Figma | Design Agent Adds Custom Tools, Context, and Skills
Figma is expanding its design agent with custom tools, stronger project context, and reusable skills for creative teams. The update makes the agent more useful directly on the canvas, where designers can create plugins, build shader effects, reference team files, and turn repeated prompts into shared design workflows.
Figma Design Agent adds more control for creative work
Figma Design Agent is becoming more practical for designers who want AI help without leaving the canvas. The update focuses on three areas: custom tools, richer context, and reusable skills. Together, these features make the agent less like a simple prompt box and more like a workspace assistant that can understand the project, build tools, and support repeatable design decisions.
For designers, the important shift is control. Instead of only asking the agent to create a static result, teams can use it to build reusable plugins, generate visual effects, reference existing Figma files, bring in outside context, and save useful prompts as commands that can be shared across a team.
Custom tools let designers build directly on the canvas
One of the main additions is the ability to prompt the design agent to build reusable plugins. This matters because many designers have workflow ideas that would normally require development support, a plugin setup, or switching between different tools. With generative plugins, the agent can help create tools that live directly on the Figma canvas.
These plugins use PropsKit, so they are designed to feel native inside Figma. They are useful for canvas-based tasks, layout helpers, asset organization, data visualization, and other repeatable design operations. For anything that needs to reach outside Figma Design, such as AI services or third-party APIs, Figma still points designers toward classic plugins.
Shaders bring more visual experimentation into Figma
Figma is also giving designers more room to experiment with shader effects and shader fills. Shaders are small customizable programs powered by WebGPU that define how pixels render, which makes them useful for visual effects such as dither, liquid metal, fractal noise, lens distortion, halftone looks, dynamic fills, and layered image treatments.
This is especially relevant for creators working on posters, campaign visuals, social assets, experimental layouts, and image-driven compositions. Shader effects can be stacked and adjusted with properties, while shader fills can go beyond static colors and gradients. The result is a more flexible visual playground inside Figma instead of a workflow that always depends on external editing tools.
Better context helps the agent understand the actual project
The update also gives the design agent more context to work with. Designers can attach files directly in the agent chat, reference other Figma files, search the web, and connect external tools through MCP connectors. This can help the agent understand project materials, design systems, content requirements, team context, and real data before generating something on the canvas.
For creative teams, context is what makes the difference between a generic AI result and something that fits the project. Referencing components, tokens, layouts, styles, UX copy, reports, or connected tools can help the agent produce work that follows an existing direction instead of starting from a blank prompt each time.
Skills turn repeated prompts into reusable commands
Figma is also introducing custom skills for the design agent. Skills turn repeatable prompts into reusable slash commands, which can be used for tasks such as applying a visual style, reviewing designs, improving accessibility, or automating a process the team already uses often.
This is useful because good prompts often contain a team's creative direction, design philosophy, review process, or production habits. When those prompts become shared skills, they can help a team work more consistently without forcing every designer to rewrite the same instructions from memory.
IMPORTANT: Figma Design Agent is available in open beta for Full seat users on Professional, Organization, and Enterprise plans. Collab, Dev, and View seats can use the agent in their drafts. The agent is free during beta, but prompting the agent to build plugins or shaders will cost AI credits once it becomes generally available.{alertWarning}
Daisuki's Take: What This Means for Designers
This update is important because it moves AI closer to the practical parts of design work. The value is not only that Figma can generate something from a prompt. The stronger idea is that designers can create their own tools, visual effects, and reusable commands inside the same space where they already build interfaces, assets, and layouts.
For teams, the context features may be just as important as the custom tools. A design agent that can reference files, understand design systems, use connected tools, and work from real project material has a better chance of producing something useful instead of something that only looks polished at first glance.
We would treat this as a workflow feature, not only an AI feature. The biggest benefit will come when designers use the agent to reduce repeated setup, test visual directions faster, and preserve team thinking through skills. The main thing to watch is access, AI credit usage after beta, and how well these tools behave in real production files.
Sources and Recommended Links
- Figma's design agent, now with custom tools and greater context | Figma Official Blog
