The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill provides patterns and usage examples for generating and executing JavaScript automation scripts using Playwright. This is the primary mechanism for the skill's testing capabilities.
[INDIRECT_PROMPT_INJECTION]: The skill processes content from external websites and browser console logs, representing an indirect prompt injection attack surface.
Ingestion points: Untrusted data enters the agent context via page.goto() calls to external URLs and by monitoring browser console messages using page.on('console', ...).
Boundary markers: The instructions do not provide explicit delimiters or warnings to the agent to disregard instructions that might be embedded in the target web page's HTML or console output.
Capability inventory: The skill uses tools for browser navigation, user interaction (clicking, typing), and writing local files (screenshots).
Sanitization: There is no evidence of content sanitization, filtering, or validation before the agent processes information gathered from the browser environment.
[EXTERNAL_DOWNLOADS]: The documentation states that the Playwright dependency is automatically installed if it is not already present on the host system.
[COMMAND_EXECUTION]: The browser automation workflows involve executing system-level commands to spawn and control browser instances and their associated drivers.

webapp-testing