McpStore

https://github.com/whillhill/mcpstore
Admin
McpStore is an open-source, high-quality MCP service management tool designed to simplify tool usage...
Tokens:8,316
Snippets:37
Trust Score:4.1
License:MIT
Update:1 year ago
Show doc for...
Tokens:
Raw
### Install McpStore

Source: https://github.com/whillhill/mcpstore/blob/main/README_zh.md

Installs the McpStore package using pip. This is the primary method for getting started with the library.

```bash
pip install mcpstore
```

--------------------------------

### Install McpStore

Source: https://github.com/whillhill/mcpstore/blob/main/README.md

Installs the McpStore package using pip. This is the primary method for getting started with the library.

```bash
pip install mcpstore
```

--------------------------------

### Start MCPStore RESTful API

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Installs and runs the MCPStore RESTful API service. Provides access to 38 API endpoints upon startup.

```bash
pip install mcpstore
mcpstore run api
```

--------------------------------

### Run McpStore API

Source: https://github.com/whillhill/mcpstore/blob/main/README.md

Starts the McpStore RESTful API service from the command line. This command requires the McpStore package to be installed.

```bash
pip install mcpstore
mcpstore run api
```

--------------------------------

### Start McpStore API Server

Source: https://github.com/whillhill/mcpstore/blob/main/README.md

Initializes and starts the McpStore API server. This allows for managing MCP services through a web interface or other clients.

```python
from mcpstore import MCPStore
prod_store = MCPStore.setup_store()
prod_store.start_api_server(host='0.0.0.0', port=18200)
```

--------------------------------

### Start McpStore API Server

Source: https://github.com/whillhill/mcpstore/blob/main/README_zh.md

Sets up and starts the McpStore backend API server. This allows for managing MCP services and tools via API calls.

```python
from mcpstore import MCPStore
prod_store = MCPStore.setup_store()
prod_store.start_api_server(host='0.0.0.0', port=18200)
```

--------------------------------

### LangChain Integration Example

Source: https://github.com/whillhill/mcpstore/blob/main/README_zh.md

An example of integrating McpStore with LangChain to create an agent that can use tools. It sets up an LLM, defines a prompt, and creates an agent executor.

```python
from langchain.agents import create_tool_calling_agent, AgentExecutor
from langchain_core.prompts import ChatPromptTemplate
from langchain_openai import ChatOpenAI
from mcpstore import MCPStore
# ===
store = MCPStore.setup_store()
store.for_store().add_service({"name":"mcpstore-wiki","url":"https://mcpstore.wiki/mcp"})
tools = store.for_store().for_langchain().list_tools()
# ===
llm = ChatOpenAI(
    temperature=0, model="deepseek-chat",
    openai_api_key="****",
    openai_api_base="https://api.deepseek.com"
)
prompt = ChatPromptTemplate.from_messages([
    ("system", "你是一个助手，回答的时候带上表情"),
    ("human", "{input}"),
    ("placeholder", "{agent_scratchpad}"),
])
agent = create_tool_calling_agent(llm, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
# ===
query = "北京的天气怎么样？"
print(f"\n   🤔: {query}")
response = agent_executor.invoke({"input": query})
print(f"   🤖 : {response['output']}")
```

--------------------------------

### LangChain Integration Example

Source: https://github.com/whillhill/mcpstore/blob/main/README.md

Shows how to integrate McpStore tools with a LangChain Agent for natural language interaction. It sets up an LLM, defines a prompt, creates an agent, and executes a query.

```python
from langchain.agents import create_tool_calling_agent, AgentExecutor
from langchain_core.prompts import ChatPromptTemplate
from langchain_openai import ChatOpenAI
from mcpstore import MCPStore
# ===
store = MCPStore.setup_store()
store.for_store().add_service({"name":"mcpstore-wiki","url":"https://mcpstore.wiki/mcp"})
tools = store.for_store().for_langchain().list_tools()
# ===
llm = ChatOpenAI(
    temperature=0, model="deepseek-chat",
    openai_api_key="****",
    openai_api_base="https://api.deepseek.com"
)
prompt = ChatPromptTemplate.from_messages([
    ("system", "You are an assistant, respond with emojis"),
    ("human", "{input}"),
    ("placeholder", "{agent_scratchpad}"),
])
agent = create_tool_calling_agent(llm, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
# ===
query = "What's the weather like in Beijing?"
print(f"\n   🤔: {query}")
response = agent_executor.invoke({"input": query})
print(f"   🤖 : {response['output']}")
```

