patrol-test-architecture
Pass
Audited by Gen Agent Trust Hub on Jun 24, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill outlines a modular framework for organizing Patrol E2E tests, including Modules for features, a System class for native interactions, and ApiClients for testing external services. These patterns follow established best practices for test automation engineering and code organization.\n- [COMMAND_EXECUTION]: The skill integrates specific tools such as patrol-run, patrol-screenshot, patrol-status, and patrol-native-tree. These commands are legitimate components of the Patrol testing framework used for test execution, debugging, and environment inspection.\n- [PROMPT_INJECTION]: No attempts to manipulate agent behavior, override safety instructions, or bypass platform constraints were detected. The instructions are focused strictly on the domain of test development and maintenance.\n- [DATA_EXFILTRATION]: No patterns involving the unauthorized extraction of credentials, environment variables, or sensitive local files were identified. Mentioned 'keys' refer to Flutter UI widget keys used for testing, not cryptographic or API secrets.
Audit Metadata