ego lite
A free tool that lets an AI coding assistant browse the web using your already-logged-in browser sessions, without you having to log it in separately or hand over your passwords.
🔗 Visit ego lite
Description
When an AI agent needs to check something on the web — your GitHub dashboard, an internal admin panel, a site you're logged into — it usually needs its own separate, logged-out browser, which means either it can't see what you can, or you have to log it in yourself with real credentials. ego lite solves this by letting the agent work inside the same browser instance as you, in a parallel workspace, inheriting your existing login sessions and cookies without ever touching your password.
It's an open-source (MIT licensed) macOS app (Apple Silicon and Intel supported, Windows/Linux on the roadmap) that gives AI agents a JavaScript-based control layer and page-snapshot capability so they can perceive and act on web pages the same way a human user would, while multiple agents can run in parallel without interfering with each other or with your own browsing. It's built to work with Claude Code, Codex, and Cursor, and the project claims benchmarks showing 2.5x faster task completion compared to Vercel's agent-browser tool. It has around 5.9k GitHub stars and 293 forks.
💬 Our review
The short version: if you've ever had to manually log an AI agent into a site just so it could check something, ego lite's session-inheriting approach removes a genuinely annoying step, and it's free.
Compared to giving an agent raw credentials or API keys to log in itself (a real security exposure), inheriting your existing session is a meaningfully safer default — the agent never sees your password. Against Vercel's agent-browser, the project's own claimed 2.5x speed advantage is worth verifying yourself rather than taking at face value, as is common with young open-source benchmarks, but the core idea of parallel workspaces without breaking your own browsing session is sound and useful regardless of the exact multiplier. The main limitation right now is platform support: macOS only, with Windows and Linux still on the roadmap, so it's not yet a fit for teams standardized on other operating systems.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Open-source (licence MIT), aucun coût
Pros
Hérite des sessions/cookies déjà connectés, sans exposer de mot de passe
Espaces de travail parallèles pour agents et utilisateur simultanément
Gratuit et open-source (MIT)
Compatible Claude Code, Codex, Cursor
Cons
macOS uniquement pour l'instant (Windows/Linux sur la roadmap)
Le gain de vitesse revendiqué (2.5x) vient du projet lui-même, à vérifier soi-même
Projet jeune, communauté encore réduite comparée à des outils établis