Figma is making it easier for AI models to transform design ideas into code by launching its new Dev Mode Model Context Protocol (MCP) server. This innovative tool offers large language models (LLMs) and coding assistants like Copilot in VS Code and Claude Code a clearer view of a design’s inner workings. In beta for Dev and Full seat users, the MCP server gives these tools precise design information—everything from detailed graph figures to the exact shades of colour—so you aren’t left to guess based on images alone.
Imagine trying to recreate a painting without knowing the artist’s technique. That’s the difference here—by understanding both the image and the methods behind a design, AI can code more accurately to match the original vision. Figma’s approach not only streamlines the design-to-code process but also promises to reduce the overall need for LLM resources.
Looking ahead, Figma plans to roll out further updates for the MCP server, including remote access and tighter integration with evolving codebases. This new server builds on the momentum from the recent full release of Figma Make, the prompt-to-code platform now available for all Full seat users. Plus, starting June 12th, the Figma Sites Code Layers feature will let AI tools convert static designs into interactive website experiences.
