AI tools can now work inside your Figma files. What design teams should set up
Figma has opened its canvas to more outside AI tools through MCP, adding around 20 new partners in October. They can read your designs and, with the right seat, write to them. Here is what that means and how to stay in control.
Until recently, an AI tool that wanted to use your designs had to work from screenshots or exports. Figma is changing that. Through MCP, short for Model Context Protocol, outside AI tools can now read your Figma files directly and, in many cases, make changes to them.
What Figma announced
At Config India in Bengaluru on 15 October 2026, Figma said its MCP server can now create code layers, work with Specs, make visual assets and run Weave workflows. TechCrunch reported on 20 October that Figma added around 20 new partners and named ten of them: Amp, ClickUp, Dia, Junie by JetBrains, Linear, Meta Muse, Mistral, Notion, OpenCode and Sarvam Agents. Coding tools such as Claude Code, Cursor, Codex, VS Code and Copilot CLI were already connected.
Figma's co-founder Dylan Field told TechCrunch that MCP is not a perfect protocol. Figma's results on 4 November said MCP was its fastest-growing product by weekly users, with use up three times and changes written back to files up four times in a quarter.
Reading and writing
- Reading: tools can pull design context, such as variables, components and layout, and turn designs into code.
- Writing: tools can add and change things on the canvas in Figma Design, FigJam and Slides. This needs Figma's remote MCP server, and a Full seat to write anywhere beyond your own drafts.
- Price: Figma's help centre says writing to the canvas is free during the beta and will later become a paid feature based on use.
Why it matters to a business
If a developer or a marketing tool can read your Figma files, the website, app or campaign it builds can follow your real designs more closely. The same access means a tool can change a file nobody meant it to touch. A little setup goes a long way.
What to set up
- Decide which tools may connect and who approves them. Treat a connection like giving someone access to the file.
- Let AI tools write in drafts or a copy first, and move work into the main file after a person has checked it.
- Keep your colours, type and spacing as variables and your parts as components. AI tools follow a tidy design system far better than loose shapes.
- Use version history. Name a version before you let a tool make big changes, so you can go back.
- Watch for the move to paid use once the beta ends.
Want your design files ready for the AI tools your team is starting to use? ALXO can tidy your design system and set up a safe way to work. Get in touch.
