Skip to content

Figma Design to Code skill

Figma's skill for turning a Figma frame into code in your project, reusing your components and assets instead of pasting generated markup.

Official · Figma
AuthorFigma
LicenceNo licence file; Figma Developer Terms apply
Sourcefigma/mcp-server-guide · SKILL.md on GitHub
Project websitedevelopers.figma.com/docs/figma-mcp-server
Works inClaude Code · Codex, Gemini CLI, Cursor, Copilot
Read from the source on

The figma-design-to-code skill is Figma's official agent skill for implementing a Figma design as code: you give the agent a link to a frame, and it builds that screen or component in your project. Figma makes it a mandatory prerequisite for the MCP server's get_design_context tool, so it is loaded every time an agent pulls a design out of Figma.

It exists because the code Figma returns is a reference, not a finished component. The skill makes the agent treat it as a high-fidelity, non-interactive prototype: translate absolute positioning into your layout system, reuse your existing components, tokens and Code Connect mappings, place every image and SVG exactly as designed, and then check the result against a screenshot of the design. It only reads from Figma; it is the opposite of figma-generate-design, which writes code into Figma.

From the skill's SKILL.md: treat returned code as a high visual fidelity non-interactive prototype

This is an official skill from Figma. Like any skill it gives the agent instructions and may include scripts it runs, so install it from the source linked here, not from a copy elsewhere.

What it does

  1. Asks for a screenshot with the first get_design_context call (or calls get_screenshot) and uses it as the visual target the finished code must match.
  2. When a response is flagged as sparse, matches its child node IDs to the screenshot and requests the visible children in one parallel batch, then implements only from those detailed responses.
  3. Inspects your project and installed design-library packages for components, assets and tokens that match, and reuses or composes them instead of writing raw markup, inline styles or hard-coded values.
  4. Uses the component mapped by Code Connect wherever one exists, even when wiring it up takes more effort.
  5. Downloads every static image and SVG to your project and keeps each in its exact place and size, leaving no temporary Figma asset URLs in the code; data-driven images stay dynamic.
  6. Verifies only the requested screen or component: every asset file, its slot, its call site and its rendered size, fixing in-scope mismatches and noting, not changing, unrelated ones.

What you get

  • Code changes in your repository: components or a page in your stack (React and Tailwind by default from Figma, adapted to your project), plus downloaded image and SVG assets.
  • No changes to the Figma file.

When to use it

  • A designer hands over a Figma frame and you want it implemented in your existing codebase and design system.
  • You use Code Connect and want mapped components reused instead of re-created.
  • Previous design-to-code attempts produced absolute-positioned markup, missing icons or links to expired Figma images.

When not to use it

  • You want the opposite direction (code to Figma): use figma-generate-design.
  • The design is animated and you need the motion too: add figma-implement-motion, which handles get_motion_context alongside this skill's structure.
  • The target is SwiftUI: Figma's figma-swiftui skill covers SwiftUI and Figma translation specifically.
  • You need pixel-perfect output with no existing components: the skill still works, but its main value is reuse of what your project already has.

What it needs

  • The Figma MCP server connected and signed in. Figma recommends the remote server; get_design_context also exists on the desktop server, which needs the Figma desktop app and a Dev or Full seat on a paid plan.
  • A Figma account with access to the design. This skill only reads, so no write access is needed, but read tools are rate-limited: up to 20 calls a month on the Starter plan or a View or Collab seat, and daily and per-minute limits on paid Dev and Full seats.
  • Your project open in the agent, so it can find components, tokens and the right place for assets.
  • A link to the frame or layer (Figma URLs carry the file key and node ID).

Files in the skill

File or folderWhat it is for
SKILL.mdThe whole skill, in three short sections: context (screenshots, sparse responses), implementation (reuse, Code Connect, assets) and verification.

Licence as found in the repository on 2026-11-15: No licence file; Figma Developer Terms apply.

Example prompts

Type one of these after installing; naming the skill makes sure it is used.

  • Implement this Figma frame as a React component in src/components using our existing Button and Card: <frame URL>.
  • Build the settings page from this Figma link in our Next.js app, with Tailwind and our design tokens, and download the icons into public/icons.
  • Turn this Figma design into plain HTML and CSS, matching the screenshot exactly: <frame URL>.
  • Update ProfileHeader.tsx to match the latest Figma design; reuse the Avatar component Code Connect maps to.

How to install figma-design-to-code

Connect the Figma MCP server

In Claude Code, Figma's plugin from Anthropic's official marketplace adds the remote MCP server and this skill with the rest of Figma's skills. After installing, run /mcp and sign in to Figma. Or add the server with the second command and copy just this folder: the skill has no other files and no companion skill to copy.

claude plugin install figma@claude-plugins-official
# or, by hand:
claude mcp add --transport http figma https://mcp.figma.com/mcp

Cursor, Gemini CLI, VS Code and the desktop server

Figma's guide installs the same plugin in Cursor (/add-plugin figma) and Gemini CLI (an extension, then /mcp auth figma), and connects VS Code through an HTTP MCP server. If your organisation uses the desktop server instead, its address is http://127.0.0.1:3845/mcp while the Figma desktop app is open with the server enabled in Dev Mode.

/add-plugin figma
gemini extensions install https://github.com/figma/mcp-server-guide

Not for claude.ai uploads or the Claude API

The skill only guides calls to the Figma MCP server's design-context tools. The Claude API skills sandbox has no network access and we have not verified a claude.ai setup.

