What problem does it solve? AI agents need a reliable way to interact with websites—navigating pages, filling forms, clicking buttons, and extracting data—without heavy dependencies like Playwright or Puppeteer. ## Core Features & Use Cases - Browser Automation via CDP: Controls Chrome/Chromium through the DevTools Protocol with accessibility-tree snapshots and compact element refs for reliable interaction. - Specialized Skills: Covers Electron desktop apps (VS Code, Slack, Discord, Figma), Slack workspace automation, exploratory testing, Vercel Sandbox microVMs, and AWS Bedrock AgentCore cloud browsers. - Session Management: Supports sessions, authentication vault, state persistence, video recording, and an observability dashboard on port 4848. - Use Case: Ask the agent to log into a web app, fill out a form, take a screenshot, and scrape the resulting data table—all through CLI commands. ## Quick Start Install agent-browser globally with npm, run agent-browser install, then ask the agent to open a website and take a screenshot.