I wanted to explore how WebMCP could be used to make websites easier for AI agents to interact with. While working with WebMCP, I felt there was a missing developer tool for seeing what tools a webpage exposes and what happens when those tools are called.
So I built gpuiXwebmcp, a native debugger for WebMCP using Rust and GPUI. It connects to a WebMCP-enabled Chrome tab and lets you see the available tools, inspect their schemas, execute them, and view the execution results.
The main idea is simple: instead of guessing what an agent is doing inside a browser, developers should be able to see the tools and their execution clearly.
I built the project to learn more about WebMCP, GPUI, browser communication, and building native developer tools with Rust. One of the main challenges was connecting the browser environment with a native GPUI application and keeping the tool and execution state synchronized.
This project is still an early experiment. The next things I want to explore are agent execution, tool-call replay, better debugging information, and connecting local AI models to the WebMCP tools.
Built With
- chrome
- gpui
- javascript
- json
- rust
- webmcp
- websockets
Log in or sign up for Devpost to join the conversation.