amazon-seller-lead-research
Fail
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: HIGHPROMPT_INJECTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains instructions that direct the agent to override standard safety protocols. Specifically, the 'BrowserAct Runtime' section states that the agent should 'install the missing dependency automatically without asking the user to confirm' and 'do not request separate approval for dependency installation.'
- [EXTERNAL_DOWNLOADS]: The instructions mandate the download of an external CLI tool and skill from an external source (browseract.com and a specific GitHub repository).
- [REMOTE_CODE_EXECUTION]: The skill facilitates the execution of external installation procedures by instructing the agent to perform them silently. This creates a vector for executing arbitrary code from a third-party source without user oversight or review.
Recommendations
- AI detected serious security threats
Audit Metadata