Loading page…
The coordinated Config updates connect written design guidance, interactive prototypes and outside tools. More predictable AI output is the goal—not a guaranteed result.
Listen to this story
Figma introduced Specs, a shared Markdown reference for design rules and supporting materials that its own canvas agent and third-party agents can use. The goal is more predictable output, not guaranteed adherence. At its first Config event in India, Figma also moved code layers from private to public beta, letting teams generate and iteratively edit code for interactive prototypes. The additions give teams ways to provide agents with design context and test code-based workflows, while integrations extend Figma Make to dynamic website content.
Specs can stand alone or sit on the canvas, linking to designs, FigJam diagrams and code layers alongside design libraries, code libraries and guidelines.
Code layers were announced in private beta in June; the public beta lets users ask the agent to generate code and revise parts of it.
Figma added support for more than 20 tools through MCP connections, including Open Code, Dia, Grok Bot and Sarvam.
Figma is giving teams a place to spell out the design rules its AI agents should follow. At its first Config event in India, the company introduced Figma Specs alongside public-beta code layers, more connections to outside tools and a content-management integration—pairing AI creation with guidance about what teams want to build.
The announcement follows a week in which Figma made its agent on the canvas available to more users. The company is now adding tools intended to steer that agent, and third-party agents, toward a brand’s design rules. That is a response to AI’s variable output: Figma’s stated aim is more predictable results, not a promise that every generated design will follow the guidance.
Specs supplies the written reference. It is a Markdown-based editor where users can define their design language for both colleagues and agents. As TechCrunch describes it, a Specs document can live as a standalone Figma file or sit within the canvas. Teams can therefore keep the instructions in their own document or alongside the visual work those instructions describe.
The document can point to elements on the canvas and embed FigJam diagrams, code layers and designs. That makes Specs more than a separate page of written instructions: it can include references to the actual material a team is discussing. Figma also says teams can bring in design libraries, code libraries and guidelines, giving agents a centralized source of context for the design system.
We see Figma as the place where software teams come to decide what they build and ship.
Loredana Crisan, Figma’s chief design officer, speaking to TechCrunch
The code-layer update changes who can test another part of that workflow. Figma first announced the feature in private beta in June. At the India event, it moved code layers into public beta, opening the feature to more users. It remains a beta release, rather than a finished general-availability launch.
Code layers add code to designs on the canvas so teams can create more interactive prototypes. Users can ask the agent to turn a design into code, then request changes to part of that code as they revise the prototype. The feature supports a process of building and adjusting a design, rather than only viewing a static canvas.
The same announcement extends Figma’s connections beyond its own agent. The company said it is supporting more than 20 new tools through its MCP connections, including Open Code, Dia, Grok Bot and Sarvam. Those additions widen the set of outside tools supported by the platform, alongside the new place for teams to record their design context.
Figma is also integrating Payload, its content-management system, with Figma Make. The integration is intended to let users generate dynamic websites and directly manage their content. It adds a content-management path to the creation product, distinct from the code-layer beta’s focus on interactive prototypes.
India is a substantial setting for these platform updates. Figma says the country is its largest user market outside the United States. It also says it works with 40% of India’s top 100 companies listed on the Bombay Stock Exchange, giving its first Config event there a sizable existing customer backdrop.
Story updates
Figma CEO Dylan Field is questioning the rules behind an AI connection standard his company is adopting more widely. In an October 20 TechCrunch report, Field said the Model Context Protocol favors frontier AI labs and needs improvement. He also defended limits on data access, citing abuse risks, as Figma added 20 MCP partners.
MCP standardizes how AI agents connect to outside data and tools. For Figma, that includes letting other services bring design context into their workflows. Field’s objections concern both the balance of that exchange and how much information Figma should expose through it.
The criticism followed a public exchange about openness. On September 30, Field asked why Anthropic, which created MCP, did not make various resources available through the protocol. TechCrunch also quoted his complaint that Anthropic did not expose artifacts and other cloud resources through MCP. That was Field’s characterization of Anthropic’s approach, not an independently established finding about its services.
His objection was more specific than opposition to an open standard. Field argued that a debate framed simply as open versus closed overlooks whether the participants share context on equal terms. He described that imbalance as “asymmetric context sharing.”
The debate is framed as open vs closed. Everyone always prefers open… obviously. And meanwhile issues of asymmetric context sharing are ignored.
At a media briefing, TechCrunch’s Ivan Mehta asked Field about that debate. Field said he was excited about opening Figma’s ecosystem, but called MCP “not a perfect protocol” with “tons of room for improvement.” He said it advantages frontier labs—the developers of leading AI models—without describing a specific protocol change in the remarks TechCrunch published.
He also acknowledged the user’s side of the argument: people want a protocol they can use to access their Figma data, and he wants them to use it. That support did not extend to making every resource available without restriction.
Field’s explanation for holding back was the potential for abuse. He said exposing more data through MCP increases that risk. Most users do not intend to use MCP or Figma data maliciously, he acknowledged, but Figma still has to account for those who might. His argument therefore places a limit on openness even when ordinary users have legitimate reasons to want broader access.
Those reservations have not stopped Figma from adding connections. TechCrunch reported that the company announced 20 new MCP partners at its first Config conference in India. Named participants included Amp, ClickUp, Dia, Junie by JetBrains, Linear, Meta Muse, Mistral, Notion, OpenCode and Sarvam Agents.
Figma’s own October 15 announcement confirms the Sarvam partnership and describes its purpose: allowing users to add Figma context to their workflow. It identifies Sarvam’s Indus as joining an ecosystem that already includes Claude Code, ChatGPT, Cursor, GitHub Copilot, Xcode and VS Code.
Sarvam CEO and co-founder Pratyush Kumar said combining his company’s AI work with Figma’s design context could help teams improve product quality, reduce rework and speed development. Those are the partnership’s stated goals, not measured results. For now, Figma is widening participation while Field argues that the underlying exchange still needs better rules.
figma.comLoading discussion...
Join the conversation
Explain where consistency should end and creative freedom should begin.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.