playwright
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions and code templates for performing end-to-end testing, which involves the agent interacting with and extracting data from untrusted web applications. This creates a potential attack surface where malicious content on a target page could attempt to influence the agent's behavior through indirect prompt injection.\n
- Ingestion points: The skill documents techniques for reading content from web pages using various Playwright locators (e.g.,
page.getByText()andpage.getByRole()as seen inloc-text-selectors.mdandloc-role-selectors.md).\n - Boundary markers: The provided examples and guidelines do not include explicit instructions to use delimiters or specific warnings to ignore instructions that might be embedded in the application's data.\n
- Capability inventory: The skill facilitates extensive capabilities including full browser control, network request interception and modification, and local file system access for managing authentication state.\n
- Sanitization: There is no evidence of sanitization or validation patterns for data retrieved from external applications before it is processed by the agent or used in assertions.\n- [EXTERNAL_DOWNLOADS]: The skill references documentation and best practices from trusted technology organizations including Microsoft (Playwright), Vercel (Next.js), and BrowserStack. These are standard industry references for testing and are documented neutrally without alarming language.\n- [COMMAND_EXECUTION]: The documentation includes standard CLI commands for running test suites and development servers, such as
npx playwright test. These commands are appropriate for the skill's stated purpose of guiding automated testing workflows.
Audit Metadata