--------------------------------

### Install McpStore

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Provides the command to install the McpStore library using pip, which is the standard package installer for Python.

```bash
pip install mcpstore
```

--------------------------------

### Multi-tenant MCPStore Setup

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Demonstrates setting up separate MCPStore instances for different tenants using distinct configuration files.

```python
# Tenant A configuration
tenant_a_store = MCPStore.setup_store(
    mcp_config_file="tenant_a_mcp.json",
    debug=False
)

# Tenant B configuration
tenant_b_store = MCPStore.setup_store(
    mcp_config_file="tenant_b_mcp.json",
    debug=False
)

# Accessing services for each tenant
tenant_a_tools = tenant_a_store.for_store().list_tools()
tenant_b_tools = tenant_b_store.for_store().list_tools()
```

--------------------------------

### JSON File Format Examples

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Examples of JSON file formats for service configuration, including 'mcpServers' definitions and individual service definitions.

```json
{
  "mcpServers": {
    "mcpstore-wiki": {
      "url": "http://mcpstore.wiki/mcp"
    },
    "howtocook": {
      "command": "npx",
      "args": [
        "-y",
        "howtocook-mcp"
      ]
    }
  }
}

{
    "name": "mcpstore-wiki",
    "url": "http://mcpstore.wiki/mcp"
}
```

--------------------------------

### Integrate MCPStore Tools with LangChain Agent

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Shows a complete example of integrating MCPStore tools into a LangChain Agent. It covers initializing the store, adding services, retrieving LangChain-compatible tools, setting up a LangChain agent with an LLM and prompt, and executing a query.

```python
from langchain.agents import create_tool_calling_agent, AgentExecutor
from langchain_core.prompts import ChatPromptTemplate
from langchain_openai import ChatOpenAI
from mcpstore import MCPStore

store = MCPStore.setup_store()
store.for_store().add_service({"name":"mcpstore-wiki","url":"http://mcpstore.wiki/mcp"})

tools = store.for_store().to_langchain_tools()

llm = ChatOpenAI(
    temperature=0, model="deepseek-chat",
    openai_api_key="sk-****",
    openai_api_base="https://api.deepseek.com"
)

prompt = ChatPromptTemplate.from_messages([
    ("system", "You are an assistant, answer with emojis"),
    ("human", "{input}"),
    ("placeholder", "{agent_scratchpad}"),
])

agent = create_tool_calling_agent(llm, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)

query = "How's the weather in Beijing?"
print(f"\n   🤔: {query}")
response = agent_executor.invoke({"input": query})
print(f"   🤖 : {response['output']}")
```

--------------------------------

### Multi-Agent Isolation Example

Source: https://github.com/whillhill/mcpstore/blob/main/README_zh.md

Demonstrates how to create isolated toolsets for different agents using McpStore's `for_agent` context manager. This ensures that each agent has its own dedicated set of tools.

```python
# 初始化Store
store = MCPStore.setup_store()

# 为“知识管理Agent”分配专用的Wiki工具
# 该操作在"knowledge" agent的私有上下文中进行
agent_id1 = "my-knowledge-agent"
knowledge_agent_context = store.for_agent(agent_id1).add_service(
    {"name": "mcpstore-wiki", "url": "http://mcpstore.wiki/mcp"}
)

# 为“开发支持Agent”分配专用的开发工具
# 该操作在"development" agent的私有上下文中进行
agent_id2 = "my-development-agent"
dev_agent_context = store.for_agent(agent_id2).add_service(
    {"name": "mcpstore-demo", "url": "http://mcpstore.wiki/mcp"}
)

# 各Agent的工具集完全隔离，互不影响
knowledge_tools = store.for_agent(agent_id1).list_tools()
dev_tools = store.for_agent(agent_id2).list_tools()
```

--------------------------------

### Multi-Agent Isolation Example

Source: https://github.com/whillhill/mcpstore/blob/main/README.md

Demonstrates how to create isolated toolsets for different agents using McpStore's `for_agent` context manager. This ensures that each agent has its own dedicated set of tools.

```python
# Initialize Store
store = MCPStore.setup_store()

# Assign dedicated Wiki tools for "Knowledge Management Agent"
# This operation is performed in the private context of "knowledge" agent
agent_id1 = "my-knowledge-agent"
knowledge_agent_context = store.for_agent(agent_id1).add_service(
    {"name": "mcpstore-wiki", "url": "http://mcpstore.wiki/mcp"}
)

# Assign dedicated development tools for "Development Support Agent"
# This operation is performed in the private context of "development" agent
agent_id2 = "my-development-agent"
dev_agent_context = store.for_agent(agent_id2).add_service(
    {"name": "mcpstore-demo", "url": "http://mcpstore.wiki/mcp"}
)

# Each Agent's toolset is completely isolated without interference
knowledge_tools = store.for_agent(agent_id1).list_tools()
dev_tools = store.for_agent(agent_id2).list_tools()
```

