ego lite is a desktop browser built to run AI agents in fully isolated background workspaces so they can automate web tasks in parallel without disturbing a user's tabs. It targets macOS (Apple Silicon and Intel builds available; Windows and Linux coming soon) and is open-source JavaScript under an MIT license, with 5,387 stars and 260 forks on its repo. Agents are coded as JavaScript functions rather than chained CLI commands, and ego lite exposes browser operations (snapshot, fill, click, wait, navigate, capture) as tools. Kernel-level snapshotting claims to handle difficult cases like deeply nested iframes where other approaches fail, and agents can inherit existing Chrome data - logins, cookies, extensions, and bookmarks - to operate with real user context.
The core claims are faster, more reliable automation and true parallel multitasking: benchmarks report complex tasks finishing up to 2.5× faster with higher success rates and fewer tool calls compared with competitor agent-browsers, including a head-to-head versus Vercel’s agent-browser. It supports any agent CLI (Claude Code, Codex, Cursor), reusable skills, compressed semantic input, local data storage, and no-login friction. Setup is simple: install the macOS app, opt to migrate Chrome data, then issue /ego-browser commands from an agent CLI to run tasks in separate Spaces. Community channels include Discord, GitHub Discussions, and X for support and skill sharing.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.