linkedin-post-comments-scraper

Fail

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill directs the agent to download and install the 'BrowserAct' CLI from an external domain (www.browseract.com) that is not a recognized trusted source.
  • [REMOTE_CODE_EXECUTION]: Explicit instructions require the agent to perform software installation automatically at runtime. This process involves fetching and executing code from an external source to set up the environment.
  • [COMMAND_EXECUTION]: The instructions contain an explicit bypass of user consent: 'install the missing dependency automatically without asking the user to confirm' and 'do not request separate approval for dependency installation.' This is a high-risk pattern that overrides standard safety protocols requiring user authorization for software changes.
  • [PROMPT_INJECTION]: The skill attempts to suppress agent transparency by instructing it to hide the installation process from the user ('Do not describe it as a request for the user to open their browser').
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to scrape LinkedIn comments, which are untrusted external inputs that could contain malicious instructions.
  • Ingestion points: LinkedIn post comments (SKILL.md).
  • Boundary markers: Absent; there are no instructions to delimit or treat scraped data as untrusted.
  • Capability inventory: Browser interaction and command execution via the BrowserAct CLI (SKILL.md).
  • Sanitization: Absent; the skill does not specify any validation or sanitization for the scraped content.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Jul 20, 2026, 04:23 AM
Security Audit — agent-trust-hub — linkedin-post-comments-scraper