M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
U

Unreal Engine MCP Server

Community
Automate Unreal Engine 5 with AI.
How to install ↓ Compare GitHub source repository ↗
Category
Dev Tools #219 of 438
Stars
★ 958 Very popular
Transport
stdio (local process) · Streamable HTTP
Runtime
Node.js 18+ · Docker · Prebuilt binary
Credentials
No credential needed
License
MIT
Last commit
Tools
23
53FMRS · D

This is a broad Unreal Engine 5 automation server covering workflows from assets and Blueprints to environments, AI, effects, audio, networking, and cinematics. It requires an Unreal plugin and project-level editing access; LAN access should be configured carefully.

Strongest · Documentation 13/20 Weakest · Reliability 9/20

Reliability
9/20
Security and permissions
10/20
Maintenance
9/20
Documentation
13/20
Setup experience
12/20
Why each score
Reliability 9/20
The materials describe stdio, native HTTP, and WebSocket paths, with 23 tools plus timeout and rate-limit settings. However, they provide no CI workflow, committed test evidence, or verifiable execution evidence for MCP initialization, tool listing, or key failure paths. Native HTTP is also absent from the supplied server.json package definition, and operation depends on an Unreal plugin and project configuration. The happy path is plausible, but edge behavior and failure handling are not established.
Security and permissions 10/20
The default loopback binding, explicit non-loopback opt-in, TLS option, capability-token support, message limits, and dangerous-console-command filtering are meaningful controls. Deductions apply because authentication is optional, while Python execution, UBT, project settings, console commands, and asset deletion expose high-impact capabilities. The materials do not establish per-operation confirmation, fine-grained permissions, path isolation, or complete command-filter coverage; LAN safety depends on correct user configuration. No evidence of malware or credential theft reaches the red-line threshold.
Maintenance 9/20
The project identifies an author, MIT license, GitHub repository, and issue tracker; it also reports version 0.5.30 and documents development and test commands. Deductions apply because the supplied material contains no release history, commit cadence, issue-response evidence, dependency-update policy, CI results, or dedicated security-response channel. Stars and open-issue counts are discovery signals only and do not add points.
Documentation 13/20
The README covers installation, plugin compilation and prebuilt binaries, multiple transports, environment variables, client configuration, security warnings, tool categories, Docker, troubleshooting, and development commands. Deductions apply because it lacks per-tool parameter and return-value documentation, detailed permission boundaries, verifiable test evidence, and cost or resource-impact guidance. The manifest and README also present transport capabilities inconsistently.
Setup experience 12/20
NPX, source-build, Docker, and native HTTP paths are documented, with Claude Code and Cursor configuration examples and Windows/macOS/Linux guidance. Deductions apply because working use requires installing and configuring an Unreal plugin, a code target or version-specific prebuilt binary, multiple UE plugins, and project settings. Version compatibility and compilation requirements add substantial complexity, and no execution verification is supplied.

Static review · not runListed 2026-07-29

Read the FMRS scoring method →

Fit and risk

What it can accessReads local filesWrites / deletes local filesRuns commands or codeUses the network

Best for

  • Teams developing games, levels, animation, or visual effects with Unreal Engine 5.
  • Users who want AI to perform repetitive editor and content-production tasks.
  • Users comfortable managing Unreal projects, plugins, and local development environments.

Not for

  • Projects that do not use Unreal Engine.
  • Blueprint-only environments that cannot install a pre-built plugin or compile a code plugin.
  • Public or remote exposure without appropriate network security configuration.

Required permissions

  • Access to and control of the specified Unreal project.
  • Installation and execution of the MCP Automation Bridge plugin and required Unreal plugins.
  • Some features require Python, Editor Scripting Utilities, Niagara, Gameplay Abilities, Smart Objects, or other plugins.
  • Node.js 18 or later is required for the TypeScript Bridge.

