Atlas MCP Server
Enterprise-Grade Agentic AI Pipeline for Modern IDEs
接入你的智能体
發布者自己的指令。不是 PlugMCP 端點。
用我們的金鑰與日誌託管一份副本。
npx -y [email protected]Claude Code 沒有本機指令的安裝連結。複製下方指令並在終端機中執行。
OLLAMA_BASE_URL=<value> OPENAI_API_KEY=<secret> ANTHROPIC_API_KEY=<secret> claude mcp add io.github.iamnishant51-atlas-pipeline -- npx -y atlas-pipeline-mcp@1.0.2在終端機中執行。
{ "mcpServers": { "io.github.iamnishant51-atlas-pipeline": { "command": "npx", "args": [ "-y", "[email protected]" ], "env": { "OLLAMA_BASE_URL": "<value>", "OPENAI_API_KEY": "<secret>", "ANTHROPIC_API_KEY": "<secret>" } } }}claude_desktop_config.json
遠端 URL 還是本機指令? 兩種安裝方式的差異
Advanced AI/ML-powered development tools with RAG search, ML bug prediction, intelligent code migration, and comprehensive test generation.
Installation • Features • Tools • Usage • Documentation
Atlas MCP Server transforms your IDE with intelligent development capabilities through a comprehensive analysis pipeline:
Intent → Context → Decomposition → Variants → Critique → Optimization
Natively integrated with Cursor, Windsurf, Claude Desktop, GitHub Copilot, and VS Code.
The latest version introduces six powerful tools designed to address complex development challenges:
New Advanced Analysis Tools:
Enhanced Development Experience:
Comprehensive suite for modern frontend development:
Expert-level guidance and analysis:
Advanced machine learning capabilities:
Essential development utilities:
Step 1: Install Globally
Open your terminal and run:
npm install -g atlas-pipeline-mcp
Step 2: Run Auto-Setup
Automatically configure your IDE (Cursor or VS Code):
atlas-mcp-setup
Step 3: Restart IDE
Restart your editor. The Atlas server will be automatically connected and ready to use.
Once installed, interact with your AI Assistant (Copilot Chat or Cursor Chat). The appropriate tool is automatically activated based on your request.
If you prefer running models locally (e.g., Ollama) or want to use your own API keys instead of your IDE's subscription, configure the server manually in your settings:
{
"atlas": {
"command": "npx",
"args": ["-y", "atlas-pipeline-mcp"],
"env": {
"OLLAMA_BASE_URL": "http://localhost:11434",
"OPENAI_API_KEY": "sk-...",
"ANTHROPIC_API_KEY": "sk-..."
}
}
}
Note: If no API keys are provided, Atlas defaults to Client Sampling mode, delegating generation to your IDE's AI capabilities.
We welcome contributions to improve the Atlas pipeline.
MIT License - see LICENSE file for details
Built by Nishant Unavane
🌐 Check out my portfolio: nishantunavane.qzz.io
For support, documentation, or questions, visit our GitHub repository.
7-stage AI pipeline MCP server: Intent→Context→Git→Decompose→Variants→Critique→Optimize
{ "mcpServers": { "io.github.iamnishant51-atlas-pipeline": { "command": "npx", "args": [ "-y", "[email protected]" ], "env": { "OLLAMA_BASE_URL": "<value>", "OPENAI_API_KEY": "<secret>", "ANTHROPIC_API_KEY": "<secret>" } } }}工具
尚無工具清單。遠端伺服器會在首次健康檢查後取得工具清單。
常見問題
在本頁複製適用於 Cursor、Claude Desktop、Claude Code 或 VS Code 的 stdio 設定。它使用發布者的安裝方式,而不是 PlugMCP 代理伺服器。
發布者列出了以下環境變數:OLLAMA_BASE_URL, OPENAI_API_KEY, ANTHROPIC_API_KEY。
Atlas Pipeline 透過 stdio 在本機執行。
上次檢查時間:2026-11-09。 7-stage AI pipeline MCP server: Intent→Context→Git→Decompose→Variants→Critique→Optimize