browser-qa
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to navigate to and interact with external URLs to perform QA tests. This creates a vulnerability surface where malicious instructions embedded in a target website (e.g., in HTML comments, hidden text, or metadata) could attempt to influence the agent's behavior.
- Ingestion points: The skill reads and processes content from external websites via browser automation tools (e.g., navigating to URLs, checking console logs, and interacting with forms).
- Boundary markers: The instructions do not define specific delimiters or "ignore" instructions for the data fetched from web pages to prevent the agent from obeying embedded commands.
- Capability inventory: The skill utilizes powerful browser automation capabilities including navigation, clicking, form submission, and screenshot generation.
- Sanitization: While the skill correctly advises redacting PII and credentials from screenshots, it does not include instructions for sanitizing or escaping the text content extracted from the DOM before the agent processes it.
Audit Metadata