--------------------------------

### Basic McpStore Usage

Source: https://github.com/whillhill/mcpstore/blob/main/README_zh.md

Demonstrates basic usage of the McpStore SDK, including adding a service and listing available tools.

```python
store = MCPStore.setup_store()
store.for_store().add_service({"name":"mcpstore-wiki","url":"https://mcpstore.wiki/mcp"})
tools = store.for_store().list_tools()
# store.for_store().use_tool(tools[0].name, {"query":'hi!'})
```

--------------------------------

### Basic McpStore Usage

Source: https://github.com/whillhill/mcpstore/blob/main/README.md

Demonstrates basic operations within McpStore, including adding a service and listing available tools.

```python
store = MCPStore.setup_store()
store.for_store().add_service({"name":"mcpstore-wiki","url":"https://mcpstore.wiki/mcp"})
tools = store.for_store().list_tools()
# store.for_store().use_tool(tools[0].name, {"query":'hi!'})
```

--------------------------------

### MCPStore Initialization and Tool Call

Source: https://github.com/whillhill/mcpstore/wiki/Home

Demonstrates how to initialize MCPStore, register a service, and call a tool. This involves creating an MCPStore instance, registering a service like 'filesystem' with its command and arguments, and then invoking a tool such as 'read_file' with specific parameters.

```python
from mcpstore import MCPStore

# 创建 Store 实例
store = MCPStore()

# 注册服务
store.register_service("filesystem", {
    "command": "npx",
    "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/files"]
})

# 调用工具
result = store.call_tool("read_file", {"path": "example.txt"})
print(result)
```

--------------------------------

### Initialize MCPStore and Add Service

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Demonstrates the basic steps to initialize MCPStore, set up a store, and register an external MCP service. This is the foundational step for integrating MCP capabilities.

```python
from mcpstore import MCPStore
store = MCPStore.setup_store()
store.for_store().add_service({"name":"mcpstore-wiki","url":"http://mcpstore.wiki/mcp"})
```

--------------------------------

### MCPStore Parameter Details

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Details the parameters for MCPStore's setup_store method, including default values and behavior.

```APIDOC
MCPStore.setup_store(mcp_config_file: str = None, debug: bool = False) -> MCPStore

  Parameters:
    mcp_config_file (str, optional): Path to a custom MCP configuration file. 
      If not specified, defaults to 'src/mcpstore/data/mcp.json'.
    debug (bool, optional): Enables debug logging mode. Defaults to False.
      When True, log level is DEBUG and includes timestamps and logger names.
      When False, log level is ERROR and format is '%%(levelname)s - %%(message)s'.

  Returns:
    MCPStore: An initialized MCPStore instance.
```

--------------------------------

### MCPStore Initialization

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Initializes an MCPStore instance. Supports custom configuration file paths and debug mode for flexible configuration.

```python
from mcpstore import MCPStore

# Initialize with default configuration
store = MCPStore.setup_store()

# Initialize with a custom configuration file
store_custom = MCPStore.setup_store(mcp_config_file="path/to/your/mcp.json")

# Initialize with debug mode enabled
store_debug = MCPStore.setup_store(debug=True)
```

--------------------------------

### MCPStore Service Registration

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Shows how to register services with MCPStore, either by loading from a configuration file or by providing service details directly.

```python
store = MCPStore.setup_store()

# Register all services from mcp.json
store.for_store().add_service()

# Register a service via URL
store.for_store().add_service({
    "name": "mcpstore-wiki",
    "url": "http://mcpstore.wiki/mcp",
    "transport": "streamable-http"
})

# Register a local command-based service
store.for_store().add_service({
    "name": "local_assistant",
    "command": "python",
    "args": ["./assistant_server.py"],
    "env": {"DEBUG": "true", "API_KEY": "your_key"},
    "working_dir": "/path/to/service"
})
```

