browser-qa
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to automate interaction with external websites, which are untrusted data sources. Malicious instructions hidden in a target website's HTML, console logs, or metadata could potentially attempt to influence the agent's behavior during the automated testing session.
- Ingestion points: Target URL navigation, console error monitoring, network request verification, and accessibility scanning (axe-core).
- Boundary markers: The skill includes clear instructional boundaries such as "redact credentials/tokens/PII before saving any screenshot" and "read-only by default," though it lacks explicit instructions to ignore natural language commands found within the target site's content.
- Capability inventory: The skill utilizes browser automation tools (claude-in-chrome, Playwright, Puppeteer) which allow for site navigation, form submission, and data extraction.
- Sanitization: The skill advocates for the use of seeded test credentials and the redaction of sensitive information before data persistence.
Audit Metadata