# Add DoubleCrawl to OpenAI Agents SDK

> Give agents built with the OpenAI Agents SDK web access that works when a plain fetch fails. One minute to set up, 5,000 free credits, pay only for delivered pages.

### 1. Install the SDK

The SDK speaks MCP out of the box:

```bash
pip install openai-agents
```

### 2. Connect DoubleCrawl

Use your API key from the dashboard ([get one with 5,000 free credits](https://doublecrawl.com/signup?utm_source=integration&utm_content=openai-agents)):

```python
import os
from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp

async with MCPServerStreamableHttp(params={
    "url": "https://mcp.doublecrawl.com/mcp",
    "headers": {"Authorization": f"Bearer {os.environ['DOUBLECRAWL_API_KEY']}"},
}) as doublecrawl:
    agent = Agent(name="Researcher", mcp_servers=[doublecrawl])
    result = await Runner.run(agent, "How many flats are for sale in Warsaw?")
```

### 3. Run it

The agent can call `fetch_page` when it needs a page.

### Try asking

How many flats are for sale in Warsaw right now, and what's the median price per m²?

Web version: https://doublecrawl.com/integrations/openai-agents.html
