Figma co-founder Dylan Field thinks MCP is not a perfect protocol

Figma's cautious embrace of MCP means designers will not yet have seamless access to all their Figma data through AI agents.

Figma co-founder Dylan Field has voiced concerns about the Model Context Protocol (MCP), an open-source standard for AI agents connecting to external data and tools. Despite these reservations, Figma announced 20 new MCP partners at its Config India conference, expanding its support for the protocol. Field previously questioned why MCP's creators, Anthropic, do not expose all their resources via MCP, citing issues of asymmetric context sharing. He elaborated that while Figma aims for openness, MCP has significant room for improvement and is currently more advantageous to "frontier labs." Field also cited the risk of abuse as a reason for not exposing all of Figma's data through MCP, even though most users have no malicious intent.

Field's nuanced stance on MCP highlights a critical tension in the AI ecosystem: the drive for open standards versus the imperative of data security and preventing misuse. By expanding MCP partnerships while simultaneously expressing reservations, Figma signals its intention to participate in the evolving landscape of AI agent interoperability without fully committing to a protocol it perceives as potentially flawed or unbalanced.

This cautious approach suggests that while the industry is moving towards more standardized AI agent interactions, companies like Figma are also prioritizing their own risk mitigation strategies. Users seeking full integration of their design data with AI tools may need to await further protocol development or alternative solutions that address Figma's stated concerns about asymmetric context sharing and potential for abuse.