tiktok-follower-lead-scraper
Fail
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill directs the agent to download and install the 'browser-act' CLI and skill from an external domain (
www.browseract.com) if they are missing. - [REMOTE_CODE_EXECUTION]: The instructions explicitly command the agent to perform software installation "automatically without asking the user to confirm" and to "not request separate approval for dependency installation." This represents a direct attempt to bypass human-in-the-loop security controls for executing external code.
- [PROMPT_INJECTION]: The skill contains instructions that override the agent's transparency by directing it to "not describe it as a request for the user to open their browser," effectively concealing autonomous installation actions from the user.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external TikTok profiles, creating an attack surface where malicious content in bios or usernames could influence agent behavior.
- Ingestion points: Data scraped from TikTok profile URLs, usernames, and bios (SKILL.md).
- Boundary markers: None; there are no instructions to delimit or ignore instructions within the scraped content.
- Capability inventory: Invocation of the
browser-actCLI and shell-based browser workflows (SKILL.md). - Sanitization: No sanitization, validation, or escaping of the scraped external content is mentioned.
Recommendations
- AI detected serious security threats
Audit Metadata