Connecting to LLMs¶
🤖 AI Summary
Integration options by difficulty: Easy - Claude Desktop (config JSON), Claude Code CLI (.mcp.json), FastMCP Client (Python). Medium - LangChain, LangGraph, CrewAI, AutoGen, MCP SDK. Hard - Anthropic API, OpenAI API (manual tool loop). For quick chat → Claude Desktop. For Python apps → FastMCP/LangChain. For multi-step agents → LangGraph. For multi-agent → CrewAI.
This MCP server can connect to any LLM that supports the Model Context Protocol or tool calling.
Feature Support by Client¶
Different MCP clients support different features:
| Client | Tools | Resources | Prompts | Notes |
|---|---|---|---|---|
| Claude Desktop | ✅ | ✅ | ❌ | No prompts support |
| Claude Code CLI | ✅ | ✅ | ✅ | Resources via @, prompts via /mcp__ |
| Cursor | ✅ | ❌ | ❌ | Tools only |
| Windsurf | ✅ | ❌ | ❌ | Tools only |
| Zed | ✅ | ✅ | ❌ | No prompts |
| Continue.dev | ✅ | ✅ | ❌ | No prompts |
| VS Code + Copilot | ✅ | ✅ | ⚠️ | Partial support |
| LangChain | ✅ | ✅ | ⚠️ | Manual prompt loading |
| OpenAI API | ✅ | ❌ | ❌ | Tools via function calling |
Legend:
- ✅ Supported
- ⚠️ Partial/manual implementation
- ❌ Not supported
See Prompts for available coaching prompts.
Sources: Claude Code MCP Docs, MCP Sampling Status
Native MCP Support¶
These clients have built-in MCP support:
| Client | Setup Difficulty | Best For |
|---|---|---|
| Claude Desktop | Easy | Interactive chat with tools |
| Claude Code CLI | Easy | Development workflows |
Agentic Frameworks¶
Use MCP tools with popular agent frameworks:
| Framework | Setup Difficulty | Best For |
|---|---|---|
| LangChain | Medium | Complex agent pipelines |
| LangGraph | Medium | Stateful multi-step agents |
| CrewAI | Medium | Multi-agent collaboration |
| AutoGen | Medium | Conversational agents |
Direct API Integration¶
For custom implementations:
| Method | Setup Difficulty | Best For |
|---|---|---|
| FastMCP Client | Easy | Python scripts |
| MCP SDK | Medium | Custom clients |
| Anthropic API | Hard | Full control |
| OpenAI API | Hard | OpenAI models |
Which Should I Use?¶
Just want to chat with match analysis? → Claude Desktop
Building a Python application? → FastMCP Client or LangChain
Need complex multi-step analysis? → LangGraph
Want multiple specialized agents? → CrewAI