Figma Adds Code Layers, Native Animations, and AI Plugins

Figma has spent years as the place where design work gets done before engineers take over. With a new update reported by TechCrunch, the company is pushing further into the space between those two groups, adding features that let design teams pull live repository code into the canvas, build animations natively, and generate custom plugins through plain-language prompts.

The headline addition is Code Layers, which lets teams clone repositories and extract code flows into design layers for testing. Chief Product Officer Yuhki Yamashita was direct about what this is for and what it is not: "this is an environment where you don't really care about the quality of the code." That framing matters; this is a prototyping and exploration tool, not a path to shipping production code from inside Figma.

The update also brings native animations, transitions, and 3D transforms, removing the previous dependency on external software and manual code conversion steps. On the AI side, users can now generate shader effects and fills through text prompts, and create repeatable automations through natural language that connect to external tools including Notion, Granola, Excel, and GitHub.

This builds on moves Figma has been making for a while. The company acquired Weavy, described as an AI-powered media generation tool, in 2025, and announced integrations with Claude Code and Codex in February 2026. The direction is consistent: Figma is positioning itself as the hub where design and development workflows intersect, not just a handoff layer.

The honest caveat is that Yamashita's own words underscore the gap between what Code Layers does and what a real engineering workflow requires: code quality is explicitly not the point. What the reporting doesn't give you is how these features are priced or whether they're gated to specific plans. Who benefits most near-term are design teams currently context-switching between Figma, separate animation tools, and version control systems; consolidating that into one canvas is a real efficiency gain if the native implementations hold up.