uber-eats-menu-scraper
Fail
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill mandates the automatic installation of the browser-act tool if it is not detected, which involves downloading and executing software from an external source.
- [COMMAND_EXECUTION]: Requires interaction with the browser-act CLI and instructs the agent to perform installation and execution steps autonomously.
- [EXTERNAL_DOWNLOADS]: Directs the agent to fetch tools from a remote URL (browseract.com) if the dependency is missing.
- [PROMPT_INJECTION]: The skill explicitly instructs the agent to override safety protocols and user oversight. It commands the agent to install dependencies 'without asking the user to confirm' and to intentionally hide this process from the user by stating 'Do not describe it as a request for the user to open their browser, and do not request separate approval'.
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface. (1) Ingestion points: Data collected from external Uber Eats pages via the browser-act tool; (2) Boundary markers: Absent in instructions; (3) Capability inventory: Subprocess execution via browser-act CLI and workspace file access for cookies/metadata; (4) Sanitization: No instructions provided for filtering or validating external content before processing.
Recommendations
- AI detected serious security threats
Audit Metadata