Figma Widens MCP Support While Its Co-Founder Questions the Protocol's Design

Figma is betting on an open standard its own co-founder calls imperfect. That contradiction, rather than any single product announcement, is the real story out of the company's first Config conference in India.

At the event, Figma unveiled 20 new partners supporting MCP (model context protocol), the open source standard that lets AI agents connect to external data and tools. The list includes Amp, ClickUp, Dia, Junie by JetBrains, Linear, Meta Muse, Mistral, Notion, OpenCode, and Sarvam Agents, among others. On paper, it is a significant expansion of Figma's AI ecosystem and a signal that the company wants its design data accessible to a broad range of agentic tools, not just one or two favored partners.

But co-founder Dylan Field has been airing doubts about MCP for weeks, and those doubts came to a head just before the announcement. One of MCP's original creators, David Soria Parra, publicly criticized Figma for not supporting the open source tool Pi. Field responded on X with a pointed observation: Anthropic, the company that created the MCP standard, does not expose its own artifacts or cloud resources through MCP.

"I find it interesting that Anthropic (creators of MCP standard) do not expose artifacts/other cloud resources created via MCP," Field wrote. "The debate is framed as open vs closed. Everyone always prefers open... obviously. And meanwhile issues of asymmetric context sharing are ignored." He later said Figma plans to support Pi after all, but the underlying critique stood unretracted.

The asymmetry argument

Field's core complaint is not that MCP fails technically. It is that the protocol, in practice, favors the frontier labs that helped build it over the platforms expected to plug into it.

Speaking at a media briefing, Field elaborated on the tension. "I'm really excited about what we're doing in terms of opening up our ecosystem. I also think that MCP is not a perfect protocol. And I think that it is something that has tons of room for improvement," he said. "It is advantaged more to frontier labs. Also, I understand everyone online who just wants a protocol that they can use when they want to use their Figma data, and I want them to use it, too."

That is a notably candid position for a company simultaneously expanding its commitment to the standard it is criticizing. Field's logic seems to be that adoption and critique are not mutually exclusive. Figma can support MCP broadly while pushing, publicly, for the protocol to evolve in ways that do not simply funnel proprietary data toward the handful of labs that set the rules.

The abuse-risk argument adds another layer. Field said the company's reluctance to go fully open isn't about control for its own sake. The more design data Figma exposes through MCP, the greater the surface area for misuse. He was careful to note that most users have no malicious intent, but platforms handling sensitive design and product data have to plan for the exceptions, not just the norm. That is a standard risk-management posture, familiar to anyone who has watched API providers wrestle with rate limits, scopes, and permissioning as their ecosystems grow.

The broader context matters here too. MCP has become something close to a default standard in under two years, with adopters ranging from coding agents to productivity suites. Developer Mario Zechner's observation, cited in the same discussion thread, that MCP is "better compared to a year ago, but also worse" captures a sentiment shared by many builders in the space: the protocol has matured in adoption while accumulating new friction points around security, data exposure, and inconsistent implementation quality across vendors.

Figma's position inside that landscape is unusual. It holds a trove of structured design data that AI agents want to act on: components, variables, prototypes, comment threads. That makes Figma a desirable MCP server for coding assistants and design tools alike. It also makes Figma a prime target for the abuse scenarios Field is worried about. The company has leverage most MCP implementers lack, and it appears to be using that leverage to negotiate terms rather than simply ship an open firehose.

The Bottom Line

The 20-partner expansion is a meaningful near-term win for Figma's AI integration strategy, broadening distribution of its data into tools developers already use daily, from Linear to Notion to Mistral's agents. That is good for engagement and stickiness, the kind of metric that matters for a public company managing investor expectations around AI relevance.

The more consequential signal, though, is Field's willingness to criticize MCP's governance even as Figma deepens its reliance on it. For investors and partners watching Figma's AI strategy, the message is twofold: expect continued, broad support for open standards, but do not expect Figma to cede full data access without a fight over terms. The company is positioning itself as both participant and critic, a stance that carries some reputational risk if partners read the criticism as foot-dragging, but one that also protects Figma's core data asset from becoming commoditized on someone else's terms. Watch how Anthropic and other frontier labs respond to the asymmetry complaint; if MCP's governance shifts toward more balanced data-sharing norms, Figma's early advocacy could prove to be a smart positioning bet rather than a sideline gripe.