiers, interactive state coverage, WCAG 2.1 AA compliance, detached instance detection, naming quality, and structured descriptions. Scores each component 0–100.
Build mode with six phases
Discovery → Foundations (variables, text styles) → File structure → Components with slots → Validation → Optional code export. Pauses between phases for your review.
Slots for compound components
Uses named slots instead of detach patterns for Card, Modal, Dialog, and similar components. Agents and users can't accidentally detach, and slot dropzones stay visible to AI.
Phase 6: Code export (optional)
Generates tokens.css with three-layer indirection, CI-ready audit script, and scoped AI rules for Claude Code, Cursor, or Codex. Phase 6 is off by default and only runs on explicit request.
Mandatory structured descriptions
Every public component requires PURPOSE, BEHAVIOR, COMPOSITION, USAGE, CODE NOTES sections. Figma MCP passes these to agents as context, so they understand intent instead of guessing from visual structure.
Stop wasting rep time on weak intent signals and mismatched accounts. The workflow ranks ICP-fit buyers by readiness, finds the right contact, and supplies tailored outreach angles for the strongest opportunities.
Ishani D.
Account Manager at G2
Shoppers get accurate answers about fit, compatibility, ingredients, and delivery even when your team is offline, so fewer leave before checkout. Conversation insights also reveal which product-page details could prevent the next unanswered question.
Urska B.
GTM Lead at Yuma AI (YC W23)
Catch a few unusual tickets before a product or fulfillment issue turns into hundreds of refunds, chargebacks, and bad reviews. Cluster transcript evidence by SKU, order date, location, and carrier so operations can trace the cause and support can reach affected customers early.
Urska B.
GTM Lead at Yuma AI (YC W23)