Figma rolled out an update on Wednesday that adds a code layer, native animation support, and an AI-powered plug-in builder to its collaborative canvas. The release pulls coding, motion design, and AI workflow tooling into a single surface, continuing a multi-year push to make Figma the place where designers, product managers, and engineers iterate together rather than hand specs across silos.
The headline addition is code layers. Teams can now clone repositories directly onto the canvas and extract flows from code into design layers for testing and exploration. Previously, moving between a working prototype in code and a manipulable design required a round trip through other tools. Code layers collapse that loop into the same multiplayer file where designers already work.
Figma laid the groundwork for this last year with Figma Make, a prompt-based prototyping tool, followed by integrations with Claude Code and Codex to smooth the design-to-implementation hand-off. The code layer is the next step in the same arc — treating code not as the final artifact at the end of a design process, but as a medium designers can poke at while ideas are still loose.
Key facts
- 01Figma's June 24 update adds code layers, native animations, 3D transforms, and shader fills directly to the collaborative canvas.
- 02Designers can now clone repositories and extract flows from code into design layers, removing a hand-off step with engineers.
- 03Figma integrates with [Claude](/claude) Code and [Codex](/openai) for the design-to-code hand-off, with custom AI plug-ins arriving in this release.
- 04Weavy, the node-based AI workflow tool Figma acquired last year, will be generable directly inside Figma later in the year.
- 05The AI assistant gains repeatable skills and connectors for Notion, Granola, Excel, and GitHub.
Yuhki Yamashita, Figma's chief product officer, said the point is not production-grade code but exploration.
The update also brings animations, transitions, and 3D transforms into Figma natively. Designers historically built motion in After Effects, Rive, or Lottie and converted the output into something the design tool could display. That conversion step is gone — animation now lives in the same file as the static design, with AI available to generate shader effects and fills on demand.
Figma is also tightening its embrace of Weavy, the node-based AI workflow tool it acquired last year. Weavy lets designers pipe inputs through different models and compare outputs side by side. In an update rolling out later in the year, users will be able to generate Weavy workflows directly within Figma, rather than treating the two products as separate apps.
The AI assistant gets a meaningful upgrade as well. Users can write text prompts to create repeatable skills that AI agents inside Figma can reuse, turning one-off operations into building blocks. The assistant can also connect to Notion, Granola, Excel, and GitHub, or accept attached files, giving it the project context it needs to act usefully rather than guess.
The final piece is a custom plug-in builder. Users can describe a tool — a layout generator, a vector path tracer, a naming-convention enforcer — in plain language and have Figma generate the plug-in. That moves plug-in creation from a developer task to something a designer can do mid-session, and it positions Figma's plug-in ecosystem to expand faster than the population of people willing to write TypeScript against an API.
The bet is not without friction. Code layers, native animations, and AI-generated plug-ins each introduce new failure modes — an extracted flow that doesn't quite match the repo, an animation curve that doesn't survive export, a generated plug-in that misreads the user's intent. Figma's pitch is that these features exist for exploration, where being approximately right is more valuable than being precisely wrong; the test is whether teams actually treat them that way or expect production-grade output from a tool that's built for sketching.
Figma's strategy here is clear: keep widening the surface area of what happens inside a Figma file so that fewer ideas need to leave it. Each addition — Make, Weavy, Claude Code and Codex hooks, now code layers and the plug-in builder — chips away at the boundary between designing a product and building one. For Figma, that boundary is where customers churn to other tools and where the next generation of AI-native design startups would otherwise carve out a wedge. Collapsing it is both a defensive move against AI-first competitors and an offensive bid to make Figma the default canvas for software made with models in the loop.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
