Browser and Capture
Sero has two related visual surfaces:
- the visible Explorer browser, which opens workspace-scoped browser tabs
- the UI-backed app/capture bridge, which can switch apps, inspect UI, capture screenshots, preview dev servers, and record short MP4s
They are useful for development previews, support reports, and agent/operator workflows, but they are not a replacement for a full browser automation framework.
Quick path: preview a dev server
- Start the server in the selected workspace runtime. Keep this terminal running.
For a Host workspace, use the development server's normal local bind:
For a container workspace, bind the server to all container interfaces:
- In a second terminal, register the running server. Use the command that matches the selected runtime.
- Open the listed URL in the visible browser or in a capturable app preview.
- Capture what Sero sees.
For container networking details, see Containers and Dev Servers.
Visible browser
The browser namespace controls loaded in-app browser tabs.
Important behavior:
- browser tabs are scoped to the current workspace
- explicit tab ids cannot control another workspace's tabs
- only
http:andhttps:URLs are accepted for open/navigate get-textandscreenshotdefault to the active tab in the current workspace- tabs appear in
listonly after their browser view is loaded
Use the visible browser for normal preview and page-reading workflows. Use sero app preview when you specifically need the preview inside the app panel for app screenshot/record capture.
App navigation and screenshots
The app namespace goes through Sero's renderer bridge.
Screenshots return an image block to the agent and can optionally save a PNG to disk. Relative --save paths resolve from the command cwd.
Failures usually mean the app panel is not available or not visible. Open the target app first, wait for it to render, then retry.
DOM interactions
Sero can perform simple interactions against the currently visible app panel.
click, type, scroll, scroll-to, select, and hover auto-capture a screenshot after the action. inspect, visible, snapshot, scroll-containers, and get-text return text/JSON and skip the post-action screenshot.
inspect and snapshot include temporary element refs, which can be reused with commands such as sero app click --ref e17 or sero app scroll --ref e42 --y 500.
Limitations:
- selector support depends on the rendered DOM in Sero's app panel
- coordinate actions are relative to the app screenshot, not the entire desktop
- typing requires an input, textarea, or contenteditable target
- hidden, offscreen, sandboxed, or cross-origin content may not expose useful DOM targets
Recording MP4s
Use recording for short visual evidence of UI behavior.
Recording captures at 2 FPS. The default save location is:
If MP4 output is not available for a capture, Sero may save a PNG frame folder instead. Recording state is bridge-local and ephemeral; do not expect it to survive renderer reloads or app restarts.
Screenshots for support reports
A useful support report usually includes:
Redact private code, tokens, account data, and personal content before sharing screenshots or recordings.