--- name: hyperbrowser-mcp description: Hyperbrowser Model Context Protocol (MCP) server for cloud headless browser automation, dynamic SPA crawling, session recording, and stealth web scraping (hyperbrowserai/mcp). Use when the agent needs to render complex JavaScript single-page apps, crawl websites, or execute browser agents. --- # Hyperbrowser MCP (Cloud Headless Browser Automation & Crawling) Based on [hyperbrowserai/mcp](https://github.com/hyperbrowserai/mcp): an MCP server providing cloud headless browser automation, dynamic JavaScript rendering, full-site crawling, and structured web data extraction for AI agents. --- ## 🎯 When to Use - **Complex Single-Page Applications (SPAs):** Scraping websites that rely on heavy client-side JavaScript, React/Vue hydration, or WebGL where static HTML extractors return blank templates. - **Deep Site Crawling:** Recursively following links across web applications, documentation trees, or API portals. - **Stealth Browsing & Session Recordings:** Recording video replays and interaction traces of web navigation for verification and audit trails. - **Browser Agent Delegation:** Powering browser agents (such as Claude Computer Use, OpenAI CUA, or Browser Use). --- ## 🚀 Setup & Registration ### MCP Configuration Register Hyperbrowser MCP in your agent's configuration (`.mcp.json` or IDE settings): ```json { "mcpServers": { "hyperbrowser": { "command": "npx", "args": ["-y", "hyperbrowser-mcp"], "env": { "HYPERBROWSER_API_KEY": "YOUR_HYPERBROWSER_API_KEY" } } } } ``` --- ## 🛠️ Key Capabilities & Tools - `crawl_page`: Crawls dynamic web applications and returns fully rendered DOM contents and structured link trees. - `extract_data`: Uses vision and LLM parsing to turn visual elements on web pages into clean JSON data models. - `record_session`: Records full interactive browser sessions into shareable artifacts for test reports and audits. --- ## ⚡ Integration with REA & Security Audits - **Auditing Client-Side Rendered Applications:** Render Next.js, Nuxt, or Vite web apps in Hyperbrowser to intercept WebSocket connections, runtime-injected scripts, and dynamic API calls that do not appear in raw HTML. - **Visual Regression Verification:** Capture live renders of web applications after applying security headers or CSP rules to confirm user interface integrity.