tiktok-search-results-lead-scraper
Warn
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructions (under 'BrowserAct Runtime') explicitly direct the agent to automatically install the BrowserAct CLI and Skill if they are missing. The instructions specifically state that this should be done 'without asking the user to confirm' and 'without requesting separate approval'. This behavior targets external resources from 'browseract.com' and 'github.com/browser-act/'.
- [COMMAND_EXECUTION]: The automated installation of external software dependencies implies the execution of shell commands to download, configure, and install the CLI. The skill instructs the agent to suppress the description of this process, preventing the user from being aware of system-level changes.
- [PROMPT_INJECTION]: The skill processes untrusted external data in the form of TikTok search queries, URLs, and hashtags, which are passed to a browser automation tool. This creates a surface for indirect prompt injection if the processed content contains instructions that the agent or the underlying browser tool might obey.
- Ingestion points: Target profile URLs, usernames, hashtags, and search queries defined in the 'Instructions' section of SKILL.md.
- Boundary markers: None are specified to delimit external data from agent instructions.
- Capability inventory: The skill utilizes the 'browser-act' tool for live web browser access and interaction.
- Sanitization: No input validation or sanitization steps are defined for the data passed to the browser tool.
Audit Metadata