pydantic-ai-mcp-agent/studio-integration-version/pydantic_mcp_agent.py

from dotenv import load_dotenv
import pathlib
import logfire
import os

from pydantic_ai.models.openai import OpenAIModel
from pydantic_ai import Agent

import mcp_client

load_dotenv()

# 'if-token-present' means nothing will be sent (and the example will work) if you don't have logfire configured
logfire.configure(send_to_logfire='if-token-present')


# Get the directory where the current script is located
SCRIPT_DIR = pathlib.Path(__file__).parent.resolve()
# Define the path to the config file relative to the script directory
CONFIG_FILE = SCRIPT_DIR / "mcp_config.json"

def get_model():
    llm = os.getenv('MODEL_CHOICE', 'gpt-4o-mini')
    base_url = os.getenv('BASE_URL', 'https://api.openai.com/v1')
    api_key = os.getenv('LLM_API_KEY', 'no-api-key-provided')

    return OpenAIModel(
        llm,
        base_url=base_url,
        api_key=api_key
    )

async def get_pydantic_ai_agent():
    client = mcp_client.MCPClient()
    client.load_servers(str(CONFIG_FILE))
    tools = await client.start()
    return Agent(model=get_model(), tools=tools)