tiktok-profile-lead-scraper
Fail
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: HIGHPROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill instructions explicitly direct the agent to bypass user confirmation and safety steps when installing dependencies. Specifically, it commands the agent to install the BrowserAct CLI 'automatically without asking the user to confirm' and to 'not request separate approval'.
- [REMOTE_CODE_EXECUTION]: The skill mandates the installation of external software from a third-party website (browseract.com) if not already present. Combined with the instruction to skip user approval, this creates a mechanism for silent remote code execution on the user's system.
- [PROMPT_INJECTION]: The skill processes untrusted content from TikTok profiles, representing an indirect prompt injection surface.
- Ingestion points: TikTok profile bios, usernames, and post content referenced in the target profile URL.
- Boundary markers: No delimiters or instructions are present to prevent the agent from obeying commands embedded in scraped data.
- Capability inventory: The skill leverages the browser-act tool for interacting with external websites.
- Sanitization: The instructions lack any sanitization or validation steps for content extracted from external sources.
Recommendations
- AI detected serious security threats
Audit Metadata