Coreshub MCP Server
About
mcp server for coreshub: https://coreshub.cn/
Details
- Author
- coreshub
- GitHub stars
- 10
- Downloads
- 201
- Categories
- Other
Jump to
- Query EPFS file systems and their billing info
- Retrieve container instance details with filtering
- Get SSH connection info for specific instances
- List distributed training tasks with time ranges
- Fetch detailed logs for distributed training jobs
- Query inference services and their logs
- Extensible architecture for custom tools and prompts
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
Coreshub MCP ServerCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
Install via uvx coreshub-mcp-server or clone from GitHub and run with uv run coreshub-mcp-server. Configure environment variables QY_ACCESS_KEY_ID, QY_SECRET_ACCESS_KEY, and CORESHUB_USER_ID with your Jishi Computing credentials. In Cherry Studio, add an MCP server with stdio type and the appropriate command for your OS. For CLI usage, run uv run src/coreshub_mcp_server with optional flags like --debug.
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"coreshub mcp server": {
"mcp-server-coreshub": {
"command": "uv",
"args": [
"run",
"src/coreshub_mcp_server"
]
}
}
}
}
McpServers
{
"mcp-server-coreshub": {
"command": "uv",
"args": [
"run",
"src/coreshub_mcp_server"
]
}
}
Coreshub MCP Server
1、项目结构
src/coreshub_mcp_server/
├── plugins/ # 插件目录,所有工具和提示插件
├── utils/ # 工具函数
│ └── signature.py # 签名工具函数
├── base_plugin.py # 工具和提示基类
├── settings.py # 配置管理
└── server.py # MCP服务器实现
> ##### 开始之前请确保安装好 python 和 uv
2、运行
场景一:在Cherry Studio中运行
> 注⚠️:为保证工具的正确调用,建议使用32B参数以上的模型服务
(1)一键拉取使用
> 在Cherry Studio的设置——MCP服务器——编辑MCP配置
推荐从pypi拉取
{
"mcpServers": {
"coreshub-mcp-server-来自pypi包": {
"type": "stdio",
"registryUrl": "http://mirrors.aliyun.com/pypi/simple/",
"command": "uvx",
"args": [
"coreshub-mcp-server"
],
"env": {
"QY_ACCESS_KEY_ID": "基石智算的AK",
"QY_SECRET_ACCESS_KEY": "基石智算的SK",
"CORESHUB_USER_ID": "基石智算的账户ID"
}
}
}
}
或者从github拉取
{
"mcpServers": {
"coreshub-mcp-server": {
"type": "stdio",
"registryUrl": "http://mirrors.aliyun.com/pypi/simple/",
"command": "uvx",
"args": [
"--from",
"git+https://github.com/coreshub/mcp-server-coreshub",
"coreshub-mcp-server"
],
"env": {
"QY_ACCESS_KEY_ID": "基石智算的AK",
"QY_SECRET_ACCESS_KEY": "基石智算的SK",
"CORESHUB_USER_ID": "基石智算的账户ID"
}
}
}
}
(2)从github下载到本地后使用
> 在Cherry Studio的设置——MCP服务器——添加服务器,进入编辑模式
对于macOS系统:
类型选择:
stdio
命令填写:
sh
参数填写:
-c
cd 项目根目录路径 && uv run coreshub-mcp-server
环境变量填写:
QY_ACCESS_KEY_ID=基石智算的AK
QY_SECRET_ACCESS_KEY=基石智算的SK
CORESHUB_USER_ID=基石智算的账户ID
对于windows系统:
类型选择:
stdio
命令填写:
cmd
参数填写:
/c
cd 项目根目录路径 && uv run coreshub-mcp-server
环境变量填写:
QY_ACCESS_KEY_ID=基石智算的AK
QY_SECRET_ACCESS_KEY=基石智算的SK
CORESHUB_USER_ID=基石智算的账户ID
场景二:命令行操作(需实现client)
(1)首先配置环境变量
可以在代码中配置,在settings.py中配置
class Settings:
access_key = os.getenv("QY_ACCESS_KEY_ID", "基石智算的AK")
secret_key = os.getenv("QY_SECRET_ACCESS_KEY", "基石智算的SK")
user_id = os.getenv("CORESHUB_USER_ID", "基石智算的账户ID")
或者在本机系统环境变量配置
export QY_ACCESS_KEY_ID="基石智算的AK"
export QY_SECRET_ACCESS_KEY="基石智算的SK"
export CORESHUB_USER_ID="基石智算的账户ID"
(2)在项目根目录使用 uv检查服务状态
uv run src/coreshub_mcp_server
命令行参数
- --debug: 启用调试模式,输出详细日志
- --list-plugins: 列出所有已加载的插件
- --log-file: 指定日志文件路径
3、开发
1、添加新工具
在 src/coreshub_mcp_server/plugins 目录下创建新的Python文件,然后实现 BaseTool 和/或 BasePrompt
的子类。工具和提示现在是分离的概念,可以根据需要只实现其中一种或两种。
(1)工具实现示例:
```python
from coreshub_mcp_server.base_plugin import BaseTool
class MyTool(BaseTool):
tool_name = "my_tool"
tool_description = "我的自定义工具"
@staticmethod
def model_json_schema():
return {
"type": "object",
"properties": {
"param": {
"type": "string",
"description": "参数描述"
}
}
}
async def execute_tool(self, arguments):
# 实现工具逻辑
pass
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.



