e2e-testing
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill contains instructions for installing several development and testing tools from well-known sources. These include the Maestro CLI via a shell script from mobile.dev, and the browse CLI via the npm registry. It also recommends injecting the axe-core accessibility testing library from a public CDN (cdnjs.cloudflare.com) during UI audits.
- [COMMAND_EXECUTION]: The instructions make extensive use of CLI commands to manage the testing lifecycle, including scaffolding projects (npx e2e init), running test suites (npx e2e run, maestro test, patrol test), and performing automated exploratory testing (npx e2e explore). It also uses shell utilities like xargs for parallelizing test execution.
- [INDIRECT_PROMPT_INJECTION]: As an automated testing tool, the agent ingests data from the applications it tests, such as accessibility trees, UI text, and screenshots. This constitutes an indirect prompt injection surface. The skill proactively mitigates this risk through a 'pixel taint' security feature that redacts screenshots and limits model visibility once a secret (like a password) has been entered into the application UI.
Audit Metadata