oxylabs

Oxylabs MCP Server exposes a set of web-scraping and AI-powered data extraction tools that you can access through the MCP framework. The server automatically surfaces different tool groups depending on which credentials you provide: Web Scraper API tools if you supply Oxylabs Web Scraper API credentials, AI Studio tools if you supply an Oxylabs AI Studio API key, or all tools if you provide both. The available tools include universal_scraper, google_search_scraper, amazon_search_scraper, amazon_product_scraper for general scraping tasks, and ai_scraper, ai_crawler, ai_browser_agent, ai_search for AI-driven data extraction and browser-enabled interactions. Use these tools to fetch, parse, and format content from websites, or to drive AI-assisted data collection across multiple pages and prompts. The MCP server will handle authentication with Oxylabs services and return structured results tailored for model consumption.

Prerequisites:

• A supported MCP runtime (uvx/uv/Smithery integration as described in the repository).
• Access credentials for Oxylabs Web Scraper API or Oxylabs AI Studio.
• Internet access from the machine where you run the MCP server.

Install and run using uvx (recommended in this repo):

1. Install the uvx package manager:

# macOS and Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

2. Install Oxylabs MCP configuration (already provided in this README) and start the MCP server using the config below:

{
  "mcpServers": {
    "oxylabs": {
      "command": "uvx",
      "args": ["oxylabs-mcp"],
      "env": {
        "OXYLABS_USERNAME": "OXYLABS_USERNAME",
        "OXYLABS_PASSWORD": "OXYLABS_PASSWORD",
        "OXYLABS_AI_STUDIO_API_KEY": "OXYLABS_AI_STUDIO_API_KEY"
      }
    }
  }
}

3. Run the MCP server with your config in a file named mcp-config.json (or as appropriate for your setup) and point uvx at it to launch Oxylabs MCP.

If you prefer the Smithery or OAuth-based setup, follow the specific steps in the README to configure via Smithery or via OAuth2/URL parameters. The essential prerequisites remain having valid Oxylabs credentials and a running MCP runner compatible with your workflow.

Tips and notes:

• Environment variables control which tool groups are exposed. Do not leave placeholder values active if you do not have corresponding credentials, as this may expose non-working tools.
• The LOG_LEVEL environment variable can be set to control verbosity (e.g., INFO, DEBUG).
• If both Web Scraper API credentials and AI Studio API key are provided, all tools will be exposed. If only one is provided, only the corresponding tools will be available.
• When using Smithery or OAuth2 configuration, ensure you follow the authentication flow to obtain and refresh tokens as needed.
• Ensure network access to Oxylabs services and that any required API quotas are understood for your use case.
• For local testing, you can stub environment variables with dummy values, but real operations require valid credentials.