Research

Completed

Browser Use agent scan · Meridian Health Systems

14 pages scanned

Every page of the client portal was crawled and captured.

212 elements documented

UI components indexed across the scanned pages.

1 structured JSON output

Passed directly into the Design page's UI/UX audit.

FlowOps first uses Browser Use, the leading open-source framework for AI browser automation. It runs entirely within the client's secure cloud environment, so client data never leaves the security perimeter. It's prompted to "log into this client portal, click every sidebar menu, expand all dropdowns, and document the available features" — going through the existing webapp the same way a human tester would.

BROWSER USE — AGENT EXECUTION LOG
✓Step 1: Logged into client portal
✓Step 2: Navigated to Settings tab
✓Step 3: Expanded all sidebar dropdowns
✓Step 4: Extracted user roles table
✓Step 5: Captured screenshots of 14 unique pages
✓Step 6: Compiled structured JSON output

Default outputs provided by Browser Use

When a Browser Use agent finishes navigating a web app, it natively returns a structured Python object containing three raw "ingredients":

Final history text

A string or JSON array summarizing what the agent did at every step — e.g. “Step 1: Logged in. Step 2: Clicked Settings tab. Step 3: Extracted user roles dropdown table.”

Media assets

Full screenshots captured at every click, or the entire browser execution session recorded as a WebM/MP4 video file.

Structured extractions

Any specific text, tables, or fields requested, formatted into clean Markdown, JSON, or CSV spreadsheets.

Output: Browser Use agent output in JSON array — passed to the Design page's UI/UX audit next.

Meridian Health Systems

Search workflows, files, agents...