Mage MCP Server

Mage.ai的MCP集成項目，為AI助手提供瀏覽、編輯和執行Mage數據管道的功能，支持管道創建、代碼修改和搜索等操作。

開發者工具人工智能聊天機器人 #數據管道 #AI集成 #代碼編輯 #管道執行 .Python

評分 : 2.5分

下載量 : 5.9K

更新時間 : 2025-04-28

打開站點

What is Mage.ai MCP Integration?

This service connects AI assistants to Mage.ai data pipelines, allowing you to browse, edit, and execute pipelines through natural language commands.

How to use Mage.ai MCP?

Simply install the server, connect it to your Mage.ai instance, and start interacting with your pipelines through supported AI assistants.

When to use this service?

Ideal for data teams who want to manage pipelines more efficiently using AI assistance, especially for routine pipeline operations and quick edits.

Key Features

Pipeline Exploration

Browse all your Mage.ai pipelines and their components with simple commands

Code Editing

View and modify block code directly through the AI interface

Pipeline Execution

Run entire pipelines or individual blocks with a single command

Smart Search

Find blocks by name, type or content using natural language queries

Pipeline Creation

Create new pipelines and blocks from scratch with AI assistance

Advantages

Saves time by allowing natural language interaction with pipelines

Reduces need for direct coding for routine pipeline operations

Provides quick access to pipeline information without navigating UI

Enables AI-assisted code editing and suggestions

Limitations

Requires initial setup and configuration

Limited to supported AI assistants (currently Claude)

Complex pipeline modifications may still require manual coding

Dependent on Mage.ai API availability

Getting Started

Install the server

Clone the repository and install required dependencies

Configure your connection

Set up the connection to your Mage.ai instance by creating a .env file

Start the server

Run the server with your preferred configuration

Connect to your AI assistant

Install the server with your AI assistant (like Claude Desktop) using the mcp CLI

Example Interactions

Exploring Pipelines

Ask the AI assistant to show you available pipelines

Editing Pipeline Code

Request changes to a specific block in your pipeline

Creating New Pipeline

Ask the assistant to create a new data processing pipeline

Frequently Asked Questions

What AI assistants are supported?

Is my pipeline data secure?

Can I undo changes made through the AI?

What programming languages are supported?

Additional Resources

Mage.ai Official Documentation

Complete documentation for Mage.ai platform

MCP Protocol Specification

Technical details about Model Context Protocol

Example Pipelines Repository

Sample Mage.ai pipelines for inspiration

🚀 Mage.ai 數據管道與模型上下文協議 (MCP) 集成

這是一個專為 Mage.ai 設計的全面模型上下文協議 (MCP) 集成方案，能讓 AI 助手（例如 Anthropic 的 Claude）高效地與 Mage 數據管道進行交互。

✨ 主要特性

瀏覽管道：可探索所有 Mage.ai 管道及其組件。
查看和編輯代碼：能閱讀並修改塊代碼。
執行管道：可運行整個管道或單獨的塊。
搜索功能：能按名稱、類型或內容查找塊。
管道創建：支持從零開始新建管道和塊。

📦 安裝指南

# 克隆倉庫
git clone https://github.com/yourusername/mage-mcp.git
cd mage-mcp

# 安裝依賴項
pip install -r requirements.txt

📚 詳細文檔

配置說明

服務器可通過環境變量或命令行參數進行配置：

MAGE_API_URL：Mage.ai API 的 URL 地址（默認值：http://localhost:6789/api/）
MAGE_API_KEY：用於身份驗證的 API 密鑰（如有需要）

創建一個 .env 文件以實現持久化配置：

MAGE_API_URL=http://localhost:6789/api/
MAGE_API_KEY=your_api_key_if_needed

使用方法

啟動服務器

使用提供的腳本來啟動 MCP 服務器：

# 使用標準輸入輸出傳輸（默認）
python run_server.py

# 使用 SSE 傳輸
python run_server.py --transport sse --host localhost --port 3000

# 其他選項
python run_server.py --api-url http://your-mage-instance:6789/api/ --api-key your_key --debug

與 Claude Desktop 集成安裝

為了獲得最佳體驗，直接通過 Claude Desktop 安裝此服務器：

# 使用 mcp CLI
mcp install run_server.py --name "Mage.ai 管理器"

# 帶環境變量配置
mcp install run_server.py -v MAGE_API_URL=http://localhost:6789/api/ -v MAGE_API_KEY=your_key

開發模式

用於測試和開發：

mcp dev run_server.py

MCP 資源與工具

資源地址

mage://pipelines - 列出所有管道
mage://pipelines/{pipeline_uuid} - 獲取管道詳細信息
mage://pipelines/{pipeline_uuid}/blocks - 獲取管道中的塊
mage://pipelines/{pipeline_uuid}/blocks/{block_id} - 獲取特定塊的詳細信息
mage://pipelines/{pipeline_uuid}/execute - 執行管道
mage://pipelines/{pipeline_uuid}/execute/{execution_id} - 獲取執行狀態

工具函數

list_all_pipelines()：返回所有管道的列表。
get_pipeline_details(pipeline_uuid)：獲取指定管道的詳細信息。
execute_pipeline(pipeline_uuid, parameters)：執行管道，可傳入參數。
search_blocks(query)：按查詢內容搜索塊。

💻 使用示例

基礎用法

# 創建新管道
pipeline = create_pipeline("數據分析管道", "數據處理和分析")

# 添加塊到管道
add_block_to_pipeline(pipeline.uuid, "數據清洗", "清洗數據步驟")

# 執行管道
execution_id = execute_pipeline(pipeline.uuid, {"input_data": "raw_data.csv"})

# 獲取執行狀態
status = get_execution_status(pipeline.uuid, execution_id)
print(status)  # 輸出執行結果