Figma opens Community publishing for Config-built plugins and shaders

• Generative plugins and shaders made with Figma’s design agent now have a public Community publishing route, and Figma's announcement says Organization and Enterprise customers can also share them privately.
• Figma’s design agent can now generate animated, interactive effects on the design canvas, according to the feature announcement.
• Generated code can leave Figma: one reply says creators own their shaders, while Figma’s build diary describes editing their source in a third-party agent.
• January’s FigGPU prototype, documented in the FigGPU videos, tested composable effects, tangible controls, and LLM-suggested parameters before becoming part of the product work.

Figma’s build diary starts with a WebGPU sandbox that put effects on vectors, 3D layers, and particle systems. Its Config documentation describes MCP export as raw JS/WGSL, while the quick-start guide says agent-generated tools are free during beta.

Community profiles can belong to an individual, team, or organization. Figma’s permissions guide says a file’s location, access settings, and plan tier determine publishing eligibility, and people with edit access can publish to a personal profile.

Figma’s follow-up post directs readers to the build diary. Figma product designer Rogie King called the rollout “shader city” in one reply.

Figma defines shaders as GPU programs that manipulate pixels in real time, applied as fills or effects. A cursor drags and adjusts a liquid-like effect directly on the canvas in the canvas demo.

Figma lays out four handoffs in the feature announcement:

1. View the code behind a generated plugin or shader.
2. Download that code for editing in a third-party agent.
3. Use the Figma MCP server to read and write generative tools with an agent.
4. Export a shader to an agent that implements it in React.

Figma’s Config documentation says the MCP export exposes JS/WGSL as “crude/raw” code. The same document says shader effects do not yet appear in Dev Mode.

King’s January FigGPU video post showed a vibe-coded environment where typed code immediately changed live canvas graphics. His build diary identifies three design targets:

• Composability: one shader’s output can feed the next.
• Editor ergonomics: generated controls let a designer manipulate the result beyond a prompt.
• Agentic creation: LLMs can propose parameters worth exposing.

The team found working prototypes useful for testing visual assumptions, then used static Figma mocks to map all states and align stakeholders before returning to code prototypes. King also wrote production code and submitted pull requests in Figma’s codebase.

Figma’s quick-start guide calls generated plugins and shaders an open beta, warns that features and performance may change, and requires WebGPU-capable systems and browsers for shaders.

• Full-seat users can access the agent across plans. Collab, Dev, and View seats can use it in drafts, according to the build diary.
• Figma says the agent is free during beta, then will begin consuming AI credits in a few weeks.
• King asked for feedback on a possible performance UI, including FPS settings or labels from “fast” to “laptop melt,” in rogie's performance question.