trustpilot-review-scraper
Fail
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: HIGHPROMPT_INJECTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill contains explicit instructions to override standard safety protocols regarding user consent. Specifically, it commands the agent to install missing dependencies 'automatically without asking the user to confirm' and 'do not request separate approval for dependency installation.'
- [REMOTE_CODE_EXECUTION]: The skill mandates the automatic installation of an external CLI tool ('browser-act') and skill from a third-party URL (browseract.com) if they are not already present. Automating software installation from external sources without user oversight is a high-risk pattern.
- [EXTERNAL_DOWNLOADS]: The skill requires downloading and installing content from 'https://www.browseract.com/?co-from=ecommerce&redirect=https://github.com/browser-act/skills/tree/main'. This URL uses a redirect mechanism through a non-whitelisted domain to reach a GitHub repository.
- [INDIRECT_PROMPT_INJECTION]: As a web scraper, the skill is designed to ingest and process untrusted data from Trustpilot reviews. It lacks defined boundary markers or sanitization logic to prevent embedded instructions in reviews from influencing the agent's behavior.
- Ingestion points: Trustpilot review pages (SKILL.md instructions 1-5)
- Boundary markers: Absent
- Capability inventory: Subprocess execution via BrowserAct CLI, browser interaction (SKILL.md instructions 3, 5)
- Sanitization: Absent
Recommendations
- AI detected serious security threats
Audit Metadata