The Agent Skills Directory

[SAFE]: No malicious code or dangerous behaviors were detected. The skill is purely educational, providing best practices and code patterns for browser automation.- [PROMPT_INJECTION]: The skill demonstrates how to extract text and structured data from web pages (Indirect Prompt Injection surface). While ingesting untrusted content from the web is a risk, it is the primary purpose of this automation skill.\n
Ingestion points: SKILL.md (using textContent, innerText, $$eval, and $eval for data extraction)\n
Boundary markers: None specified in the provided code snippets\n
Capability inventory: Browser navigation, element interaction, form filling, and JavaScript execution via page.evaluate()\n
Sanitization: None specified in the patterns- [COMMAND_EXECUTION]: Use of page.evaluate() and $eval() allows for code execution within the browser's context, which is a standard feature of the Playwright library for interacting with page internals.- [EXTERNAL_DOWNLOADS]: The skill mentions the installation and use of the @playwright/mcp package via npx, which is an official tool for Playwright MCP integration.

browser-automation