Vault MCP Server
Enables AI assistants like Claude to browse, search, and read data from an Autodesk Vault server by exposing its REST API operations as tools.
README
Vault MCP Server
An MCP (Model Context Protocol) server that exposes Autodesk Vault REST API operations as tools, enabling AI assistants like Claude to browse, search, and read data from an Autodesk Vault server.
Prerequisites
- Python 3.10+
- An Autodesk Vault server with REST API access enabled
- Vault user credentials with appropriate permissions
Installation
pip install -r requirements.txt
Configuration
Edit config.json before running the server. All fields are required.
{
"vault": {
"servername": "VaultServer",
"username": "Administrator",
"password": "your-password-here",
"database": "Inventor"
},
"server": {
"host": "0.0.0.0",
"port": 8080
},
"logging": {
"level": "INFO",
"file": "Log/mcp_server.log"
}
}
| Field | Description |
|---|---|
vault.servername |
Hostname or IP of your Vault server |
vault.username |
Vault login username |
vault.password |
Vault login password |
vault.database |
Vault database name (e.g. Inventor, Vault) |
server.host |
Bind address for SSE mode (0.0.0.0 = all interfaces) |
server.port |
Port for the SSE HTTP server (default 8080) |
logging.level |
Log verbosity: DEBUG, INFO, WARNING, or ERROR |
logging.file |
Path to the rotating log file |
Running the Server
SSE Mode (Claude Code / remote clients)
This is the default mode. It starts an HTTP server that MCP clients connect to over the network.
python app.py
The server will be available at:
- WebServer:
http://localhost:8080 - SSE endpoint:
http://localhost:8080/sse - Messages endpoint:
http://localhost:8080/messages
To use a custom config file:
python app.py --config path/to/my_config.json
stdio Mode (Claude Desktop)
For Claude Desktop, the server runs as a subprocess communicating over stdin/stdout.
python app.py --transport stdio
Connecting to Claude Code
Add the server to your Claude Code MCP configuration (.claude/settings.json or via claude mcp add):
{
"mcpServers": {
"vault": {
"type": "sse",
"url": "http://localhost:8080/sse"
}
}
}
Or use the CLI:
claude mcp add --transport sse vault http://localhost:8080/sse
Connecting to Claude Desktop
Add the following to your Claude Desktop configuration file (claude_desktop_config.json):
{
"mcpServers": {
"vault": {
"command": "python",
"args": [
"C:/path/to/Vault MCP/app.py",
"--transport", "stdio"
]
}
}
}
Replace C:/path/to/Vault MCP/ with the actual path to this project directory. Claude Desktop will launch the server automatically as a subprocess.
Available Tools
Once connected, the following tools are exposed to the AI assistant:
| Tool | Description |
|---|---|
vault_get_server_info |
Get Vault server version and metadata |
vault_sign_in |
Authenticate with different credentials |
vault_sign_out |
Invalidate the current session |
vault_list_vaults |
List all accessible vaults |
vault_get_vault |
Get details for a specific vault |
vault_get_folder_contents |
List files and sub-folders in a folder |
vault_get_folder |
Get metadata for a specific folder |
vault_get_file |
Get metadata for a specific file |
vault_get_file_versions |
List all versions of a file |
vault_get_file_download_url |
Get the download URL for a file |
vault_search_files |
Keyword search across vault files |
vault_advanced_search |
Structured search using property criteria |
vault_list_groups |
List all groups in the vault |
vault_get_group |
Get details for a specific group |
vault_list_users |
List all users in the vault |
vault_get_user |
Get details for a specific user |
vault_list_property_definitions |
List user-defined property definitions |
vault_get_property_definition |
Get a specific property definition |
vault_search_items |
Search for engineering/BOM items |
vault_get_item |
Get details for a specific engineering item |
vault_list_lifecycle_definitions |
List lifecycle definitions |
vault_list_category_definitions |
List category definitions |
Logs
Logs are written to Log/mcp_server.log (rotating, max 5 MB × 5 files). In SSE mode, logs also print to the console. In stdio mode, only the file is written to (stdout is reserved for the MCP protocol).
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。