tl;dr: Penpot launches its open beta for the MCP server. Take part as a tester by writing to support@penpot.app with the subject “MCP test volunteer”
Something new is coming to Penpot, and we would love the community to help us test it.
Introducing Penpot MCP server
Penpot MCP server connects Penpot with external tools, workflows, and AI agents in a new way. Unlike typical design to code solutions, Penpot’s design expressed as code approach lets AI agents read and work directly with file data allowing multi-directional workflows; turning the canvas into a programmable space for automation and experimentation.
Teams can connect any AI agent or client, integrate Penpot into existing workflows, and build custom processes through the open API. The MCP server runs in your own environment, helping keep data secure while avoiding vendor lock in.
Fully open and free, Penpot MCP gives teams complete control over their models, infrastructure, and integrations. It will be able to be use locally or remotely, we want to know about your experience.
Here are just a few examples of the multi-directional workflows that you can do with MCP:
Design to code
Design to Design
Code to Design
And many more capabilities that you can discover in the full playlist.
Objective of the test
We are looking forward to not only validating MCP in everyday workflows but see how users can push it to its limits. We want to see how MCP behaves under complex, unconventional, and demanding scenarios that challenge what design tooling can normally do, especially when tackling sophisticated UX and UI challenges.
Furthermore, we invite testers to experiment freely, connect MCP, locally or remotely, to unexpected workflows, automate at scale, and explore edge cases that may reveal both its strengths and its breaking points. The goal is to stress the MCP server in real conditions, uncover friction early, and learn where performance, usability, and documentation must evolve.
How Penpot MCP server works
For this test, you will use an MCP compatible AI client to interact with your design or code using natural language. The Penpot MCP Server will act as an open translation layer, enabling external AI agents and tools to communicate with the Penpot API and execute actions in your workspace.
Penpot does not include proprietary AI models or built-in AI services. Instead, you will be able to connect third party AI providers or agents such as Claude, Cursor, or any other LLM based tool you prefer.
How to join the Penpot MCP open beta test
If you want to be part of the beta, send an email to support@penpot.app with the subject “MCP test volunteer”. In the message, tell us a bit about your role and how you usually use Penpot.
Thank you for trusting us and for building Penpot together with us 
The Penpot Product team