--------------------------------

### Direct Tool Usage without LangChain

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Illustrates how to use MCPStore tools directly without relying on LangChain. This involves initializing the store, adding a service, listing available tools, and then invoking a specific tool with provided arguments.

```python
from mcpstore import MCPStore

store = MCPStore.setup_store()
store.for_store().add_service({"name":"mcpstore-wiki","url":"http://mcpstore.wiki/mcp"})

tools = store.for_store().list_tools()
print(store.for_store().use_tool(tools[0].name,{"query":'Beijing'}))
```

--------------------------------

### Add Executable File Service

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Registers an executable file as a service. Requires 'name' and 'command'. 'args' are optional command-line arguments.

```python
store.for_store().add_service({
    "name": "binary_service",
    "command": "./mcp_server",
    "args": ["--config", "config.json"]
})
```

--------------------------------

### MCP Configuration File Format

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Defines the structure for MCPStore configuration files, supporting URL-based and command-based service definitions.

```json
{
  "mcpServers": {
    "mcpstore-wiki": {
      "url": "http://mcpstore.wiki/mcp"
    },
    "howtocook": {
      "command": "npx",
      "args": [
        "-y",
        "howtocook-mcp"
      ]
    }
  }
}
```

--------------------------------

### Register MCPConfig Dictionary

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Registers services using an MCPConfig dictionary format. Services can be defined with URLs or commands to execute.

```python
store.for_store().add_service({
  "mcpServers": {
    "mcpstore-wiki": {
      "url": "http://mcpstore.wiki/mcp"
    },
    "howtocook": {
      "command": "npx",
      "args": [
        "-y",
        "howtocook-mcp"
      ]
    }
  }
})
```

--------------------------------

### Add Node.js Service

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Registers a Node.js service with specified command, arguments, and environment variables. Requires 'name', 'command', and 'args'. 'env' is optional.

```python
store.for_store().add_service({
    "name": "node_service",
    "command": "node",
    "args": ["server.js", "--port", "8080"],
    "env": {"NODE_ENV": "production"}
})
```

--------------------------------

### McpStore API Endpoints

Source: https://github.com/whillhill/mcpstore/blob/main/README.md

Lists the main RESTful API endpoints provided by McpStore for service and tool management, as well as monitoring.

```APIDOC
# Service Management
POST /for_store/add_service          # Add service
GET  /for_store/list_services        # Get service list
POST /for_store/delete_service       # Delete service

# Tool Operations
GET  /for_store/list_tools           # Get tool list
POST /for_store/use_tool             # Execute tool

# Monitoring & Statistics
GET  /for_store/get_stats            # System statistics
GET  /for_store/health               # Health check
```

--------------------------------

### MCPStore Loading Screen CSS

Source: https://github.com/whillhill/mcpstore/blob/main/vue/index.html

Defines the styles for the loading screen overlay, including its position, background gradient, centering of content, and a spinning animation for the loader.

```css
#loading {
  position: fixed;
  top: 0;
  left: 0;
  width: 100%;
  height: 100%;
  background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
  display: flex;
  justify-content: center;
  align-items: center;
  z-index: 9999;
  transition: opacity 0.3s ease;
}

.loading-content {
  text-align: center;
  color: white;
}

.loading-spinner {
  width: 50px;
  height: 50px;
  border: 3px solid rgba(255, 255, 255, 0.3);
  border-top: 3px solid white;
  border-radius: 50%;
  animation: spin 1s linear infinite;
  margin: 0 auto 20px;
}

@keyframes spin {
  0% {
    transform: rotate(0deg);
  }
  100% {
    transform: rotate(360deg);
  }
}

.loading-text {
  font-size: 18px;
  font-weight: 500;
  margin-bottom: 10px;
}

.loading-subtitle {
  font-size: 14px;
  opacity: 0.8;
}
```

--------------------------------

### Monitoring System API Endpoints

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Provides endpoints for monitoring system status, configuring monitoring, and restarting monitoring tasks.

```APIDOC
GET  /monitoring/status              # Get monitoring status
POST /monitoring/config              # Update monitoring configuration
POST /monitoring/restart             # Restart monitoring tasks
```

--------------------------------

### Register Service Name List

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Registers multiple services by their names from existing configurations. Can register a list of names or a single name.

```python
# Register specified services
store.for_store().add_service(['mcpstore-wiki', 'howtocook'])

# Register single service
store.for_store().add_service(['howtocook'])
```

