AI agents can now write directly to Figma files through the Figma MCP server, with skills shaping how they apply a team's design decisions.
Canvas access via MCP
Figma's MCP server now lets AI agents design directly on the canvas. Through the use_figma tool, MCP clients such as Claude Code and Codex can generate and modify design assets linked to an existing design system. The same server that makes this possible also means agents work within Figma's security and reliability model while accessing surfaces like Code Connect, Figma Draw, and FigJam through the Plugin API.
Codex can find and use all the important design context in Figma to help us build higher quality products more efficiently.“Codex can find and use all the important design context in Figma to help us build higher quality products more efficiently.”
The capability is native to the Figma MCP server. It works with MCP clients including Augment, Claude Code, Codex, Copilot CLI, Copilot in VS Code, Cursor, Factory, Firebender, and Warp.
Two tools, different jobs
The existing generate_figma_design tool translates HTML from live apps and websites into editable Figma layers. The new use_figma tool takes the next step: agents operate on the canvas using your design system, applying your components and variables.
The tools are meant to be used together. When designs drift out of sync with code, generate_figma_design pulls the latest UI into Figma for iteration; use_figma can then edit those designs or create new assets from system components.
Skills: encoding design intent for agents
Skills are sets of instructions written as markdown files, and they now directly shape how agents build on the Figma canvas. They spell out how a workflow runs — which steps to take, what sequencing to follow, which conventions to respect — and are meant to bridge the gap between raw model output and the specialized knowledge needed for brand-aligned design. Anyone can author a skill; no plugin work or code is required. Several of the skills launching today come from community practitioners.
The foundational skill
At the core is /figma-use, a foundational Figma skill that every other skill builds on. It gives agents a shared understanding of how Figma works, from its structure to its core principles. Teams can customize and iterate on it to match their own agent workflows.
Nine examples to start with
/figma-generate-library: create new components in Figma from a codebase/figma-generate-design: create new designs using existing components and variables/create-voice: generate screen reader specs (VoiceOver, TalkBack, ARIA) from UI specs (Ian Guisard, Uber)/cc-figma-component: generate Figma components from a structured JSON contract (Nick Villapiano, One North)/apply-design-system: connect existing designs to system components (Chris Goebel, Edenspiekermann)/rad-spacing: apply hierarchical spacing with variables and fallbacks (Nolan Perkins, Rad Collab)/edit-figma-design: orchestrate Figma design workflows using Warp and Oz (Warp)/sync-figma-token: sync design tokens between code and Figma variables with drift detection (Firebender)/multi-agent: run parallel workflows and implement designs in Augment (Augment Code)
Skills also govern how output gets refined. In a self-healing loop, an agent that generates a screen can take a screenshot and iterate on what doesn't match — and because it works against real structure, components, variables, and auto layout, those adjustments act on the system itself rather than on visual output alone. Skills also make agent behavior more predictable: since AI models are non-deterministic and the same prompt can yield different results, encoding specific steps, guidelines, and code turns conventions into rules agents follow during their work.
Availability and roadmap
The capability is free during the beta period. Figma says it will eventually be a usage-based paid feature, and it will be a paid API after the beta as the company works out how to account for agentic behavior in its paid seats. The company draws a direct line between this release and today’s larger theme of code and canvas: switching between a coding agent, the Figma canvas, or the command line should keep both in the same shared context, with Code Connect implementation covered in Figma's existing documentation.
Future work will expand what agents can do in Figma, including native AI functionality on the canvas and easier sharing and reuse of skills. Figma also plans to add more to the tool toward parity with the Plugin API, starting with image support and custom fonts. Implementation guidance is available in Figma's MCP server guide, and skill authoring is documented in the developer docs.



