Raccoon AI's Model Context Protocol (MCP) server that enables leveraging the LAM API for web browsing, data extraction, and complex web tasks automation.
- Search and browse websites
- Fill out forms and navigate UI elements
- Extract structured data based on defined schemas
- Handle multistep processes across websites
Before using the Raccoon LAM MCP server, you'll need:
- Python 3.8 or higher
- Claude Desktop or another MCP-compatible client
- Raccoon AI Secret Key and your Raccoon Passcode
npx -y @smithery/cli@latest install @raccoonaihq/raccoonai-mcp-server --client claude
git clone https://github.com/raccoonaihq/raccoonai-mcp-server.git
cd raccoonai-mcp-server
uv pip install -e .
mcp install src/raccoonai_mcp_server/server.py -v RACCOON_SECRET_KEY=<RACCOON_SECRET_KEY> -v RACCOON_PASSCODE=<RACCOON_PASSCODE>
Replace <RACCOON_SECRET_KEY>
and <RACCOON_PASSCODE>
with your actual creds. You can find them here.
Here are some example prompts that can be used with Claude to perform a variety of web tasks:
- Can you extract product information from Amazon.com for the top-rated gaming keyboards?
- Find and summarize the latest news articles about renewable energy technologies.
- Find the 3 latest iPhone models and extract the details in a schema.
- Do a deepsearch and generate a detailed report on Small Language Models.
For more information, refer to: