TestMu AI (Formerly LambdaTest) Agent icon

TestMu AI (Formerly LambdaTest) Agent

Give your AI agent a real browser to work with. Opens URLs, clicks elements, fills forms, and captures screenshots — all through a single tool.

TestMu AI (Formerly LambdaTest) Agent

AI-generated

Summary

Initiates and controls a cloud browser session to perform web actions like navigation, clicking, typing, and capturing content.

Inputs

  • browserName — Browser to use (e.g., Chrome, pw-chromium). Defaults to Chrome.
  • platformName — Operating system for the session (e.g., Windows 11, macOS). Defaults to Windows 11.
  • browserVersion — Browser version (e.g., latest, beta). Defaults to latest.
  • sessionScope — Scope of the browser session: 'execution' (shared across tool calls in one run) or 'call' (fresh per tool call). Defaults to 'execution'.
  • action (required) — The operation to perform: navigate, snapshot, click, type, get_text, or screenshot.
  • url — Absolute URL to navigate to. Required if action is 'navigate'.
  • ref — Element reference number from the latest snapshot. Required for click, type, and optionally get_text.
  • text — Text to type into the element. Required if action is 'type'.
  • submit — Whether to press Enter after typing. Defaults to false.
  • maxLength — Maximum length for get_text result. Defaults to 4000.
  • fullPage — Whether to capture the full scrollable page. Defaults to false.

Output shape

A single item containing the action result, session metadata, and optionally a snapshot.

Returns session details (sessionId, dashboardUrl, current URL, title) and the specific action result (e.g., clicked, typed, text, image). For 'snapshot', returns a formatted list of elements. 'get_text' returns truncated text. For 'screenshot', returns a base64 PNG string.

Examples

Example 1: Navigate to a website

Set action to 'navigate' and provide the target URL.

Example 2: Interact with a web page

Run a 'snapshot' to get refs, then use 'click' or 'type' with the specific ref number.

Example 3: Capture a screenshot

Set action to 'screenshot' and optionally enable 'fullPage'.

Links

Discussion