ADLS2 MCP Server 🚀
Microsoft Azure Data Lake Storage MCP Server
erikhoward
README
ADLS2 MCP Server 🚀
A Model Context Protocol (MCP) server implementation for Azure Data Lake Storage Gen2. This service provides a standardized interface for interacting with ADLS2 storage, enabling file operations through MCP tools.
Setup 🛠️
Installation 📦
Requires Python 3.13 or higher.
Install the package using uv
:
uv pip install adls2-mcp-server
MCP Configuration ⚙️
Claude Desktop Configuration
1 - Edit Claude Desktop Configuration:
Open claude_desktop_config.json
and add the following configuration.
On MacOs, the file is located here:
~/Library/Application Support/Claude Desktop/claude_desktop_config.json
.
On Windows, the file is located here:
%APPDATA%\Claude Desktop\claude_desktop_config.json
.
{
"mcpServers": {
"adls2": {
"command": "adls2-mcp-server",
"env": {
"LOG_LEVEL": "DEBUG",
"UPLOAD_ROOT": "/path/to/store/uploads",
"DOWNLOAD_ROOT": "/path/to/store/downloads",
"AZURE_STORAGE_ACCOUNT_NAME": "your-azure-adls2-storage-account-name",
"READ_ONLY_MODE": "false"
}
}
}
}
The following is a table of available environment configuration variables:
Variable | Description | Default |
---|---|---|
LOG_LEVEL |
Logging level | INFO |
UPLOAD_ROOT |
Root directory for file uploads | ./uploads |
DOWNLOAD_ROOT |
Root directory for file downloads | ./downloads |
AZURE_STORAGE_ACCOUNT_NAME |
Azure ADLS2 storage account name | None |
AZURE_STORAGE_ACCOUNT_KEY |
Azure ADLS2 storage account key (optional) | None |
READ_ONLY_MODE |
Whether the server should operate in read-only mode | true |
If AZURE_STORAGE_ACCOUNT_KEY
is not set, the server will attempt to authenticate using Azure CLI credentials. Ensure you have logged in with Azure CLI before running the server:
az login
2 - Restart Claude Desktop.
Available Tools 🔧
Filesystem (container) Operations
list_filesystems
- List all filesystems in the storage accountcreate_filesystem
- Create a new filesystemdelete_filesystem
- Delete an existing filesystem
File Operations
upload_file
- Upload a file to ADLS2download_file
- Download a file from ADLS2file_exists
- Check if a file existsrename_file
- Rename/move a fileget_file_properties
- Get file propertiesget_file_metadata
- Get file metadataset_file_metadata
- Set file metadataset_file_metadata_json
- Set multiple metadata key-value pairs using JSON
Directory Operations
create_directory
- Create a new directorydelete_directory
- Delete a directoryrename_directory
- Rename/move a directorydirectory_exists
- Check if a directory existsdirectory_get_paths
- Get all paths under the specified directory
Development 💻
Local Development Setup
1 - Clone the repository:
git clone https://github.com/erikhoward/adls2-mcp-server.git
cd adls2-mcp-server
2 - Create and activate virtual environment:
Linux/macOS:
python -m venv .venv
source .venv/bin/activate
Windows:
.venv\Scripts\activate
3 - Install dependencies:
uv pip install -e ".[dev]"
4 - Copy and configure environment variables:
cp .env.example .env
Edit .env with your settings.
AZURE_STORAGE_ACCOUNT_NAME=your_azure_adls2_storage_account_name
AZURE_STORAGE_ACCOUNT_KEY=your_azure_adls2_storage_key (optional)
DOWNLOAD_ROOT=/path/to/download/folder
UPLOAD_ROOT=/path/to/upload/folder
READ_ONLY_MODE=True
LOG_LEVEL=INFO
If AZURE_STORAGE_ACCOUNT_KEY
is not set, the server will attempt to authenticate using Azure CLI credentials. Ensure you have logged in with Azure CLI before running the server:
az login
5 - Claude Desktop Configuration
Open claude_desktop_config.json
and add the following configuration.
On MacOs, the file is located here:
~/Library/Application Support/Claude Desktop/claude_desktop_config.json
.
On Windows, the file is located here:
%APPDATA%\Claude Desktop\claude_desktop_config.json
.
{
"mcpServers": {
"adls2": {
"command": "uv",
"args": [
"--directory",
"/path/to/adls2-mcp-server/repo",
"run",
"adls2-mcp-server"
],
"env": {
"LOG_LEVEL": "DEBUG",
"UPLOAD_ROOT": "/path/to/store/uploads",
"DOWNLOAD_ROOT": "/path/to/store/downloads",
"AZURE_STORAGE_ACCOUNT_NAME": "your-azure-adls2-storage-account-name",
"READ_ONLY_MODE": "false"
}
}
}
}
6 - Restart Claude Desktop.
Contributions 🤝
Contributions are welcome! Please feel free to submit a Pull Request.
- Fork the repository
- Create your feature branch (
git checkout -b feature/AmazingFeature
) - Commit your changes (
git commit -m '✨ Add some AmazingFeature'
) - Push to the branch (
git push origin feature/AmazingFeature
) - Open a Pull Request
License ⚖️
Licensed under MIT - see LICENSE.md file.
This is not an official Microsoft product.
推荐服务器
Google Drive MCP Server
启用与 Google Drive 的集成,用于列出、读取和搜索文件,支持各种文件类型,并为 Google Workspace 文件提供自动导出功能。

AWS MCP Server
一个模型上下文协议服务器实现,使 Claude 能够通过自然语言命令在 S3 和 DynamoDB 服务上执行 AWS 操作。
Google Drive MCP Server
与 Google Drive 集成,以实现文件列表、搜索和读取,以及 Google 表格的读取和写入。
Azure MCP Server
这个服务器实现了模型上下文协议,可以与 Azure Blob 存储和 Cosmos DB 无缝交互,从而实现操作的自动日志记录和审计跟踪。
Box MCP Server
Box MCP 服务器方便使用开发者令牌认证在 Box 中搜索和读取 PDF 和 Word 文件。
MCP JSON Document Collection Server
一个模型上下文协议服务器,允许创建和管理多个 Fireproof JSON 数据库,具备 CRUD 操作、查询功能以及云同步能力,以便与他人共享数据库。 (Breakdown of the translation:) * **一个模型上下文协议服务器 (Yī gè móxíng shàngxiàwén xiéyì fúwùqì):** A Model Context Protocol server * **允许 (yǔnxǔ):** allows * **创建和管理 (chuàngjiàn hé guǎnlǐ):** creation and management * **多个 (duō gè):** multiple * **Fireproof JSON 数据库 (Fireproof JSON shùjùkù):** Fireproof JSON databases * **具备 (jùbèi):** with, possesses * **CRUD 操作 (CRUD cāozuò):** CRUD operations (Create, Read, Update, Delete) * **查询功能 (cháxún gōngnéng):** querying capabilities * **以及 (yǐjí):** and * **云同步能力 (yún tóngbù nénglì):** cloud synchronization capability * **以便 (yǐbiàn):** in order to * **与他人共享 (yǔ tārén gòngxiǎng):** sharing with others * **数据库 (shùjùkù):** databases

Unstructured Document Processor MCP
一个模型上下文协议服务器,它使大型语言模型能够从各种文件格式的非结构化文档中提取和使用内容。
Firebase MCP Server
一个提供统一接口以与 Firebase 服务(包括身份验证、Firestore 和存储)交互的服务器。
MCP Server for Apache OpenDAL™
一个模型上下文协议服务器,通过 Apache OpenDAL™ 提供对包括 S3、Azure Blob 存储和 Google Cloud Storage 在内的多个存储服务的无缝访问。
Cloudinary MCP Server
这个服务器提供了一些工具,可以使用 Claude/Cline 直接将图片和视频上传到 Cloudinary,从而方便资源管理,并提供可定制的选项,例如资源类型和公共 ID。