AssetBundleMCP 0.1.11

AssetBundleMCP

日本語

AssetBundleMCP is an MCP (Model-Context-Protocol) server for easily and efficiently analyzing Unity AssetBundles through conversation with an AI assistant (such as gemini-cli).

By using this tool, developers and QA engineers can quickly grasp information such as a list of assets and texture details contained in an AssetBundle simply by asking questions in natural language.

Main Features

• Interactive AssetBundle Analysis: Analyze AssetBundles just by talking to your AI assistant.
• Easy Installation: Provided as a NuGet package, so you can easily introduce it by adding it to your configuration file.
• Flexible Data Access: You can also write SQL queries directly to the AI and freely extract information from the database-formatted analysis results.

Requirements

• .NET 9.0 SDK or later

Installation & Setup

If .NET 10 is installed (Recommended)

Starting from .NET 10 preview 6, you can run it directly using dnx.

Please follow the documentation for your AI assistant to set up AssetBundleMCP as an MCP server.

• For Visual Studio Code: .vscode/mcp.json
• For Visual Studio: .mcp.json

{
  "servers": {
    "AssetBundleMCP": {
      "type": "stdio",
      "command": "dnx",
      "args": [
        "AssetBundleMCP",
        "--yes"
      ]
    }
  }
}

• For Gemini Cli: .gemini/settings.json

{
  "mcpServers": {
    "AssetBundleMCP": {
      "command": "dnx",
      "args": [
        "AssetBundleMCP",
        "--yes"
      ]
    }
  }
}

Before .NET 10 preview 6 (Not Recommended)

1. Clone the Repository:

Clone with the --recurse-submodules option to also fetch the Git submodules.

git clone --recurse-submodules [https://github.com/hanachiru/AssetBundleMCP.git](https://github.com/hanachiru/AssetBundleMCP.git)
cd AssetBundleMCP

2. Build the Project:

dotnet build -c Release

3. Configure the MCP server:

Please follow the documentation for your AI assistant to set up AssetBundleMCP as an MCP server.

• For Visual Studio Code: .vscode/mcp.json
• For Visual Studio: .mcp.json

{
  "servers": {
    "AssetBundleMCP": {
      "type": "stdio",
      "command": "dotnet",
      "args": [
        "run",
        "--project",
        "<PATH TO PROJECT DIRECTORY>/src/AssetBundleMCP/AssetBundleMCP.csproj"
      ]
    }
  }
}

• For Gemini Cli: .gemini/settings.json

{
  "mcpServers": {
    "AssetBundleMCP": {
      "command": "dotnet",
      "args": [
        "run",
        "--project",
        "<PATH TO PROJECT DIRECTORY>/src/AssetBundleMCP/AssetBundleMCP.csproj"
      ]
    }
  }
}

Usage

1. Loading AssetBundles:

Specify the path to the directory containing the AssetBundles or the AssetBundle files you want to analyze, then instruct the system to load them.

Please load the AssetBundles in C:/path/to/your/assetbundles

The tool will analyze the AssetBundles and save the results to a temporary database file. If you do not specify an output destination, an SQLite file will be created in the current directory.

2. Retrieving Information:

Once loading is complete, you can ask various questions.

• To get a list of assets:

  Please show me the list of assets in the AssetBundle

• To get a list of textures:

  Please show me the list of textures in the AssetBundle

3. Finishing Analysis:

When you are done analyzing, unload the database and release resources with the following command:

Please unload the AssetBundle database

Available Tools

License

This project is licensed under the MIT License.

Acknowledgements

The core analysis functionality of this tool uses UnityDataTools. Many thanks to the developers of this excellent library.

Notes

None of the repo, the tool, nor the repo owner is affiliated with, or sponsored or authorized by, Unity Technologies or its affiliates.
This is intended for use in developing your own products, and should not be used for reverse engineering other companies' products.