UnrealMCPHub
Central management platform for Unreal Engine MCP instances that bridges AI agents with Unreal Engine, handling plugin installation, project compilation, editor launch, crash recovery, and transparent proxy of tool calls.
README
UnrealMCPHub
Central management platform for Unreal Engine MCP instances — bridge between AI agents and UE across the full development lifecycle.
Works with RemoteMCP (UE Editor plugin that exposes engine capabilities via MCP protocol). The Hub manages everything around it: install the plugin, compile, launch the editor, detect crashes, and transparently proxy all tool calls — so the AI agent only needs one MCP endpoint.
AI Agent ──► UnrealMCPHub ──► UE Editor + RemoteMCP plugin
(this repo) (blackplume233/UnrealRemoteMCP)
RemoteMCP runs inside UE Editor and provides 17+ tools across 6 domains (level, blueprint, umg, edgraph, behaviortree, slate). UnrealMCPHub runs outside UE Editor and can compile, launch, monitor, and proxy — even when UE is not running.
Features
- Project setup — Configure
.uprojectonce; engine auto-detected from registry - Build & launch — Compile via UBT, launch editor, wait for MCP readiness
- Plugin install — One-click RemoteMCP installation (local copy or GitHub download)
- Instance discovery — Scan ports to find running UE editors
- UE tool proxy —
ue_run_python,ue_call,ue_list_toolsand domain dispatch - Crash resilience — Crash detection, report retrieval, restart flow
- Multi-instance — Switch between multiple UE editors via
use_editor - Session notes — Persist context for crash recovery
- One-click overview —
hub_statusshows project, plugin, instances, and watcher state
Quick Install
Option A: One-line install from PyPI (recommended)
# uv (recommended)
uv tool install unrealhub
# or pip
pip install unrealhub
Then add to .cursor/mcp.json:
{
"mcpServers": {
"unrealhub": {
"command": "unrealhub",
"args": ["serve"]
}
}
}
Option B: Zero-install (auto-fetch on first launch)
No download, no clone, no install. Paste this single block into .cursor/mcp.json and restart Cursor:
{
"mcpServers": {
"unrealhub": {
"command": "uvx",
"args": ["unrealhub", "serve"]
}
}
}
Requires uv (
pip install uvorwinget install astral-sh.uv). Cursor will auto-fetch the Hub on first launch.
Option C: Standalone executable (no Python needed)
Download the latest executable for your platform from GitHub Releases, then:
{
"mcpServers": {
"unrealhub": {
"command": "/path/to/unrealhub",
"args": ["serve"]
}
}
}
| Platform | File |
|---|---|
| Windows x64 | unrealhub-windows-amd64.exe |
| macOS ARM | unrealhub-macos-arm64 |
| Linux x64 | unrealhub-linux-amd64 |
Option D: Clone for development
git clone https://github.com/blackplume233/UnrealMCPHub.git
cd UnrealMCPHub
uv sync # or: pip install -e .
{
"mcpServers": {
"unrealhub": {
"command": "uv",
"args": ["--directory", "/path/to/UnrealMCPHub", "run", "unrealhub", "serve"]
}
}
}
Quick Start
For AI Agents (Cursor / Claude / etc.)
This is the primary use case. Add UnrealMCPHub as an MCP server in your AI tool, then the agent handles everything through natural language.
Step 1: Install & Configure MCP (see Quick Install above)
HTTP mode (for shared / remote / multi-client use):
unrealhub serve --http --port 9422
{
"mcpServers": {
"unrealhub": {
"url": "http://127.0.0.1:9422/mcp"
}
}
}
Step 2: Talk to the agent
Once configured, the AI agent has full access to all Hub tools. Example conversations:
You: "帮我编译并启动 UE 项目 D:/Projects/MyGame/MyGame.uproject"
Agent: [calls setup_project, compile_project, launch_editor automatically]
You: "在 UE 里创建一个蓝图 Actor"
Agent: [calls ue_get_dispatch to find blueprint tools, then ue_call_dispatch]
You: "UE 崩溃了怎么办"
Agent: [calls get_crash_report, shows crash info, offers restart_editor]
Step 3: What happens behind the scenes
Agent → Hub (setup_project) # One-time project config, persisted
Agent → Hub (compile_project) # Compiles via UBT, even without UE running
Agent → Hub (launch_editor) # Starts UE, waits for MCP readiness
Agent → Hub (ue_run_python) # Hub forwards to UE's RemoteMCP
↓
UE Editor (port 8422) # Executes Python, returns result
The agent only needs to know about the Hub — it never talks to UE directly.
