The idea is:
Run a plugin in Figma
My use case:
the entire sue case is running a plgin in figma and getting the output so i can work on it, give it to ai agetn and output it as code changes to github
Run a plugin in Figma
the entire sue case is running a plgin in figma and getting the output so i can work on it, give it to ai agetn and output it as code changes to github
Hi! This is a very interesting use case. However, there is a fundamental technical constraint in Figma’s architecture you should be aware of:
The “Pull” Problem: Figma does not allow external triggers to “start” a plugin. The Figma REST API is great for reading file data, but it cannot execute plugin code on the canvas. Plugins run in a sandboxed browser environment and require an active user session (or at least the file being open in a browser/app with the plugin running).
The Solution: The “Push” Workflow: Since you cannot call a plugin from the outside, the workflow must start inside Figma. You need a bridge that “pushes” the design data out to your AI agent.
How to achieve this currently: I developed a plugin called MKitFlow exactly for these types of automation gaps. Here is how you could build your pipeline:
The Bridge: The plugin gathers the required layer data/JSON and sends it via a Webhook to your automation platform (e.g., n8n, Make, or a custom Node.js server).
AI Processing: Your n8n workflow receives the Figma data, sends it to your AI agent (LLM), and receives the code output.