stealth-browser

Pass

Audited by Gen Agent Trust Hub on Mar 14, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted web content, creating a surface for Indirect Prompt Injection where malicious instructions on a website could influence the agent's behavior. Ingestion points: wrapper.js (getContent(), getTitle()). Boundary markers: None (fetched content is returned directly without delimiters or safety warnings). Capability inventory: wrapper.js (evaluate(), screenshot(), launch()). Sanitization: None (content is not sanitized or escaped before being provided to the agent context).
  • [COMMAND_EXECUTION]: The browser instance is launched with --no-sandbox and --disable-web-security flags in wrapper.js. These settings bypass standard browser security boundaries (like the sandbox and Same-Origin Policy) to facilitate scraping, which increases the risk if the browser visits a malicious site.
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of the puppeteer-extra and puppeteer-extra-plugin-stealth Node.js packages. These are well-known technology libraries and are documented neutrally as they are standard dependencies for the skill's purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 14, 2026, 01:54 AM