Risks and side effects

  • The server can modify assets, actors, levels, project settings, and other game-development content.
  • Enabling non-loopback addresses exposes the automation bridge to the local network; use trusted networks and enable capability-token authentication.
  • Incorrect console commands, scripts, or automation actions can change project content or editor state.
  • Pre-built plugin binaries are tied to specific Unreal Engine versions.

Setup

Before you start

Runtime：Node.js 18+ · Docker · Prebuilt binary

UE_PROJECT_PATH required Absolute path to your Unreal .uproject file; needed at install time
MCP_METRICS_TOKEN optionalsecret Access token for the Prometheus metrics endpoint
Other optional settings (18)
MCP_AUTOMATION_PORT optional Automation bridge WebSocket port (default: 8091)
MCP_AUTOMATION_HOST optional Automation bridge WebSocket host (default: 127.0.0.1)
MCP_AUTOMATION_ALLOW_NON_LOOPBACK optional Set true to allow binding to non-loopback addresses for LAN access; default false, security-sensitive
MCP_AUTOMATION_CLIENT_MODE optional Run as client connecting to UE (false) or server listening for UE (true)
MCP_AUTOMATION_USE_TLS optional Use wss:// for the automation bridge client connection (default false)
MCP_CONNECTION_TIMEOUT_MS optional Connection timeout in ms (default 5000)
MCP_REQUEST_TIMEOUT_MS optional Request timeout in ms (default 30000)
MCP_AUTOMATION_MAX_MESSAGES_PER_MINUTE optional Max inbound bridge messages per minute before disconnect (default 600)
MCP_AUTOMATION_MAX_AUTOMATION_REQUESTS_PER_MINUTE optional Max inbound automation request messages per minute (default 120)
LOG_LEVEL optional Logging level: debug/info/warn/error (default info)
MCP_ROUTE_STDOUT_LOGS optional Route stdout to stderr to avoid protocol corruption (default true)
UE_EDITOR_EXE optional Optional path to UnrealEditor-Cmd.exe
UE_SCREENSHOT_DIR optional Optional directory for saving screenshots
ASSET_LIST_TTL_MS optional Asset list cache TTL in ms (default 10000)
MCP_METRICS_PORT optional Optional Prometheus metrics endpoint port (default 9100)
MCP_METRICS_HOST optional Metrics endpoint host, loopback-only by default
MCP_METRICS_ALLOW_NON_LOOPBACK optional Allow metrics endpoint on non-loopback addresses; requires a token
MCP_ADDITIONAL_PATH_PREFIXES optional Comma-separated custom content mount point prefixes

For the TypeScript Bridge, install Node.js 18 or later and install the MCP Automation Bridge plugin in the Unreal project. Run npx unreal-engine-mcp-server, or clone the repository and run npm install, npm run build, and node dist/cli.js. Set UE_PROJECT_PATH to the Unreal project path. Alternatively, enable Native MCP in the plugin and connect the client to http://localhost:3000/mcp.

claude_desktop_config.json 
{"mcpServers":{"unreal-engine":{"command":"npx","args":["unreal-engine-mcp-server"],"env":{"UE_PROJECT_PATH":"C:/Path/To/YourProject"}}}}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json
{
  "servers": {
    "unreal-engine": {
      "command": "npx",
      "args": [
        "unreal-engine-mcp-server"
      ],
      "env": {
        "UE_PROJECT_PATH": "C:/Path/To/YourProject"
      }
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal
claude mcp add unreal-engine -e UE_PROJECT_PATH=C:/Path/To/YourProject -- npx unreal-engine-mcp-server

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

The inspect tool should appear in your MCP client's tool list; calling inspect should return object introspection data, and the Unreal Output Log (LogMcpNativeTransport or Automation Bridge entries) should show a successful connection handshake.

