skills/browseros-ai/skills/browseros/Gen Agent Trust Hub

browseros

Pass

Audited by Gen Agent Trust Hub on Apr 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill relies on the browseros-cli tool to interact with a Chromium browser instance. It executes commands via Bash to perform navigation, element interaction, and browser management.
  • [EXTERNAL_DOWNLOADS]: The setup instructions include downloading the browseros-cli package via NPM and the BrowserOS application binary via browseros-cli install. These resources originate from the official vendor (browseros-ai) infrastructure.
  • [REMOTE_CODE_EXECUTION]: The skill includes an eval command that allows the execution of arbitrary JavaScript code within the browser's execution context. While a standard feature for automation, it represents a high-privilege capability.
  • [DATA_EXFILTRATION]: The tool has the capability to access sensitive user data, specifically browser history (history recent/search) and bookmarks (bookmark list/search). It can also capture screenshots and export pages as PDFs.
  • [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection as it ingests untrusted data from the web (via text, snap, and dom commands) and possesses powerful capabilities like browser interaction, file writing, and JavaScript execution. No explicit sanitization or boundary markers for this data are defined in the skill instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 2, 2026, 01:19 PM