StackMCP

AI/ML Engineer Stack for Claude Desktop

Advanced5 servers11.8K tokens

Configuration

{
  "mcpServers": {
    "exa-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "exa-mcp-server"
      ],
      "env": {
        "EXA_API_KEY": "YOUR_EXA_API_KEY"
      }
    },
    "context7-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@upstash/context7-mcp"
      ]
    },
    "memory-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-memory"
      ]
    },
    "sequential-thinking-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-sequential-thinking"
      ]
    },
    "filesystem-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-filesystem",
        "/path/to/allowed/directory"
      ]
    }
  }
}

Where to save

Paste the config above into:

~/Library/Application Support/Claude/claude_desktop_config.json

Environment Variables

Replace the YOUR_ placeholders with your actual values.

EXA_API_KEYrequired

Exa API key

Used by: Exa Search MCP

What’s in this stack

Exa Search MCP3 tools

AI-powered web search and crawling with Exa. Get semantically relevant results, extract content, and find similar pages.

AI-native search that understands semantic queries — essential for finding research papers, datasets, and technical references.

Context7 MCP2 tools

Pull up-to-date documentation and code examples for any library directly into your prompt context using the Context7 API.

Pull up-to-date library docs on demand. LLM training data goes stale fast; this keeps your AI assistant current on rapidly evolving ML frameworks.

Memory MCP6 tools

Enable persistent memory for Claude through a knowledge graph. Store and retrieve entities, relations, and observations across sessions.

Persist experiment results, model architectures, and design decisions across sessions. Your AI assistant remembers what worked and what didn't.

Sequential Thinking MCP1 tools

Enable structured, step-by-step reasoning for complex problem solving. Helps AI break down problems into logical sequences.

Complex ML pipelines need structured reasoning. This prevents your AI from jumping to conclusions on multi-step data processing and model design.

Filesystem MCP11 tools

Read, write, search, and manage files on your local filesystem with secure directory-scoped access for your AI editor.

Read and write datasets, configs, and model files directly. The glue that connects your AI assistant to your actual project files.

Other editors

Claude CodeCursorVS CodeWindsurfContinue