Figma now lets users give instructions to its AI agent directly from dev mode, the view developers use to inspect designs. The change lets developers prompt the Figma agent without leaving that view.
Figma now lets you generate several images from one prompt when using Make image or Edit image. You can have one model make two, four, or six images, or run the prompt across up to three models at once and compare the results side by side.
One model can make two, four, or six images from a single prompt, so you need to retype prompts less often.
You can run one prompt on up to three image models at once and see each model's version side by side.
Figma admins can now keep a list of approved OAuth apps, including MCP clients such as Cursor and Claude Code, to control which apps their teams can use. They can also review app requests and manage approved apps from a new Extensions tab in the admin console. The feature is rolling out to Organization and Enterprise plans over the coming days.
Admins can create an allowlist of approved OAuth apps.
The allowlist covers MCP clients like Cursor and Claude Code.
A new Extensions tab in the admin console handles app requests and approved apps.
The feature is rolling out to Organization and Enterprise plans over the coming days.
Figma, the design tool, now has a 3D transform effect. You can rotate objects along the x, y, and z axes to build 3D designs. You can copy the CSS or send the design straight to a coding agent.
You rotate objects with on-canvas handles and see live previews as you turn them.
You can build 3D designs from scratch or ask the Figma agent to make them.
You adjust perspective and depth to build 3D scenes and animate flips and turns with Figma Motion.
You copy CSS from Dev Mode or send designs to a coding agent through the Figma MCP.
In Select area, Figma's AI image editing tool, you can now click an object in a picture and Figma selects it for you. There is no need to trace around it by hand.
Select area is an AI image editing tool for picking part of an image to isolate or erase it.
One click on an object now makes Figma select that object automatically.
When Figma users build a code layer, which turns design frames into working code, the agent now reads the prototype connections and interactions between screens. That means less explaining of the tricky parts of a flow. Code layers are currently an open beta.
When a code layer is created from design frames, the agent also reads the links and interactions between them.
Less time goes into describing complex flows in prompts, leaving more for polishing the prototype and finding edge cases.
Newly connected GitHub repositories in Figma will have cloning, which copies in a repository's code, switched on by default. Teams can bring code into Figma without a separate admin approval. The change is rolling out to all plans.
Cloning is switched on by default for GitHub repositories connected from now on.
Teams can import code into Figma without waiting for an extra admin approval.
Admins can still switch cloning off repository by repository, and repositories already connected keep their current settings.
Figma shaders, graphics code that creates visual effects, can now be rendered and exported at up to 4096px. They can also be carried along when designs are turned into code layers. Shaders have now fully launched for every user.
Shaders can be rendered and exported at up to 4096px to keep their look and fine detail.
Shaders can be included when the build-with-code feature turns a design into a code layer.
To see shaders correctly in a code layer, switch the layer to Play using the control at its top right.
This works only in the browser on Chrome 157 and is rolling out this week.
Users can now link Google Drive to the Figma agent and Make, pulling existing file context into Figma and creating new Drive files without leaving their work. Connectors, which hook up outside services, now work on every Figma plan, including Starter.
With the connector, the agent can sum up research, draw on real data and bring Drive assets onto the canvas to edit.
New Google Drive files can be created without leaving the Figma file you are working in.
Every Figma plan, Starter included, can now use connectors.
Connectors such as Google Drive are added through the plus menu in the prompt box while working with the agent.
Figma users can now publish Weave tools they build for use inside their own organization. These internal tools appear in the Tools tab next to Weave tools made by Figma and the Figma Community. The option is limited to Organization and Enterprise plans.
People can share Weave tools they built with others in the same Figma organization.
Internally shared tools sit in the Tools tab alongside Weave tools from Figma and the Community.
Only Organization and Enterprise plan customers can publish tools internally.
In Figma, every edit to a generative plugin or a shader, graphics code that creates visual effects, now saves a new version. Earlier versions can be viewed again or restored, and an AI agent writes a description for each one to build a change log. The update is rolling out this week.
Changing a generative plugin or shader creates a new version and keeps the earlier ones.
Stored earlier versions can be reopened or restored at any time.
An agent-written note on each version builds a change log and marks which version is live.
Every video and image model offered in Figma Weave can now be used from outside AI agents through the Figma MCP, a way of linking AI agents to other tools. With a Weave account tied to a Figma account, users can make videos and images with models such as Seedance, Kling and Nano Banana while chatting with their agent.
All of Figma Weave's video and image models can be called from an external AI agent via the Figma MCP.
Using them requires linking a Figma Weave account to your Figma account.
While chatting with an agent, you can create videos and images with models like Seedance, Kling and Nano Banana.
Generating with Weave models draws on Weave credits.
Figma added tools that help Enterprise plan admins manage paid AI credits and automate team operations. They include AI credit defaults, credit management through SCIM, a standard for managing company accounts automatically, limits on outside plan creation and admin APIs. The update is rolling out to Enterprise plans over the next few days.
Admins can give users a default level of paid AI credit access for the whole organization or for each billing group.
For SCIM-managed users, AI credit access and limits can be set from the company's identity system.
Admins can stop staff from creating separate Starter or Professional plans with company email addresses.
New admin REST APIs let companies automate the setup and upkeep of teams, folders and files.