Spec Layer
spec-layer.com
- Free plan
- Web
What Spec Layer is
Spec Layer turns Figma components, variables, and styles into specifications for teams and AI coding agents. It adds component specifications and foundation references to the Figma canvas, with measurements, variants, anatomy, and token tables. Component specifications can be copied as YAML or Markdown; variables and styles can be exported as DTCG JSON. Its CLI brings published specifications and design tokens into a repository and can produce CSS custom properties for web projects. The CLI requires Node.js 22 or newer. Other supported outputs include canonical JSON for validation, inspection, and integrations; Teams plans include custom integration scoping. The MIT-licensed plugin has no user accounts, and deterministic measurements, states, anatomy, and tokens run locally in Figma. When AI writing is requested, derived component summaries and sometimes rendered images go through Spec Layer to Anthropic. Foundation writing sends token names and resolved values without a component image. The free plan covers one Figma file and ten publishing updates per month. Pro costs $7.99 a month; a published library is limited to 5 MB.
Who it suits
Spec Layer suits teams documenting Figma design systems and sharing component guidance with coding agents. It can also suit web teams that want published design tokens delivered to a repository.
Good to know, and what to watch
- Documents measurements, variants, anatomy, and token tables on canvas.
- Exports component specifications as YAML or Markdown.
- CLI delivers tokens and specifications to a repository.
- Deterministic design details run locally in Figma.
- MIT licensed, with no user accounts.
- CLI requires Node.js 22 or newer.
- Free plan allows one Figma file.
- Free plan has ten publishing updates monthly.
- Published libraries have a 5 MB limit.
Plans
Every plan and priceTeams
Free
Pro
Free plan
The verdict
Spec Layer connects Figma documentation with formats for coding agents and repository workflows. Check its file and publishing limits, and note that requested AI writing sends derived content through Spec Layer to Anthropic.
What the maker publishes
- AI context export
- Component specifications can be copied for coding agents as YAML or Markdown, while variables and styles can be exported as DTCG JSON. source
- AI data flow
- When AI writing is requested, a derived component summary and sometimes a rendered image are sent through Spec Layer to Anthropic; foundation writing sends token names and resolved values without a component image. source
- Canvas documentation
- It creates component specifications and foundation references on the Figma canvas, including measurements, variants, anatomy, and token tables. source
- CLI requirement
- The CLI requires Node.js 22 or newer. source
- Integrations
- Canonical JSON is provided for validation, inspection, and building integrations, while Teams plans include custom integration scoping. source
- License
- Spec Layer is MIT licensed. source
- Output formats
- The product supports canvas documentation, component YAML, component Markdown, foundation DTCG JSON, web CSS tokens, and canonical JSON. source
- Privacy
- The plugin has no user accounts, and deterministic measurements, states, anatomy, and tokens run locally in Figma without sending data elsewhere. source
- Publishing limit
- A published library has a 5 MB limit, and the Pro license can publish up to 10 libraries. source
- Purpose
- Spec Layer turns Figma components, variables, and styles into precise specifications for teams and AI coding agents. source
- Repository delivery
- The CLI pulls published component specifications and design tokens into a repository and can export CSS custom properties for web projects. source
- Security architecture
- The backend runs on Cloudflare Workers, the Anthropic API key remains server-side, prompts, rendered images, and generated text are never logged, and payment cards are handled by Lemon Squeezy. source
- Support
- Spec Layer is maintained by an independent developer and support replies are expected within a few business days. source
Checked against the maker's pages on 30 Sept 2026.