The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads browser binaries from Playwright's official CDN (cdn.playwright.dev) during the setup phase using the npx playwright install command.
[COMMAND_EXECUTION]: The skill instructs the agent to execute several shell commands, including pnpm for package management, npx for Playwright installation, and node to run generated test scripts. It also involves starting a background Chrome process with the --remote-debugging-port flag and using pkill to terminate it.
[DYNAMIC_EXECUTION]: The skill generates and executes standalone .mjs scripts at runtime to perform browser automation tasks. These scripts are created from templates within the skill and executed using Node.js to verify local or remote web content.
[INDIRECT_PROMPT_INJECTION]: The skill interacts with web pages (via page.goto()) and processes their content. This creates a surface for indirect prompt injection if the visited pages contain malicious instructions designed to manipulate the browser automation or the agent's logic.
Ingestion points: Browser interaction with local and remote URLs in SKILL.md (e.g., page.goto("http://localhost:3000/some-page")).
Boundary markers: None identified in the provided templates.
Capability inventory: The skill has the ability to execute shell commands, run Node.js scripts, and control browser instances.
Sanitization: No explicit sanitization or filtering of web content is specified before the agent processes results like console logs or page measurements.

playwright-verify