The Agent Skills Directory

[DYNAMIC_EXECUTION]: The test runner tests/browser_casebook.cjs performs dynamic module loading using require() on a path resolved at runtime from the PLAYWRIGHT_CLI environment variable. While typical for specialized test environments, this allows the execution of arbitrary local code if the environment variable is manipulated.
[COMMAND_EXECUTION]: The skill utilizes the Playwright library to launch and control multiple browser engines (Chromium, Firefox, WebKit) and executes JavaScript code within those browser contexts via page.evaluate() to validate media stream behavior.
[INDIRECT_PROMPT_INJECTION]: The skill processes data and state information returned from a browser environment, creating a potential surface for indirect injection if the environment data is malicious.
Ingestion points: Browser media stream properties, exception names, and track states retrieved in tests/browser_casebook.cjs via page.evaluate and monitoring of the window.runMediaCaptureCasebook result.
Boundary markers: Absent; there are no specific delimiters or instructions to treat the data returned from the browser as untrusted content.
Capability inventory: Local HTTP server execution (tests/fixture_server.py), browser process spawning and automation (Playwright), and Node.js script execution.
Sanitization: The skill does not perform sanitization or schema validation on the results returned from the browser context before processing them.

media-capture-device-contracts