McpLLMServer

import httpx

OLLAMA_URL = "http://ollama:11434/api/generate"
MODEL = "llama3"

async def generate(prompt: str) -> str:
    async with httpx.AsyncClient(timeout=60) as client:
        response = await client.post(
            OLLAMA_URL,
            json={
                "model": MODEL,
                "prompt": prompt,
                "stream": False
            }
        )
        response.raise_for_status()
        return response.json()["response"]