agentic-browser-testing
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides legitimate instructional content focused on quality assurance and browser automation. It emphasizes security and reliability best practices, such as using deterministic model settings and explicit success oracles.
- [EXTERNAL_DOWNLOADS]: The skill references standard development tools and packages, including @playwright/test and @playwright/mcp, which are maintained by a well-known service (Microsoft) and distributed via official registries. These downloads are documented for standard setup purposes.
- [COMMAND_EXECUTION]: The provided documentation includes examples of standard shell commands for initializing testing environments and running CI/CD pipelines. These examples follow secure practices, such as utilizing secret management for API credentials and enforcing execution timeouts.
- [PROMPT_INJECTION]: As the skill involves an agent navigating external web pages, it possesses a surface for indirect prompt injection from untrusted page content. The skill effectively mitigates this vulnerability by requiring explicit success assertions and negative constraints (oracles) that are validated by the test harness rather than the agent itself.
Audit Metadata