AI Agent Decision Flow
Is project configured?
├── No → Ask user for .uproject path → setup_project()
└── Yes → Is UE Editor online?
├── No → Need plugin? → install_plugin() → compile_project() → launch_editor()
└── Yes → Use ue_* tools directly
└── Crashed? → get_crash_report() → restart_editor()
For Humans (CLI)
unrealhub setup /path/to/MyProject.uproject # Configure project
unrealhub serve # Start MCP server (stdio)
unrealhub serve --http --port 9422 # Start MCP server (HTTP)
unrealhub status # Show instance status
unrealhub discover # Discover running UE instances
unrealhub compile # Build active project
unrealhub launch # Launch editor
Tool Reference
Hub Management (always available, even without UE)
| Tool | Description |
|---|---|
setup_project |
Configure project path (once, persisted to ~/.unrealhub) |
get_project_config |
View current project configuration |
hub_status |
One-stop overview of everything |
compile_project |
Compile project via UBT |
launch_editor |
Start UE Editor, wait for MCP readiness |
restart_editor |
Restart a crashed editor |
install_plugin |
Install RemoteMCP plugin |
set_plugin_source |
Configure plugin download URL or local path |
discover_instances |
Scan ports for running UE instances |
use_editor |
Switch active UE instance (multi-instance) |
get_crash_report |
Get crash details |
add_note / get_notes |
Session notes for crash recovery context |
UE Proxy Tools (forwarded to active UE instance)
| Tool | Description |
|---|---|
ue_run_python |
Execute Python script in UE |
ue_call |
Call any UE MCP tool by name |
ue_list_tools |
List all tools from UE instance |
ue_get_dispatch |
List domain tools (level, blueprint, umg, etc.) |
ue_call_dispatch |
Call a domain-specific tool |
ue_test_state |
Test connection to UE |
ue_status |
Get active instance status |
Architecture
┌─────────────────┐ stdio/HTTP ┌──────────────────┐
│ Cursor / Agent │ ◄─────────────────► │ UnrealMCPHub │
└─────────────────┘ │ (FastMCP Server) │
└────────┬─────────┘
│
┌───────────────────────────────┼───────────────────────────────┐
│ │ │
▼ ▼ ▼
┌───────────────┐ ┌───────────────┐ ┌───────────────┐
│ Project Mgmt │ │ Lifecycle │ │ UE Proxy │
│ setup_project │ │ compile │ │ ue_run_python │
│ install_plugin│ │ launch/restart│ │ ue_call │
│ hub_status │ │ discover │ │ ue_list_tools │
└───────────────┘ └───────┬───────┘ └───────┬───────┘
│ │ │
▼ │ Streamable HTTP │
┌───────────────┐ ┌──────▼──────────────────────────────▼────────┐
│ ~/.unrealhub │ │ UE Editor + RemoteMCP (port 8422) │
│ config.json │ │ 17+ tools: run_python, get_dispatch, etc. │
│ state.json │ │ 6 domains: level, blueprint, umg, edgraph, │
└───────────────┘ │ behaviortree, slate │
└──────────────────────────────────────────────┘
Development
git clone https://github.com/blackplume233/UnrealMCPHub.git
cd UnrealMCPHub
uv sync --extra dev # Install with dev dependencies
uv run pytest tests/ -v # Run tests (129 tests)
Requirements
- Python >= 3.11
- Unreal Engine 5.x
- RemoteMCP — UE Editor plugin (Hub can auto-install it via
install_plugin)
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.