exploring-browser

Pass

Audited by Gen Agent Trust Hub on Nov 4, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements strong safety principles for browser exploration. It explicitly instructs the agent to treat exploration as read-only and lists sensitive actions (Merge, Approve, Pay, etc.) that must not be clicked without explicit user confirmation (confirmSideEffect: true).
  • [SAFE]: The skill follows secure credential handling practices. It forbids the agent from generating or typing passwords found in source code and instructs the agent to ask the user to log in manually or provide credentials via secure methods like a specific file path or the 'persons_own: true' flag.
  • [SAFE]: Network operations are restricted to a specific MCP server (robomotion-browser-mcp) for browser automation and debugging purposes (e.g., discovery of APIs to transition from browser automation to direct HTTP requests).
  • [SAFE]: Instructions for handling user data, such as browser_export_session and browser_save_session, are presented as standard session management features for automated workflows.
  • [SAFE]: The instruction set reinforces the need for human-in-the-loop validation, requiring AskUserQuestion before committing to building flow code based on the exploration results.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 4, 2026, 08:01 AM