Figma | Design Agent Lets Designers Create Reusable Skills

Figma is making reusable AI workflows easier to create directly inside the design environment. Published on August 13, 2026, the update lets designers work with the Figma agent to author custom skills from existing frames, instructions, guidelines, or repeatable processes, turning personal design judgment and team practices into workflows that can be invoked again when needed.


Figma design agent creating reusable skills for design and UI workflows

{getToc} $title={Table of Contents}

Figma skills turn repeatable design decisions into agent workflows


Skills are Markdown-based instruction files that give an AI agent a stable sequence of steps to follow. Instead of rewriting the same detailed prompt every time a designer needs to audit an interface, apply design-system rules, document a component, or review accessibility, that process can be packaged into a reusable skill.


Designers can now ask the Figma agent to create one directly. A frame can be used as reference, existing guidelines can become the basis of the instructions, or a workflow developed during a conversation can be converted into a skill. After the agent drafts it, the instructions can be previewed, tested, revised, or edited manually in Markdown.



Community examples show how specific a skill can become


Figma highlighted ten community-created examples to show that skills do not need to be generic automation commands. One skill analyzes an interface for typography, layout, accessibility, motion, and product-writing issues, while another documents a selected component with previews, variants, states, anatomy, and handoff information directly on the canvas.


Other examples focus on matching visible colors to design tokens, auditing components for duplicates and naming problems, expanding screens into loading, empty, error, disabled, and edge-case states, and performing structured design critiques based on usability, accessibility, hierarchy, and interaction quality.


Skills can capture design-system and handoff knowledge


The strongest use case for web and UI teams is consistency. A skill can package the standards a designer normally carries from project to project, such as how components should be documented, which states must exist before handoff, how tokens should be applied, or which accessibility and UX checks should be completed during review.


Once created, skills remain available across files and can be invoked with slash commands. They can also be published to a team or organization, shared through the Figma Community, exported as Markdown, or adapted from skills created by other designers.


Figma MCP can extend some skills beyond the canvas


Many skills can also work with Figma's MCP server in compatible third-party agents. This creates another path for teams that want the same design guidance available while working in development or AI-assisted coding environments rather than only inside the Figma canvas.


Skills can also be combined with connectors to bring additional context into a workflow. For example, a skill could reference design-system information from one connected source and product requirements from another before applying those rules to the current design.


IMPORTANT: Figma notes that skills improve consistency but do not make agent output deterministic. Results can still vary between runs, so design reviews, accessibility checks, and production handoff should remain subject to human validation.{alertWarning}

Daisuki's Take: What This Means for Web Designers


The interesting part of skills is that they give designers a way to preserve judgment, not just automate clicks. A good interface review often depends on dozens of small decisions that experienced designers apply almost automatically, and those decisions can now be documented as something an agent can repeatedly reference.


For web design teams, that can be especially useful around design systems, component states, accessibility, handoff, and repetitive audits. Instead of asking every designer to remember the same checklist, a shared skill can make that knowledge easier to reuse and refine over time.


We would treat skills as living design documentation rather than fixed automation. Their value increases when teams continue testing and updating the instructions as standards change, while keeping human judgment involved whenever the agent's interpretation affects the final interface.



Sources and Recommended Links