UE Agent Asset Toolkit

UE Asset Toolkit

An MCP server that lets AI coding agents inspect Unreal Engine assets without launching the Editor. Parses .uasset binaries directly, builds a search index, and exposes two tools — unreal_search and inspect_asset — over MCP.

In Action (Lyra Starter Game)

Requirements

• .NET 8 SDK or later (for the C# asset parser)
• Python 3.10+

Installation

git clone --recursive https://github.com/bradenleague/UE-Agent-Asset-Toolkit
cd UE-Agent-Asset-Toolkit
python -m venv .venv
source .venv/bin/activate  # macOS/Linux
# .venv\Scripts\activate   # Windows
pip install -e .

This installs two CLI entry points:

• unreal-agent-toolkit — index management
• unreal-agent-mcp — MCP server

Note: the UAssetAPI submodule is ~70MB, so the initial clone can take a minute. For build details, see CONTRIBUTING.md.

Optional dependencies

pip install -e ".[embeddings]"  # vector embeddings (sentence-transformers)
pip install -e ".[dev]"         # pytest + coverage

Quick Start

# 1. Build the parser and configure your project
python setup.py /path/to/YourProject.uproject

# 2. Build the search index
unreal-agent-toolkit

# 3. Connect to your MCP client (see below)

setup.py builds the C# parser, installs Python packages, and registers your .uproject. Add --index to combine steps 1 and 2.

MCP Client Setup

Register the server in your MCP client config:

{
  "mcpServers": {
    "unreal": {
      "command": "unreal-agent-mcp"
    }
  }
}

If you didn't pip install -e ., use the full Python path instead:

{
  "mcpServers": {
    "unreal": {
      "command": "/path/to/.venv/bin/python",
      "args": ["/path/to/UE-Agent-Asset-Toolkit/unreal_agent/mcp_server.py"]
    }
  }
}

See mcp-config.example.json for a copy-paste template.

MCP Tools

unreal_search

Search assets and C++ source code. The search mode is auto-detected:

inspect_asset

Returns structured data for a specific asset:

• Blueprint: parent class, components, functions, variables, graph
• WidgetBlueprint: widget tree hierarchy, bindings
• Material: parameters (scalar, vector, texture), domain
• DataTable: row structure, columns, sample data

Indexing

unreal-agent-toolkit                        # full index (default)
unreal-agent-toolkit --plugins              # include plugin content
unreal-agent-toolkit --plugins --embed      # + vector embeddings
unreal-agent-toolkit --source               # C++ source (UCLASS, UPROPERTY)
unreal-agent-toolkit --profile quick        # high-value types only
unreal-agent-toolkit --path UI              # only Content/UI/
unreal-agent-toolkit --force                # re-index everything
unreal-agent-toolkit --status               # show index stats

The --path option uses Unreal virtual paths (UI maps to Content/UI/).

Multi-Project Setup

unreal-agent-toolkit add "/path/to/MyGame.uproject"
unreal-agent-toolkit list
unreal-agent-toolkit use mygame

Each project gets an isolated SQLite database at unreal_agent/data/<project>.db.

Architecture

MCP Client (Claude Desktop, Claude Code, etc.)
        │
        ▼
MCP Server (Python, stdio)
├── unreal_search  →  SQLite FTS5 index
└── inspect_asset  →  C# AssetParser  →  .uasset binaries

The C# parser (UAssetAPI) reads .uasset files directly. The Python layer builds an FTS5 search index and serves results over MCP.

Known Limitations

• Read-only — cannot modify assets
• UE version coverage — tested against UE 5.5; newer versions may have unsupported formats
• Type detection — uses naming conventions (BP_, WBP_, DT_) by default; create a profile in unreal_agent/profiles/ for project-specific types

License

• AssetParser / unreal_agent: MIT
• UAssetAPI: See UAssetAPI/LICENSE