Skip to content
News Figma Sep 2026

Figma: how generative plugins and shaders were built

Figma published a behind-the-build post on September 1, 2026, explaining how the generative plugins and shaders feature announced at Config 2026 was actually constructed. The piece covers technical decisions, the team structure, and the publishing model that governs who can distribute these new capabilities.

The most consequential infrastructure change is the move from WebGL2 to a WebGPU rendering pipeline. WebGPU gives the shaders access to compute shaders and more direct GPU control, which is what makes real-time generative and interactive effects possible inside the canvas without a separate rendering step. The post goes into how this transition was handled without breaking existing plugin behavior.

On the distribution side, Figma introduced two publishing tiers. Community publishing lets anyone create and share generative plugins and shaders publicly, and other users can remix them — similar to how community plugins have worked, now extended to the shader layer. Organization publishing is a private tier available on Enterprise plans, where teams can distribute capabilities internally without making them public. Both tiers expose the underlying code for download, so designers with programming backgrounds can inspect and modify what they’re running.

The post also covers an updated MCP server, which third-party agents can use to call generative capabilities from outside Figma itself. This is the infrastructure path for tools like Claude Code, Cursor, or custom workflows to interact with Figma’s generative layer programmatically.

The team behind the feature included Rogie King on design, Yassir Solomah on the rendering engineering side, and Georgia Rust as product manager. For designers interested in how AI-adjacent infrastructure decisions get made at a major design platform, the post is more technically specific than most product launch coverage.