Add the godot_mcp addon to your C# Godot project and drive the editor with AI.
Godot 4.3–4.5C# / .NET 8 (mono)
01
Install the plugin
Two ways to get the plugin into your project — pick whichever fits your workflow.
>
Recommended
CLI
Install via the godot-cli tool on npm. The addon files and the two NuGet pins must already be present — install-plugin only flips the enable flag in project.godot.
1Install the CLI
Install godot-cli globally from npm.
npm install -g godot-cli
2Enable the plugin
Flip the project.godot enable flag for the godot_mcp addon.
godot-cli install-plugin ./MyGodotProject
3Wire up your AI client
Generate the MCP config for your client (e.g. Claude Code).
godot-cli setup-mcp claude-code ./MyGodotProject
4Open & wait
Open the project and wait until the editor is ready.
godot-cli open ./MyGodotProject
After opening you can also gate scripts on editor-readiness: godot-cli wait-for-ready ./MyGodotProject.
OR
◆
Asset Library
Install the godot_mcp addon from the Godot Asset Library (or the same zip from GitHub Releases), then add the two NuGet packages and enable the plugin.
Get godot_mcp from the Godot AssetLib tab, or download the zip from the GitHub Releases page, and place it in your project's addons/ folder.
2Add the NuGet packages
Add both packages to your project .csproj: com.IvanMurzak.ReflectorNet version 5.3.1 and com.IvanMurzak.McpPlugin version 6.9.1.
3Enable the plugin
In the editor: Project → Project Settings → Plugins, then enable godot_mcp.
4Open & build
Reopen the project so Godot builds the C# solution with the new packages.
02
Connect
Run through the hosted cloud endpoint, or point at your own server.
Cloud
Default
Connect through ai-game.dev using the device-code OAuth flow — no tokens to copy by hand. From the AI Game Developer window choose Cloud, click Authorize, and approve the device code in your browser. Works across firewalls and gives you usage analytics. This is the default and what we recommend for most users.
Custom
Prefer to run everything locally? Point the plugin at your own gamedev-mcp-server instance instead of the hosted endpoint. Your AI client then talks straight to your machine. Client-by-client config snippets live on the MCP endpoint page.