visual-qa-testing
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill uses the browser MCP to navigate URLs, capture screenshots, and audit network traffic for debugging purposes. These are standard features for UI testing.
- [COMMAND_EXECUTION]: The instructions include running 'npm run dev' to start a local development server. This is a standard project command in web development environments.
- [INDIRECT_PROMPT_INJECTION]: The skill navigates to and processes content from web pages, which represents an ingestion point for untrusted data. Malicious content on a web page could potentially contain instructions aimed at influencing the agent's behavior. 1. Ingestion points: Web content and console logs via 'browser_navigate' and 'browser_console_messages' in SKILL.md. 2. Boundary markers: None present in the instructions to distinguish between application content and agent instructions. 3. Capability inventory: Shell execution ('npm run dev'), network inspection, and browser interaction tools. 4. Sanitization: No explicit sanitization or filtering of web content is described.
Audit Metadata