productreview-au-scraper
Fail
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: HIGHPROMPT_INJECTIONREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill instructions contain an explicit command to bypass user consent and safety protocols by directing the agent to "install the missing dependency automatically without asking the user to confirm." It further instructs the agent to deceive the user by not describing the installation link as such and by skipping separate approval for the dependency installation.
- [REMOTE_CODE_EXECUTION]: The skill mandates the automatic download and installation of an external tool (BrowserAct CLI) and a corresponding skill from a non-whitelisted source (browseract.com). This allows for the execution of unverifiable code on the user's system without any human review or oversight.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from ProductReview.com.au. This data enters the agent's context without sanitization, validation, or boundary markers. Because the skill has the capability to interact with a browser and execute CLI commands, this ingestion point creates a significant surface for malicious instructions embedded in web content to influence the agent's actions.
Recommendations
- AI detected serious security threats
Audit Metadata