A skill is a folder with a SKILL.md file. You install it once; the agent reads its short description at the start of each session and loads the full instructions only when a request matches. Pick the place you use:

Claude Code: from the plugin marketplace

  1. Start Claude Code in a terminal with claude.
  2. This plugin is in Anthropic's official marketplace, which Claude Code adds for you the first time you start it.
  3. Install the plugin. Claude Code opens its details first: check what it will add (skills, hooks, MCP servers), then choose whether it is for you, for everyone on this repository, or for you in this repository only.
    /plugin install figma@claude-plugins-official
  4. Type / and look for /figma:figma-design-to-code, or simply ask for the task.

From your shell instead of a session:

claude plugin install figma@claude-plugins-official

The plugin installs as a whole, so other skills come with it, including figma-3d-transforms, figma-code-connect, figma-code-workflows, figma-create-new-file, figma-generate-design, figma-generate-diagram, figma-generate-library, figma-generative-plugins, figma-implement-motion, figma-shaders, figma-spec, figma-swiftui, figma-use, figma-use-figjam, figma-use-motion, figma-use-slides.

Claude Code: copy the folder into your skills folder

Put the skill's folder in ~/.claude/skills/ to use it in every project on this computer, or in .claude/skills/ inside a repository to share it with everyone who works there (commit it).

git clone --depth 1 https://github.com/figma/mcp-server-guide.git
mkdir -p ~/.claude/skills
cp -R mcp-server-guide/skills/figma-design-to-code ~/.claude/skills/

Claude Code notices new or changed skills in these folders during the session. If the skills folder did not exist when the session started, run /reload-skills once. Then type /figma-design-to-code or just describe the task.

On Windows, run the commands in Git Bash or WSL, or copy the folder into the same place with File Explorer.

Codex, Gemini CLI, Cursor and GitHub Copilot

These agents read the same SKILL.md folders. The one folder all four accept inside a project is .agents/skills/, so this copies the skill there:

git clone --depth 1 https://github.com/figma/mcp-server-guide.git
mkdir -p .agents/skills
cp -R mcp-server-guide/skills/figma-design-to-code .agents/skills/
AgentProject folderYour user folderNotes
OpenAI Codex
.agents/skills/
~/.agents/skills/
Type $ or run /skills to pick a skill; Codex also chooses one when your task matches.
Gemini CLI
.agents/skills/
.gemini/skills/
~/.agents/skills/
~/.gemini/skills/
Or install from Git with gemini skills install <git URL> --path <folder> --consent. Google replaced Gemini CLI with Antigravity CLI for free-tier and Google One users on 18 June 2026; Antigravity uses .agents/skills/ in a project and ~/.gemini/config/skills/ for you.
Cursor
.agents/skills/
.cursor/skills/
~/.agents/skills/
~/.cursor/skills/
Type / in Agent chat to pick a skill. Cursor also reads Claude Code's .claude/skills/ folders.
GitHub Copilot
.github/skills/
.agents/skills/
.claude/skills/
~/.copilot/skills/
~/.agents/skills/
Works in Copilot CLI, the cloud agent, and agent mode in VS Code, Visual Studio and JetBrains (IDE support depends on the version). Or run gh skill install OWNER/REPOSITORY SKILL (public preview, GitHub CLI 2.90.0 or later).
Only install skills you trust. Anthropic's advice is to use skills only from trusted sources, and to read every file of any other skill before using it: a skill can direct the agent to run commands, read files or send data in ways its description does not mention.

Sources for these steps

Checked against each vendor's documentation on 2026-11-14:

Reuse beats regeneration

Most of the skill is about not reinventing your project. Before editing, the agent must look through likely project paths and installed design-library packages, reuse matching components, assets and tokens, and change them only when they cannot express the design. A Code Connect mapping is always used at its mapped node unless the component truly cannot be configured.

Absolute positioning from Figma is translated into your project's own layout unless the design really needs fixed positions, and interactive parts of the design become interactive code.

Assets are where most attempts fail

The skill is strict about images and SVGs: each is downloaded as the design-context response describes and used in its exact position, never redrawn, inlined, swapped for a near match or left pointing at a temporary Figma URL. SVGs keep their own width and height. A single substituted or misproportioned asset counts as a failure in its verification step.

Questions about figma-design-to-code

How is figma-design-to-code different from figma-generate-design?

They go in opposite directions. figma-design-to-code reads a Figma design and writes code in your repository. figma-generate-design reads your code and builds editable screens in Figma. Many teams install both through Figma's plugin.

Does the code come out as React?

Figma's get_design_context returns React and Tailwind by default and can return other frameworks such as Vue, plain HTML and CSS or iOS when asked. The skill then adapts that output to your project's stack, conventions and existing components rather than pasting it.

Do I need a paid Figma seat?

Not to write, because this skill only reads. But read tools are limited: up to 20 calls a month on the Starter plan or with a View or Collab seat. Dev and Full seats on Professional, Organization and Enterprise plans get daily and per-minute limits instead, which suit regular use.

Why does it insist on a screenshot?

The screenshot is the visual target the finished code must match. The skill forbids using it as an image in the code: the implementation must come from the detailed design context, with real assets downloaded and placed.

Is the figma-design-to-code skill free?

The file is free to install but not open source: the repository has no licence file and Figma's Developer Terms apply. It needs a Figma account and counts against Figma's read-tool rate limits for your plan and seat.

More AI skills

Related skills

Skill categories