Figma expands MCP support despite co-founder criticism

Figma is expanding its support for the Model Context Protocol (MCP), the open standard that allows AI agents to connect to external tools, even as its co-founder Dylan Field publicly questions the protocol's design. According to TechCrunch, the company announced 20 new MCP partners at its first Config conference in India, signaling a commitment to the ecosystem despite internal reservations about its current state.

The new partners include major names in the developer and productivity space, such as Amp, ClickUp, Dia, Junie by JetBrains, Linear, Meta Muse, Mistral, Notion, OpenCode, and Sarvam Agents. This expansion marks a significant step for Figma, which has historically been cautious about exposing its design data to third-party AI agents. The move positions Figma as a central node in the growing MCP network, allowing users to integrate design workflows with a broader range of AI-powered applications.

The criticism from the top

Field’s hesitation is not new. Days before the announcement, Field engaged in a public debate on X regarding MCP’s limitations. The friction stemmed from a complaint by one of MCP’s co-creators about Figma’s lack of support for the open-source tool Pi. In response, Field pointed out an asymmetry in the ecosystem: Anthropic, the creator of MCP, does not expose its own cloud resources or artifacts via the protocol.

“I find it interesting that Anthropic (creators of MCP standard) do not expose artifacts/other cloud resources created via MCP,” Field wrote on X. “The debate is framed as open vs closed. Everyone always prefers open … obviously. And meanwhile issues of asymmetric context sharing are ignored.”

Field later clarified that Figma does plan to support Pi, but his comments highlighted a broader tension between the protocol’s open-source ideals and the practical realities of corporate data security.

Security and the frontier lab advantage

At a media briefing, Field elaborated on why Figma is not fully open with its MCP implementation. He argued that MCP is “not a perfect protocol” and has “tons of room for improvement.” A key concern, according to Field, is that the current standard is “advantaged more to frontier labs.”

The primary driver for Figma’s cautious approach is security. Field explained that the more data the company exposes through MCP, the higher the chance of abuse. While most users intend to use MCP and Figma data in benign ways, the company must account for malicious actors who could exploit broad data access. This risk-benefit analysis has led Figma to adopt a selective approach to MCP support, prioritizing security over total openness.

Context in the AI tooling landscape

This announcement fits into a broader trend of design and productivity tools integrating with AI agents. The Model Context Protocol, initially developed by Anthropic, has become a de facto standard for connecting LLMs to external data sources. However, the ecosystem is still maturing, with significant gaps in functionality and security. Figma’s decision to expand its MCP support while criticizing the protocol reflects a common stance among enterprise software vendors: they want the benefits of AI integration but are wary of the security and control implications of a fully open standard.

Competitors in the design and developer tool space are also exploring MCP, but Figma’s scale makes its adoption particularly significant. By partnering with companies like Notion, Linear, and Mistral, Figma is creating a network effect that could make MCP the default way for AI agents to interact with design data. However, the internal debate at Figma suggests that the protocol may need significant evolution to meet the security requirements of large-scale enterprise applications.

What it means for developers

For developers, the expansion of Figma’s MCP support means more opportunities to build AI agents that can interact with design files. The new partners, including Junie by JetBrains and OpenCode, suggest that MCP is becoming a viable option for integrating AI into professional development workflows. However, developers should be aware of the security implications. Figma’s caution highlights the risks of exposing sensitive design data to third-party AI models.

Developers using MCP with Figma should ensure that their agents are properly scoped and that they understand the data access permissions. The debate over asymmetric context sharing, as highlighted by Field, suggests that not all data sources are equally accessible via MCP. This could lead to inconsistencies in how AI agents interact with different tools. Developers should monitor Figma’s documentation for updates on MCP support and be prepared to adapt their workflows as the protocol evolves.

What to watch

1. Figma’s support for Pi: Field stated that Figma plans to support Pi, the open-source tool that sparked the initial debate. The timeline and scope of this support will be a key indicator of Figma’s commitment to the open-source MCP ecosystem.
2. MCP protocol updates: The criticism that MCP is “advantaged more to frontier labs” suggests that the protocol may undergo changes to better support enterprise use cases. Developers should watch for updates from Anthropic and the MCP community.
3. Security incidents: As more companies adopt MCP, the risk of data breaches increases. Any security incidents involving MCP-connected tools will likely influence how companies like Figma approach their MCP support.
4. Competitor adoption: The actions of other design and productivity companies will shape the future of MCP. If competitors adopt a more open approach, it could pressure Figma to expand its support further.