The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents the use of standard, official Playwright CLI commands such as npx playwright test, npx playwright install, and npx playwright merge-reports for managing test lifecycles.
[EXTERNAL_DOWNLOADS]: The instructions facilitate the use of official Microsoft Playwright Docker images (mcr.microsoft.com) and well-known open-source packages such as @playwright/test and @axe-core/playwright from the NPM registry.
[INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection as it is designed to interact with and process data from the Document Object Model (DOM) of external web applications.
Ingestion points: Untrusted data enters the agent context via browser page content fetched through page.goto() and element locators defined in SKILL.md.
Boundary markers: The skill does not implement specific delimiters or instructions to treat web-derived content as untrusted data.
Capability inventory: The agent possesses capabilities to write files (storageState), perform network operations (APIRequestContext), and execute arbitrary browser actions.
Sanitization: There are no mentioned mechanisms for sanitizing or filtering data extracted from web pages before it is processed by the agent.

playwright-automation