advai-browser

Advai Browser

Use this skill to operate a live browser through advai browser.

When to Invoke

Invoke this skill when the user wants to:

• Open or navigate real web pages
• Click, type, fill, select, or scroll in a browser
• Read DOM content, extract structured data, or execute JavaScript
• Capture screenshots, inspect cookies, or manage tabs
• Scrape public web content or validate browser-side behavior

Do not use this skill for static codebase inspection or for platforms that already have a more specific native skill.

Preconditions

Before running browser actions:

1. This skill depends on the advai CLI from https://github.com/Advai-X/advai-cli.
2. Ensure advai is installed and available on PATH.
3. Ensure the advai browser subcommand is available in the current installation.
4. If advai browser is missing, stop and guide the user to install or upgrade advai from https://github.com/Advai-X/advai-cli before continuing.
5. Ensure the browser extension/runtime is installed and connected.
6. Start with:

advai browser doctor

If browser support is unavailable, surface the doctor output and explain what is missing.

Core Workflow

Use a named session for each task, for example demo, zhihu, checkout, or research.

Basic flow:

advai browser open demo https://example.com
advai browser wait demo --selector "h1"
advai browser get demo
advai browser exec demo "document.title"

Reuse the current tab instead of creating a new one:

advai browser open demo https://example.com --replace

Important behavior:

• open creates a new tab by default
• open --replace reuses the current tab when the session already has one
• If the session has no active page, open --replace falls back to opening a new tab
• Tabs are not closed automatically after the action completes

Common Commands

Open and navigate:

advai browser open <session> <url>
advai browser open <session> <url> --replace
advai browser navigate <session> <url>
advai browser state <session>