Figma Expands MCP Support While Field Questions Its Limits

Figma announced 20 new MCP partners at its first Config conference in India, even as co-founder Dylan Field called the protocol imperfect. The debate centers on openness, unequal access to context, and the risk that exposing more Figma data could invite abuse.

WTF Index TERMINATOR
◄ Terminator 2 Idiocracy 1 ►

The story mildly leans toward Terminator because broader AI access to Figma data raises concerns about exposure and misuse.

Figma Expands MCP Support While Field Questions Its Limits

Figma is expanding support for the open source Model Context Protocol (MCP), even as co-founder Dylan Field argues that the standard has room to improve. The company’s latest announcement adds 20 partners, while its leadership weighs the benefits of opening Figma’s ecosystem against possible misuse of the data shared through it.

Figma adds 20 MCP partners

The design company announced the partners at its first Config conference in India. The list includes Amp, ClickUp, Dia, Junie by JetBrains, Linear, Meta Muse, Mistral, Notion, OpenCode, and Sarvam Agents.

MCP standardizes how AI agents connect to external data sources and tools. For Figma, supporting the standard creates a way for outside tools and agents to connect with its ecosystem. The new partners signal that the company is broadening that support, despite Field’s reservations about the protocol itself.

Field questions how open the protocol is

Field’s concerns came into focus after a debate over Figma’s support for Pi, an open source tool. One of MCP’s co-creators had complained about Figma not supporting it. In response, Field raised a separate question about Anthropic, which created MCP, and whether its own resources were available through the protocol.

On X, Field asked why Anthropic did not expose artifacts and other cloud resources through MCP. He also argued that debate over open versus closed systems can overlook “asymmetric context sharing.” That concern points to a broader question: whether organizations that build AI tools make the same information available to others that they can access themselves.

Field later said Figma plans support for Pi. At a media briefing, he reiterated that the company wants to open its ecosystem, but does not see MCP as a finished solution. He described it as a protocol with “tons of room for improvement” and said it is advantaged more to frontier labs.

Openness brings a data exposure trade-off

Field also acknowledged why users want a protocol that lets them work with their Figma data. He said he understands people who want to use such a protocol and wants them to use it, too. That interest sits alongside Figma’s responsibility to consider what can happen when more data becomes accessible.

Figma’s CEO said the company does not want to be fully open because exposing more data through MCP increases the chance of abuse. The concern is not that most users intend harm: he clarified that most do not. Rather, the company still has to account for potential risks as it decides what information to make available.

That tension shapes Figma’s approach. Broader MCP support can make the ecosystem more useful to people who want to connect AI agents and tools with Figma, while each additional opening also raises questions about how exposed data might be used. The company’s partner announcement shows continued expansion; Field’s comments make clear that expansion does not mean he considers the protocol or its risks settled.

A debate still taking shape

The discussion combines two distinct issues: whether MCP gives different organizations fair access to context, and how much data a company should expose to outside tools. Field’s remarks touch on both. His criticism of the protocol focuses on its structure and who it advantages, while the CEO’s comments focus on potential abuse of Figma data.

Figma is therefore moving forward with MCP while continuing to question how it works and where openness should stop. The 20 new partners mark a concrete expansion, but the company’s position suggests that support for an open standard can coexist with caution about the standard’s design and the data shared through it.