Unity MCP Server serves as a bridge between MCP Client and the Unity Engine. It exposes and explains Unity's tools to the LLM, enabling AI-driven workflows inside Unity. The server supports multiple transport protocols (http and stdio) and can run locally, in Docker, or as a binary, with configurable port, plugin timeout, and client transport. It is designed to be extensible: Unity projects can define custom MCP Tools, MCP Resources, and MCP Prompts in code, which are exposed to MCP Clients at runtime. MCP Tools are executable functions with typed parameters and return values, and can run on the main thread or in background threads. MCP Resources provide read-only access to project data and Unity objects, giving the AI contextual information about the current state of the project. MCP Prompts allow injecting predefined prompts into conversations. The server supports multiple LLM providers and includes runtime usage guidance with the Unity MCP Plugin and example integrations. Deployment options include Docker, OpenUPM installation, or binary distributions.