The Model Context Protocol (MCP) is changing how developers interact with their tools. Instead of switching between an AI chat window and your editor, MCP lets AI assistants directly read, modify, and create content inside Unreal Engine.

This guide covers everything you need to know to get started with AI-assisted level design using the Unreal MCP Server.

What Is MCP?

MCP (Model Context Protocol) is an open standard that connects AI models to external tools. Think of it as a universal adapter — any MCP-compatible AI client can communicate with any MCP server.

For Unreal Engine, this means an AI assistant can:

• Spawn and position actors in your level
• Create and assign materials
• Modify Blueprint properties
• Control Sequencer timelines
• Adjust lighting, audio, and post-processing
• Query your scene for context-aware suggestions

The AI doesn't just generate code for you to paste. It directly manipulates the editor.

Setting Up the MCP Server

The Unreal MCP Server installs as a standard UE5 plugin. Once enabled, it exposes 200+ editor tools across 32 categories to any MCP-compatible client.

Supported AI Clients

The server works with any client that supports MCP:

• Claude Code — Anthropic's CLI tool
• Cursor — AI-powered code editor
• Windsurf — Codeium's AI IDE
• Claude Desktop — Anthropic's desktop app
• Any custom MCP client — the protocol is open

Connection

After installing the plugin, the MCP server starts automatically when you open your project. Point your AI client to the server address, and you're connected. No API keys, no cloud dependencies — everything runs locally.

Your First AI-Assisted Level

Let's walk through a practical example: building a simple village scene using natural language.

Step 1: Scene Setup

Tell your AI assistant:

"Create a flat landscape 500m x 500m. Add a directional light for daytime. Set the sky to partly cloudy."

The AI calls the appropriate MCP tools — landscape creation, light spawning, sky atmosphere configuration — and executes them in the editor. You see the results in your viewport immediately.

Step 2: Building Placement

"Place 8 medieval house meshes in a cluster near the center. Vary the rotation so they don't look uniform. Leave space for a central market square."

The AI uses actor spawning tools with randomized transforms. It understands spatial relationships, so "near the center" and "leave space for" translate into reasonable coordinates.

Step 3: Environment Details

"Add trees around the village perimeter. Place torches along the main path. Add a well in the market square."

Each request maps to specific MCP tool calls. The AI builds on the existing scene context — it knows where the houses are, where the square is, and places new actors accordingly.

Step 4: Lighting and Atmosphere

"Change the time of day to golden hour. Add volumetric fog with low intensity. Warm up the directional light color."

The AI adjusts light rotation, fog parameters, and color temperature through the MCP tools. You see the mood shift in real time.

The 5 Tool Presets

Not every workflow needs all 200+ tools. The MCP Server ships with 5 tool presets that load only the tools relevant to your task:

• Level Design — actors, transforms, landscape, foliage
• Visual Polish — materials, lighting, post-processing, atmosphere
• Blueprint Development — Blueprint creation, component management, properties
• Cinematics — Sequencer, cameras, animation
• Full Access — all 200+ tools enabled

Using presets reduces noise for the AI model, leading to more accurate tool selection and faster responses.

Context Resources: How the AI Understands Your Scene

Raw tool access isn't enough — the AI also needs to understand what's already in your scene. The MCP Server provides 12 context resources that give the AI awareness of:

• Current level actors and their properties
• Available asset types in your project
• Material and texture libraries
• Blueprint class hierarchies
• Active Sequencer state
• Viewport camera position and selection

This context means the AI can answer questions like "what materials are the buildings using?" or "how many lights are in the scene?" without you having to explain your project structure.

Full Undo Support

Every action the AI takes through MCP is recorded in Unreal's standard undo history. If the AI places something wrong or makes an unwanted change, Ctrl+Z works exactly as expected.

This is critical for trusting AI-assisted workflows. You're never locked into an AI decision — every action is reversible through the same undo system you already use.

Practical Workflows That Save Time

Rapid Prototyping

Use AI to block out levels with placeholder geometry. Describe the space in plain language, let the AI build the rough layout, then refine manually. This cuts blockout time from hours to minutes.

Asset Auditing

Ask the AI to scan your level and report issues: "Find all actors with missing meshes," "List lights with intensity above 10," "Show me all unlit materials." The context resources make these queries instant.

Batch Operations

Need to replace every instance of one mesh with another? Adjust the intensity of every point light in a sublevel? Change the material on 50 actors? Describe the operation in natural language and let the AI handle the repetitive work.

Blueprint Scaffolding

The MCP Server includes 33 Blueprint-specific tools. Ask the AI to create Blueprint classes, add components, set default properties, and wire up basic logic. It won't write your entire game, but it handles the boilerplate.

Tips for Effective AI-Assisted Level Design

1. Be specific about quantities and distances — "place 12 trees in a 50m radius" works better than "add some trees nearby"
2. Work in passes — layout first, then details, then polish, just like manual level design
3. Use the right preset — fewer available tools means more accurate AI decisions
4. Review and iterate — treat AI output as a first draft, then refine manually
5. Combine with manual work — AI handles the repetitive parts, you handle the creative decisions

Getting Started

Install the Unreal MCP Server plugin, connect your preferred AI client, and start with a simple scene. The documentation walks you through setup in detail.

AI-assisted level design isn't about replacing designers. It's about removing the tedious parts so you can focus on what makes your game world interesting.