skills/eins78/skills/chrome-browser/Gen Agent Trust Hub

chrome-browser

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes the Google Chrome binary from its standard macOS path to launch a dedicated instance with remote debugging and profile isolation enabled. It also utilizes launchctl to install a persistent user-level LaunchAgent, ensuring the browser is available across sessions.
  • [EXTERNAL_DOWNLOADS]: Setup instructions recommend using npx to download and run the @playwright/mcp package, a well-known library for browser automation developed by Microsoft.
  • [PROMPT_INJECTION]: The skill enables an indirect prompt injection surface by allowing an agent to navigate and ingest arbitrary web content.
  • Ingestion points: External data is read into the agent context via browser navigation and snapshot capabilities.
  • Boundary markers: No specific delimiters or isolation instructions are implemented in the provided configuration scripts.
  • Capability inventory: The system allows for full browser control, including interaction with web elements and local network communication on the CDP port.
  • Sanitization: Web content is processed directly from the browser output without additional sanitization or filtering layers.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 03:29 AM