> ## Documentation Index
> Fetch the complete documentation index at: https://docs.olostep.com/llms.txt
> Use this file to discover all available pages before exploring further.

# MCP Server

> Use Olostep through the Model Context Protocol for fast, efficient web data

Olostep MCP Server offers web access tooling to your AI apps. It allows your AI to:

1. Search the web to ground AI on real world data (facts, docs, events)
2. Get the markdown content of any web page
3. Discover and filter out relevant URLs within a website.

## Installation

You need to have an API key. Get the API key from [https://www.olostep.com/dashboard/api-keys](https://www.olostep.com/dashboard/api-keys).

You can either use our remote hosted URL or run the server locally

### Remote hosted URL

```bash theme={null}
https://mcp.olostep.com/mcp
```

### Running with npx

```bash theme={null}
env OLOSTEP_API_KEY=your-api-key npx -y olostep-mcp
```

### Manual Installation

```bash theme={null}
npm install -g olostep-mcp
```

### Running on Claude Desktop

Add this to your claude\_desktop\_config.json:

```bash theme={null}
{
  "mcpServers": {
    "mcp-server-olostep": {
      "command": "npx",
      "args": ["-y", "olostep-mcp"],
      "env": {
        "OLOSTEP_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}
```

```bash theme={null}
npx -y @smithery/cli install @olostep/olostep-mcp-server --client claude
```

### Running on Claude Code

Run this in your terminal

```bash theme={null}
claude mcp add --transport http secure-api https://mcp.olostep.com/mcp \
  --header "Authorization: Bearer YOUR_API_KEY"
```

Alternative (local / stdio)

```bash theme={null}
claude mcp add --transport stdio --env OLOSTEP_API_KEY=YOUR_API_KEY olostep \
  -- npx -y olostep-mcp
```

### Running on Cursor

<a href="cursor://anysphere.cursor-deeplink/mcp/install?name=olostep-web&config=eyJjb21tYW5kIjoibnB4IiwiYXJncyI6WyIteSIsIm9sb3N0ZXAtbWNwIl0sImVudiI6eyJPTE9TVEVQX0FQSV9LRVkiOiJZT1VSX0FQSV9LRVkifX0%3D">
  <img src="https://cursor.com/deeplink/mcp-install-dark.png" alt="Add Olostep MCP server to Cursor" style={{ maxHeight: 32 }} />
</a>

#### Manual setup

To configure Olostep MCP in Cursor:

* Open Cursor Settings (click on the settings icon in the top right of the IDE)
* Click on MCP
* Click on "+ Add new global MCP Server"
* Enter the following in the mcp.json file:

```bash theme={null}
{
  "mcpServers": {
    "olostep": {
      "url": "https://mcp.olostep.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}
```

Alternative (local / stdio)

```bash theme={null}
{
  "mcpServers": {
    "olostep-web": {
      "command": "npx",
      "args": ["-y", "olostep-mcp"],
      "env": {
        "OLOSTEP_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}
```

* Remember to replace `YOUR_API_KEY` with your actual API key that you can get from the [Olostep dashboard](https://www.olostep.com/dashboard)
* You can close the settings tab now. If needed refresh the MCP server list to see the new tools

> If you are using Windows and are having issues, try `cmd /c "set OLOSTEP_API_KEY=your-api-key && npx -y olostep-mcp"`

Next time that you use Cursor it will automatically use the Olostep MCP when needed. You can also explicitly request to use the Olostep MCP when you need up to date information from the web

### Running on Windsurf

Add this to your ./codeium/windsurf/model\_config.json:

```bash theme={null}
{
  "mcpServers": {
    "olostep": {
      "serverUrl": "https://mcp.olostep.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}
```

Alternative (local / stdio)

```bash theme={null}
{
  "mcpServers": {
    "mcp-server-olostep": {
      "command": "npx",
      "args": ["-y", "olostep-mcp"],
      "env": {
        "OLOSTEP_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}
```

## Available Tools

* **scrape\_website**: Extract content from a single URL. Supports multiple formats and JavaScript rendering.
* **search\_web**: Search the web for a given query and return structured results (non-AI, parser-based).
* **answers**: Search the web and return AI-powered answers in the JSON structure you want, with sources and citations.
* **batch\_scrape\_urls**: Scrape up to 10k URLs at the same time. Perfect for large-scale data extraction.
* **get\_batch\_results**: Retrieve the status and scraped content for a batch job. If completed, fetches items and retrieves content for each URL.
* **create\_crawl**: Autonomously discover and scrape entire websites by following links from a start URL.
* **create\_map**: Get all URLs on a website. Extract URLs for discovery and site analysis.
* **get\_webpage\_content**: Retrieve content of a webpage in markdown format.
* **get\_website\_urls**: Search and retrieve relevant URLs from a website.

## Conclusion

You can find complete details about Olostep MCP [here](https://github.com/olostep/olostep-mcp-server).