Browser Use Orchestrator icon

Browser Use Orchestrator

Pilote browser-use-cdp-orchestrator (agents web, sessions, jobs, profils)

Actions16

Session → Get Session Screenshot

AI-generated

Summary

Captures a screenshot of the current page of an open multi-step browser session, without running an agent step. It calls the orchestrator's get_session_screenshot tool over MCP (POST {baseUrl}/mcp).

Inputs

  • session_id (required) — ID of the session to capture. This is the value returned by a Run Session Step operation.
  • full_page — Optional, set in Additional Fields. Captures the full page height when true. When false, captures only the viewport. The description says the server default is true. The node's own UI default is false, but it only sends the field if you add it to Additional Fields.

Output shape

A single JSON item per input item. It holds the tool's structured result if the server returns one. Otherwise it holds the parsed JSON text of the response, or { "text": "..." } if that text isn't valid JSON.

The source doesn't define the exact screenshot fields, so inspect a real run to see how the image is returned (for example, a URL). This is not a job-producing operation, so the Execution Mode and Max Wait settings don't apply. Errors from MCP, including an invalid session, throw. With Continue on Fail enabled, the item becomes { error: message } instead.

Examples

Example 1: Check what a session's page looks like between two agent steps.

Session ID = {{ $json.session_id }} from a previous Run Session Step; leave Additional Fields empty.

Example 2: Capture only the visible viewport.

Session ID set; add Full Page under Additional Fields and leave it off (false).

Links

Discussion