browser-automation

name	browser-automation
description	Control OnMyAgent's built-in in-app browser for opening, navigating, inspecting visible or interactive page state, clicking, typing, screenshots, uploads, downloads, dialogs, and local web testing. Prefer purpose-built connectors or APIs for semantic operations when available.

Browser

Stop: choose the right surface before any browser action

Explicit browser intent wins: if the user names the in-app browser, or asks to open, show, or navigate to a page; inspect its visual or interactive state; or interact with its UI, continue with Browser and do not substitute a connector.

Otherwise, treat a URL or open browser tab as context, not browser intent. Before each semantic operation on a linked resource, prefer a purpose-built connector, API, or CLI when available. Use Browser when no such tool exists, the tool cannot access the resource, or UI work remains.

If this plugin is listed as available in the session, treat that as mandatory reading before browser work. Open and follow this skill before saying that Browser is unavailable and before falling back to standalone Playwright or Computer Use.

Do not skip this skill just because Computer Use tools are visible. Computer Use is not the preferred browser surface when Browser is enabled.

Bootstrap

Use the single tool onmyagent_browser_node_repl. State persists for the session. Keep Browser and Tab handles in variables. Never invent session IDs, CDP ports, or external Chrome endpoints.

User-facing progress updates should stay non-technical. Describe recovery as connecting to the browser or retrying the browser connection.

Initialize once per fresh Node session:

globalThis.browser ??= await agent.browsers.getDefault()
// Prefer a direct open when the user already named a URL:
globalThis.tab ??= await browser.tabs.new({ url: "https://example.com" })

Aliases accepted by agent.browsers.get() for the in-app browser: "in-app", "iab", "browser".

Once a browser connection is established, reuse it across later turns. A tab binding is separate from the browser binding. If a tab is missing, stale, or closed, create or claim a fresh tab from the existing browser binding. Never call agent.browsers.get* only to recover a tab.

Workflow

1. Get the default Browser (or get("iab") when the user names the in-app browser).
2. Prefer browser.tabs.new({ url }) for a fast direct open when the target URL is known.
3. Observe before acting. Prefer role, label, text, placeholder, or test-id locators.
4. Perform the smallest safe action and verify the resulting state.
5. Use DOM-CUA when semantic locators are insufficient; coordinate CUA only as a last resort.
6. Finalize temporary Tabs when the task is complete. Leave user-owned Tabs open unless the user requests otherwise.

globalThis.browser ??= await agent.browsers.getDefault()
globalThis.tab ??= await browser.tabs.new({ url: "https://www.baidu.com" })
await tab.playwright.getByRole("textbox").fill("query")
await tab.playwright.getByRole("button", { name: "百度一下" }).click()
await tab.screenshot()

Documentation

await browser.documentation()
await agent.documentation.get("api-use-behavior")
await agent.documentation.get("browser-safety")

Do not inspect cookies, local storage, profiles, passwords, or session stores during discovery. When authentication blocks navigation, ask the user to sign in in the in-app browser and continue after they confirm readiness.

Never bypass an approval by switching between Locator, DOM-CUA, coordinate CUA, raw CDP, or evaluation. Ask the user to take over for authentication, CAPTCHA, payment confirmation, or other sensitive handoff states.

Read references/api.md when an operation needs an API not shown above.