Specs 2

Specs 2 is an ecosystem for specs driven component development. Via the plugin, turn a selected Figma component or component set into a compact, complete, schema-valid specification — computed directly from the file, not inferred from a screenshot. From the command line, generate specs from an entire Figma file, render assets back into Figma, and generate code and analysis data. One structured source of truth that engineers and AI agents build on across web, iOS, and Android..com SiteDocsFeature & Bug RequestsSlack SupportOn-canvas specs, in full detailSelect a component and see its entire structure surfaced right on the canvas: element anatomy, prop types and defaults, layout and positioning, mode-specific styling, and every style value with its token reference. No annotation, no manual redlining — just what's actually in the file.

Every variant, every layer, every detail, every time. Not a some sample or shallow DESIGN.md file.

This is computation over inference: enumerate the variants, diff the styles, record what changes — mechanical, not guesswork. Specs 2 evaluates every variant combination, every prop binding, and every slot, where Specs Classic and manual inspection alike miss multi-prop interactions and edge-case combinations.

The result is deterministic and repeatable — the same component always produces the same spec. And what you see on the canvas is the same structure you can give LLMs as YAML or JSON data.

Built for the full Specs ecosystemOutput isn't a dead end — it's structured data ready to move downstream. Copy spec data directly into LLM chats for AI-assisted development. Get started with the command line tools with your LLM's help to:generate specs in bulk, whether on a schedule, on-demand, or per releaserender a spec back into Figma via a bridge using command line or agent chatstransform Figma data into state-aware CSS, TypeScript contracts, or Storybook storiesanalyze for drift and quality of how props, variables, slots, layers and more are applied across an entire libraryAll this from the same deterministic data this plugin produces on canvas.