Troubleshooting

  1. Verify that UE_PROJECT_PATH points to a valid .uproject file or project path.
  2. Verify that MCP Automation Bridge is installed, compiled, and enabled in Unreal Editor.
  3. If a code project reports missing modules, build through Visual Studio or Xcode; Blueprint-only projects need a code target or pre-built plugin.
  4. Check MCP_AUTOMATION_HOST, MCP_AUTOMATION_PORT, and connection and request timeout settings.
  5. In Native MCP mode, check the editor status bar and LogMcpNativeTransport output.
  6. Confirm that required Unreal plugins are enabled and review the optional plugins needed by the selected tool.

Things to try

Once connected, you can ask your AI assistant things like:

  • Generate a 50x50 landscape and populate it with procedural foliage
  • Create a ThirdPerson blueprint and add a spring arm component to it
  • Take a screenshot of the current viewport and save it to the screenshot directory
  • Run all automation tests and write the results to the log

Tools 23

manage_asset writes
Manage assets, materials, render targets, behavior trees, and Blueprint structs.
manage_blueprint writes
Manage Blueprints, SCS components, graphs, UMG widgets, layouts, bindings, and animations.
control_actor writes
Spawn, delete, and transform actors, and manage physics, tags, and components.
control_editor writes
Control PIE sessions, cameras, viewports, screenshots, and bookmarks.
manage_level writes
Load and save levels and manage level streaming and lighting.
system_control writes
Run UBT, tests, logs, project settings, CVars, and Python operations.
inspect read-only
Inspect Unreal objects.
manage_tools writes
Enable or disable tools dynamically at runtime.
Show 15 more tools
build_environment writes
Manage landscapes, foliage, procedural environments, lighting, and spline roads, rivers, and fences.
manage_level_structure writes
Manage levels, sublevels, World Partition, streaming, data layers, HLOD, and volumes.
manage_geometry writes
Create and edit procedural meshes with Geometry Script.
manage_pcg writes
Manage PCG graph assets, nodes, pin connections, execution, partition grid size, and node settings.
animation_physics writes
Manage Animation Blueprints, skeletons, sockets, physics assets, cloth, vehicles, ragdolls, Control Rig, and IK.
manage_effect writes
Manage Niagara, particles, debug shapes, and GPU simulations.
manage_gas writes
Manage Gameplay Ability System abilities, effects, and attributes.
manage_character writes
Create characters and manage movement and advanced locomotion.
manage_combat writes
Manage weapons, projectiles, damage, and melee systems.
manage_ai writes
Manage AI controllers, Behavior Trees, EQS, perception, State Trees, Smart Objects, and navigation.
manage_inventory writes
Manage items, equipment, loot tables, and crafting systems.
manage_interaction writes
Manage interactables, destructibles, and triggers.
manage_audio writes
Manage audio assets, components, Sound Cues, MetaSounds, and attenuation.
manage_sequence writes
Manage Sequencer, cinematics, Movie Render Queue, media, Take Recorder, and replay controls.
manage_networking writes
Manage replication, RPCs, network prediction, sessions, split-screen, LAN, voice, game framework, and input mappings.

Use cases

Automate Unreal Engine editor and game-development workflows with natural language.
Create and manage assets, Blueprints, levels, environments, effects, animation, AI, and audio.
Control PIE, Sequencer, cinematic rendering, and editor inspection workflows.

Supported clients

Claude Desktop
Cursor
Claude Code

Listed from the project's documentation, not tested by this site.

Overview

Unreal Engine MCP Server is an MCP server built with TypeScript and C++ that lets AI assistants automate Unreal Engine through a native C++ Automation Bridge plugin. It exposes 23 tools covering assets, actors, editor control, levels, Blueprints, animation, physics, effects, AI, audio, networking, and Sequencer workflows. The server supports stdio and the plugin's native Streamable HTTP transport.

Similar servers

Context7 80 · B

Upstash's official server providing up-to-date third-party library docs for AI coding assistants

★ 63.3k · Tools 2 Compare with this →
Compare with other MCP servers

Source revision 99fbce87ddf7 Data synced 2026-10-19 Read the FMRS scoring method