The AI Game Developer ecosystem is intentionally modular — install only the parts you need. 72 capabilities across the plugin and its domain extensions (21 MCP tools plus RQL), two .NET libraries that power the bridge, the Unity MCP CLI, the desktop app and gamedev CLI, and a template for authoring your own extensions.
L1
The plugin
Drop into Unity. AI gets a steering wheel.
◆
AI Game Developer
plugin·v1.1.0
Unity Editor & runtime plugin — 72 capabilities (MCP tools + RQL) for AI to drive your project.
21 MCP tools plus RQL: one query language for scenes, GameObjects, assets, prefabs and the console
Works in Editor and at runtime inside compiled builds
Real-time skill discovery for Claude / Cursor / Copilot / Gemini
Extensible via custom [AiTool] tools and RQL type adapters
openupm add com.ivanmurzak.unity.mcp
L2
Domain extensions
Bolt on specialty tools for the work you do.
◔
AI Animation
extension·v2.0.0
Domain extension — create & modify AnimationClips and AnimatorControllers via prompts.
Bidirectional read/write for AnimationClip data
Create AnimatorController assets with auto-folders
Edit layers, states, transitions, parameters via JSON patches
Batch-create animations across multiple paths
openupm add com.ivanmurzak.unity.mcp.animation
✺
AI ParticleSystem
extension·v2.0.0
Domain extension — inspect & modify all 24 ParticleSystem modules through natural language.
Need animation, particles, or ProBuilder modelling? Drop in the matching extension package — it teaches RQL the new area and the agent can use it right away.
Under the hood the plugin uses MCP-Plugin-dotnet as a SignalR bridge, and ReflectorNet for fuzzy method discovery + JSON-schema generation. Both are open-source NuGet packages — pull them into your own .NET app to expose tools the same way Unity does.