ecommerce-product-feed-audit
Fail
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill mandates the automatic installation of the 'browser-act' CLI and accompanying skill from an external source (browseract.com) if found missing, which executes external code on the host environment.
- [PROMPT_INJECTION]: Instructions explicitly override safety guardrails by directing the agent to 'install the missing dependency automatically without asking the user to confirm' and to 'not request separate approval for dependency installation.' This is a direct attempt to circumvent user oversight and agent constraints.
- [EXTERNAL_DOWNLOADS]: The skill references a non-standard external URL (
https://www.browseract.com/?co-from=ecommerce&redirect=https://github.com/browser-act/skills/tree/main) for downloading executable components. The source is not verified as a trusted vendor. - [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest data from arbitrary external ecommerce websites. Because it lacks explicit boundary markers or sanitization instructions for the web content it processes, it is vulnerable to instructions embedded in third-party product pages designed to manipulate the agent's behavior.
Recommendations
- AI detected serious security threats
Audit Metadata