Godot MCP server

This is a small helper that connects your AI assistant to Godot, the free program people use to make games. Once it is set up, your AI can open your game project, run it, read the error messages, and even add things to your scenes. It is handy for hobbyist game makers and anyone learning Godot who wants a second pair of hands.

What this MCP server does

You ask your AI something like "run my game and tell me what went wrong". The AI sends that request to this helper program, which starts Godot in the background. The helper watches the output, collects any error messages, and hands them back to the AI. The AI then explains the errors to you in plain words and can suggest fixes. It can also make changes to your scenes, like adding a node or loading a sprite, if you ask it to.

Click to zoom

What you can do with it

• Open the Godot editor for a project
• Run your game and capture the debug output
• Stop a running project when you are done
• Find Godot projects in a folder on your computer
• Create a new scene with a chosen root node
• Add nodes to an existing scene with properties you set
• Load a sprite or texture into a Sprite2D node
• Save a scene, including as a variant

Try asking your AI

• “Open my Godot project at C:/Games/MyPlatformer and run it”
• “Run my game and show me any errors that appear”
• “Create a new scene called Level2 with a Node2D as the root”
• “Add a Sprite2D node to my Main scene and load player.png into it”
• “Stop the running project and give me the debug output”

What it gives back to you

You get back plain text in your chat: the version of Godot you have, a list of projects found in a folder, or the console output from a run, including any error messages. When it changes a scene, it tells you what it did, like which node it added or which scene it saved. It does not show you a picture of your game, just the text Godot prints and the results of the actions.

Before you start

Who it's for

Hobbyist game makers, students and anyone learning Godot who wants their AI to actually run and debug their project instead of just talking about code.