Fixed set for the use_figma scripts of the work-with-design-systems skill. Audits Figma design system components for accessibility, token compliance, naming quality, interactive states, and detached instances.
This server defines 6 audit/inspection tools for Figma design systems. All tools have descriptions (50-150 chars, within baseline 34-392), but critical gaps undermine usability: (1) Input schemas are present but minimal, only 1-2 parameters per tool, no type declarations visible in the schema objects provided (e.g., COMPONENT_SET_ID and PAGE_ID lack explicit 'type' fields in the JSON schema structure shown). (2) Parameter descriptions exist but are terse (10-30 chars), below the 72-char baseline for A+ tools. (3) No output schemas documented, LLMs cannot predict return structure. (4) No error handling guidance, tools return {error: ...} but do not explain recovery steps. (5) Tool names are action-oriented (audit-*, inventory) but descriptions lack WHEN/WHY context. (6) No mention of pagination, limits, or result caps despite tools scanning potentially large Figma files.
Module 3 — Accessibility (WCAG 2.1 AA). Audits a component set for accessibility compliance including color contrast (WCAG 1.4.3), touch target size (WCAG 2.5.5), font size minimum (12px), and focus indicator presence (WCAG 2.4.7).
Module 4 — Detached instances scanner. Walks a Figma page and collects FRAME nodes that were previously instances but are now detached from their source components, using the canonical detachedInfo property.
Module 5 — Naming quality. Flags layers inside a component set that carry Figma's auto-generated names (Frame 12, Rectangle 4, etc.), deduplicating across variants.
Module 2 — Interactive states. Infers component archetype from the component set name and checks whether variants cover the expected interactive states for that archetype (Default, Hover, Pressed, Focused, Disabled, etc.).
Module 1 — Token compliance. Audits a component set for design token usage, checking for unbound fills, strokes, padding, itemSpacing, corner radius, text styles, and raw opacity values on bound paints.
Module 0 — Component inventory. Scans a Figma page and returns a list of all component sets and standalone components with metadata including variant count, description presence, and page location.
Input schemas lack explicit type declarations. Parameters shown as {COMPONENT_SET_ID: {type: 'string', description: '...'}} but no JSON Schema 'type' field visible in structured schema objects.
No output schemas documented. Tools return complex objects (checks[], passed, total, percentage, detachedInstances[], etc.) but LLMs have no schema to predict structure. Baseline: 100% of A+ tools have documented return types.
Parameter descriptions are terse (10-30 chars). Baseline: 72 chars average for A+ tools. E.g., 'The Figma node ID of the component set to audit' is only 50 chars and lacks format/constraint details. No mention of valid ID format, length, or how to obtain IDs.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 41 | 2026-07-28+ | v2 |
Error handling lacks recovery guidance. Code returns {error: '...'} but does not tell LLM what to do next. Per pattern:recovery-guide, errors must guide the agent: 'Node not found. Verify the ID via inventory() first.'
No pagination or result limits documented. Tools scan Figma pages/component sets which may contain hundreds of items. No mention of max results, pagination tokens, or performance guidance. Baseline: tools returning lists should accept limit/offset and return total count.