Godot MCP Runtime
Related Servers
Alternatives to Godot MCP Runtime
- AlicenseBqualityDmaintenanceA Model Context Protocol server that enables AI assistants to interact with the Godot game engine, allowing them to launch the editor, run projects, capture debug output, and control project execution.146,695MIT
- AlicenseNot gradedqualityDmaintenanceProvides a comprehensive integration between LLMs and the Godot Engine, enabling AI assistants to intelligently manipulate project files, scripts, and the live editor. It supports advanced workflows including version-aware documentation querying, automated E2E game testing, and real-time visual context capture.26MIT
Related Servers
- AlicenseNot gradedqualityDmaintenanceAn MCP server that lets an AI agent work on a Godot 4.x project the way a developer does: read and edit scenes, write scripts, build, test, run, look at the result, and debug what went wrong.187 npm2AGPL 3.0
- AlicenseAqualityAmaintenanceOpen-source MCP server connecting AI agents (Claude, Cursor, GitHub Copilot, Gemini, and more) to the Godot Editor and runtime (Godot 4.x, C#). Additionally you can use AI Game Dev app (https://ai-game.dev) to generate images, video, audio, 3D, rig, animation and much more.14345Apache 2.0
- AlicenseAqualityCmaintenanceAn MCP server that lets AI coding agents work on Godot 4 projects, including reading and editing scenes, writing and validating GDScript, running the game, and capturing screenshots.2530 npmMIT
- AlicenseNot gradedqualityBmaintenanceA TypeScript MCP server bridging MCP clients to Godot 4 editor, enabling scene, node, script editing and more via WebSocket.187 npmMIT
- AlicenseNot gradedqualityBmaintenanceThe most advanced MCP server for Godot Editor that lets AI assistants operate directly inside your running Godot project for scene creation, script generation, UI authoring, and more.55MIT
- AlicenseNot gradedqualityCmaintenanceA local MCP server plus a bundled Godot editor addon that lets an AI agent create, inspect, run, debug, and export real Godot 4.6 games through tools.2PolyForm Small Business 1.0.0
TDQS
Scored across 43 tools
Most tools have clearly scoped resource-and-action boundaries, and descriptions explicitly route ambiguous pairs such as profile_project versus start_profiler/stop_profiler, launch_editor versus run_project, save_scene versus auto-save, and batch_scene_operations versus single mutations. Still, with 43 tools there are several near-overlaps (load_sprite/set_node_properties, create_scene/add_node, get_project_settings/check_project) that require reading descriptions to choose correctly.
All tool names use snake_case and almost all are verb-first (get_, set_, add_, remove_, create_, run_, stop_), giving a predictable pattern. Minor deviations such as validate (single verb) and batch_scene_operations (noun-heavy phrasing) are small and still readable.
43 tools is well above the 25-tool threshold and forces agents to choose from many edge-case operations; the scope is broad but still feels over-fragmented across runtime, profiling, scene editing, resources, autoloads, and project settings. Some specialized tools like export_mesh_library, render_movie, duplicate_node, and move_nodes earn their place only in narrow workflows.
Coverage is broad: project discovery/settings, autoload CRUD, scene/node lifecycle and signals, scripts/validation, runtime sessions, screenshots/input/UI, profiling, and movie rendering. Minor gaps remain around file-level deletion/rename and resource read/delete (create_resource and set_resource_properties exist without matching get/delete), but core workflows are covered.