Model Context Protocol

What is MCP?

Model Context Protocol (MCP) is an open standard that allows AI models to use external tools and services through a unified interface. Developer Tools provides an MCP server that enables AI assistants to inspect and interact with your Avalonia application's visual tree, search for elements, and access diagnostic information directly.

Prerequisites

Before setting up the MCP server, ensure you have:

1. DevTools .NET tool installed - Follow the Getting Started guide
2. Valid license key - Set ACCELERATE_LICENSE_KEY environment variable to a valid license key from the portal

备注

DevTools MCP is only available with a full Accelerate license.

Supported AI Assistants

DevTools MCP server works with various AI coding assistants.

• VSCode with Copilot
• Visual Studio with Copilot
• Rider Copilot
• Rider AI Assistant
• Claude Code
• Cursor

Setting Up the MCP Server

DevTools provides a STDIO type MCP server, available by running the avdt mcp command. Installation methods depend on your editor and are detailed below.

Quick Install

Use these one-click installation links to automatically configure the MCP server:

• Install for VSCode
• Install for Cursor

CLI Install

Claude Code

Run this command:

claude mcp add --scope user avalonia_devtools -- avdt mcp

Manual Configuration (mcp.json)

For editors that support mcp.json configuration files (VSCode, Visual Studio, Rider, Cursor), add the following to your configuration:

.vscode/mcp.json

{
    "servers": {
        "avalonia_devtools": {
            "type": "stdio",
            "command": "avdt",
            "args": [
                "mcp"
            ]
        }
    }
}

提示

Refer to your editor's documentation for the exact location to place or edit the mcp.json file.

Supported Tools

attach-to-app: Connects to a running Avalonia app. Lists apps if no processId is found and multiple apps are available. attach-to-file: Connects to the live previewer with the specified XAML file. This is recommended over attach-to-app for previewing XAML.

tree: Gets child elements. Null nodeId returns roots. ancestors: Gets parent chain from root. search: Finds elements by type or x:Name.

props: Gets property values for a node. set-prop: Sets a property value. Use null or unset for special values. styles: Gets applied styles and setters for a node. pseudo-class: Sets pseudo-class. Omit pseudoClass to list available options.

resources: Gets resources. Optionally scoped to a node. assets: Lists embedded assets (images, fonts). Returns URLs for open-asset. open-asset: Downloads asset by URL from assets tool.

screenshot: Captures a PNG screenshot of a UI element. input: Sends an input event to a UI element. action: Performs an action on a UI element.

Usage Examples

Simply describe what you want to accomplish in natural language. The AI assistant will use the MCP server to inspect and interact with your running Avalonia application:

Inspecting UI:

"Connect to my running app and show me the visual tree structure."

Finding Elements:

"Find all Button elements in my application."

Debugging Styles:

"What styles are applied to the MainWindow?"

Taking Screenshots:

"Take a screenshot of the login panel."

The AI assistant will guide you, execute commands, and help you understand your application's UI structure.