--------------------------------

### Multi-Agent Context Isolation

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Demonstrates how to create isolated contexts for different agents using `store.for_agent()`. This allows each agent to have its own dedicated set of MCP tools without interference from other agents.

```python
# Initialize Store
store = MCPStore.setup_store()

# Assign dedicated Wiki tools to "Knowledge Management Agent"
agent_id1 = "my-knowledge-agent"
knowledge_agent_context = store.for_agent(agent_id1).add_service(
    {"name": "mcpstore-wiki", "url": "http://mcpstore.wiki/mcp"}
)

# Assign dedicated development tools to "Development Support Agent"
agent_id2 = "my-development-agent"
dev_agent_context = store.for_agent(agent_id2).add_service(
    {"name": "mcpstore-demo", "url": "http://mcpstore.wiki/mcp"}
)

# Each Agent's tool set is completely isolated
knowledge_tools = store.for_agent(agent_id1).list_tools()
dev_tools = store.for_agent(agent_id2).list_tools()
```

--------------------------------

### McpStore API Endpoints

Source: https://github.com/whillhill/mcpstore/blob/main/README_zh.md

Lists the primary RESTful API endpoints provided by McpStore for managing services and tools, as well as for system monitoring.

```APIDOC
# 服务管理
POST /for_store/add_service          # 添加服务
GET  /for_store/list_services        # 获取服务列表
POST /for_store/delete_service       # 删除服务

# 工具操作
GET  /for_store/list_tools           # 获取工具列表
POST /for_store/use_tool             # 执行工具

# 监控统计
GET  /for_store/get_stats            # 系统统计
GET  /for_store/health               # 健康检查
```

--------------------------------

### Register JSON File Configuration

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Registers service configurations by reading from an external JSON file. If both 'config' and 'json_file' are provided, 'json_file' takes precedence.

```python
# Read configuration from file
store.for_store().add_service(json_file="./demo_config.json")

# Specify both config and json_file (json_file takes priority)
store.for_store().add_service(
    config={"name": "backup"},
    json_file="./demo_config.json"  # This will be used ⚡
)
```

--------------------------------

### General API Endpoint

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Provides a general endpoint for cross-context service queries by service name.

```APIDOC
GET  /services/{name}                # Cross-context service query
```

--------------------------------

### Store Level API Endpoints

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Provides endpoints for managing services, tools, and system statistics at the store level. Includes operations for adding, listing, deleting, updating, and restarting services, as well as executing tools and retrieving system health.

```APIDOC
# Service Management
POST /for_store/add_service          # Add service
GET  /for_store/list_services        # Get service list
POST /for_store/delete_service       # Delete service
POST /for_store/update_service       # Update service
POST /for_store/restart_service      # Restart service

# Tool Operations
GET  /for_store/list_tools           # Get tool list
POST /for_store/use_tool             # Execute tool

# Batch Operations
POST /for_store/batch_add_services   # Batch add
POST /for_store/batch_update_services # Batch update

# Monitoring & Statistics
GET  /for_store/get_stats            # System statistics
GET  /for_store/health               # Health check
```

--------------------------------

### MCPStore Loading Screen Removal Logic

Source: https://github.com/whillhill/mcpstore/blob/main/vue/index.html

Provides JavaScript functions to remove the loading screen. It includes an immediate removal function and logic to ensure removal upon DOMContentLoaded or after a 1-second timeout as a fallback.

```javascript
// 立即移除loading屏幕
const removeLoading = () => {
  const loading = document.getElementById('loading');
  if (loading) {
    loading.style.opacity = '0';
    setTimeout(() => {
      loading.remove();
    }, 50); // 极短延迟
  }
};

// 多种方式确保loading被移除
if (document.readyState === 'loading') {
  document.addEventListener('DOMContentLoaded', removeLoading);
} else {
  removeLoading();
}

// 备用方案：最多等待1秒
setTimeout(removeLoading, 1000);
```

--------------------------------

### Agent Level API Endpoints

Source: https://github.com/whillhill/mcpstore/blob/main/src/README.md

Provides endpoints for managing services at the agent level, supporting multi-tenant isolation. These endpoints mirror the functionality of the Store Level API but are scoped to a specific agent ID.

```APIDOC
# Fully corresponds to Store level, supports multi-tenant isolation
POST /for_agent/{agent_id}/add_service
GET  /for_agent/{agent_id}/list_services
# ... All Store level features are supported
```
Tokens:
Raw