Headless & Headful Browsers
Give every session, scrape or screenshot its own disposable browser microVM (headless or with a full display) that starts in milliseconds.
Bring any browser you like; the platform is browser-agnostic. Its long startup happens once: snapshot the initialised browser into a template and boot every session from there.
Templates alone would give you identical browsers. A ROM mounted at boot carries that session's own profile, cookies, proxy or script, so one warm template serves many distinct sessions.
Run browsers with a real display and GPU-backed rendering (for video, WebGL and canvas-heavy pages that a headless, software-rendered browser can't handle).
Any browser (Chromium, Firefox, WebKit) plus your automation code
Browser fully initialized, then snapshotted
Watch the video to see a browser deployed with a full display, driven over VNC and screenshotted, asleep between each interaction.
Under the hood
The platform features this use case is built on.
Snapshot the browser after its long initialisation and start instances from that state in milliseconds.
Read-only blobs mounted on a VM at start. Paired with templates they customise each instance (profile, cookies, proxy, script) without rebuilding the image.
Headful browsers can render on a GPU, so video playback, WebGL and canvas-heavy pages behave as they do on a real machine.
Clones a running browser (memory, cookies and open tabs included), so several paths can be explored from one logged-in state.
Optionally filter the browser's network traffic, and inject secrets and API keys so the browser does not need to hold them.
Caps how long a session can run, useful when a page hangs and would otherwise stall the instance.
Sign up to Unikraft Cloud then follow the guide to automate your first browser