Hyperbrowser MCP Server is Hyperbrowser's implementation of the Model Context Protocol (MCP). It provides tools to scrape webpages, extract structured data, and crawl linked content, while offering ready-made access to general purpose browser-use agents such as OpenAI's CUA, Claude Computer Use, and the Browser Use agent. The server can be integrated with the Hyperbrowser API, which supports a superset of MCP features. Core capabilities include: scraping webpage content (formatted output, including markdown and screenshots), crawling linked pages to gather LM-friendly content, and converting messy HTML into structured JSON. It also supports web search via Bing, and various browser automation agents for rapid experimentation. Profiles can be created, deleted, and listed to manage persistent Hyperbrowser sessions. The MCP server can be run locally for development (npm install, npm run build, node dist/server.js), installed via npx hyperbrowser-mcp <YOUR-API-KEY>, or deployed via Smithery for automatic installation with Claude Desktop clients. Documentation and resources are available through Hyperbrowser and MCP documentation; MIT license.