OssHub
Provides unified access to multiple cloud object storage services (Huawei OBS, Alibaba OSS, AWS S3, MinIO) enabling AI assistants to list, search, retrieve, and manage unstructured data across different storage providers.
README
OssHub - Universal Object Storage MCP Server
OssHub is a Model Context Protocol (MCP) server that provides unified access to multiple cloud object storage services. It allows AI assistants like Claude to interact with unstructured data stored in cloud object storage.
Features
-
Multi-Provider Support: Connect to multiple cloud object storage services:
- Huawei Cloud OBS (华为云对象存储)
- Alibaba Cloud OSS (阿里云对象存储)
- AWS S3 / S3-Compatible (MinIO, Backblaze B2, DigitalOcean Spaces, etc.)
- More providers coming soon (Tencent Cloud COS, etc.)
-
Multi-Source Configuration: Connect to multiple storage sources simultaneously with unique identifiers
-
Built-in Tools:
list_buckets- List all storage bucketslist_objects- List objects in a bucket with filtering and paginationget_object- Retrieve object content (text as plain text, binary as base64)get_object_metadata- Get object metadata without downloading contentsearch_objects- Search objects with advanced filters (extension, size, date, pattern)
-
Flexible Configuration: TOML configuration file or environment variables
Installation
# Clone the repository
git clone https://github.com/your-org/osshub.git
cd osshub
# Install dependencies
npm install
# or
pnpm install
# Build
npm run build
# or
pnpm build
Quick Start
1. Create Configuration File
Create osshub.toml in your project directory:
[[sources]]
id = "my_obs"
type = "obs"
endpoint = "https://obs.cn-east-2.myhuaweicloud.com"
access_key = "your-access-key"
secret_key = "your-secret-key"
region = "cn-east-2"
2. Run the Server
# STDIO mode (for MCP clients like Claude Desktop)
npm start
# HTTP mode (for web-based MCP clients)
npm start -- --transport=http --port=8080
3. Configure MCP Client
Claude Desktop Configuration
Add to your Claude Desktop config (claude_desktop_config.json):
{
"mcpServers": {
"osshub": {
"command": "node",
"args": ["/path/to/osshub/dist/index.js"],
"env": {}
}
}
}
Cursor Configuration
Add to your Cursor MCP settings:
{
"mcpServers": {
"osshub": {
"command": "node",
"args": ["/path/to/osshub/dist/index.js", "--config=/path/to/osshub.toml"]
}
}
}
Configuration
TOML Configuration (Recommended)
# Huawei Cloud OBS
[[sources]]
id = "huawei_obs"
type = "obs"
endpoint = "https://obs.cn-east-2.myhuaweicloud.com"
access_key = "your-access-key"
secret_key = "your-secret-key"
region = "cn-east-2"
# Alibaba Cloud OSS
[[sources]]
id = "aliyun_oss"
type = "oss"
endpoint = "https://oss-cn-hangzhou.aliyuncs.com"
access_key = "your-access-key"
secret_key = "your-secret-key"
region = "cn-hangzhou"
# AWS S3 / MinIO
[[sources]]
id = "minio"
type = "s3"
endpoint = "http://localhost:9000"
access_key = "minioadmin"
secret_key = "minioadmin"
path_style = true
# Tool configurations (optional)
[[tools]]
name = "list_objects"
source = "huawei_obs"
max_keys = 500
[[tools]]
name = "get_object"
source = "huawei_obs"
max_size = 5242880 # 5MB
Environment Variables
For single-source configuration:
export OSSHUB_TYPE=obs
export OSSHUB_ENDPOINT=https://obs.cn-east-2.myhuaweicloud.com
export OSSHUB_ACCESS_KEY=your-access-key
export OSSHUB_SECRET_KEY=your-secret-key
export OSSHUB_REGION=cn-east-2
Built-in Tools
list_buckets
List all storage buckets in a source.
Tool: list_buckets (or list_buckets_{source_id})
Parameters: None
list_objects
List objects in a bucket with optional filtering.
Tool: list_objects (or list_objects_{source_id})
Parameters:
- bucket: (required) Bucket name
- prefix: (optional) Filter by key prefix
- delimiter: (optional) Delimiter for hierarchical listing
- max_keys: (optional) Maximum objects to return
- continuation_token: (optional) Pagination token
get_object
Retrieve object content.
Tool: get_object (or get_object_{source_id})
Parameters:
- bucket: (required) Bucket name
- key: (required) Object key
- max_size: (optional) Maximum content size in bytes
get_object_metadata
Get object metadata without downloading.
Tool: get_object_metadata (or get_object_metadata_{source_id})
Parameters:
- bucket: (required) Bucket name
- key: (required) Object key
search_objects
Search objects with advanced filters.
Tool: search_objects (or search_objects_{source_id})
Parameters:
- bucket: (required) Bucket name
- prefix: (optional) Key prefix filter
- suffix: (optional) Key suffix filter
- extensions: (optional) File extensions [".jpg", ".png"]
- pattern: (optional) Glob pattern
- min_size: (optional) Minimum file size
- max_size: (optional) Maximum file size
- modified_after: (optional) Modified after date (ISO 8601)
- modified_before: (optional) Modified before date (ISO 8601)
- max_results: (optional) Maximum results
Development
# Development mode with hot reload
npm run dev
# HTTP mode development
npm run dev:http
# Run tests
npm test
# Build
npm run build
Provider Support
| Provider | Type | Status | SDK |
|---|---|---|---|
| Huawei Cloud OBS | obs |
✅ Supported | esdk-obs-nodejs v3.25.6 |
| Alibaba Cloud OSS | oss |
✅ Supported | ali-oss |
| AWS S3 | s3 |
✅ Supported | @aws-sdk/client-s3 |
| MinIO | s3 |
✅ Supported (S3-Compatible) | @aws-sdk/client-s3 |
| Tencent Cloud COS | cos |
🚧 Coming Soon | - |
| Backblaze B2 | s3 |
✅ (S3-Compatible) | @aws-sdk/client-s3 |
| DigitalOcean Spaces | s3 |
✅ (S3-Compatible) | @aws-sdk/client-s3 |
License
MIT License
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。