The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill makes extensive use of browser-based execution methods such as page.evaluate() and Runtime.evaluate via the Chrome DevTools Protocol (CDP). These methods allow the agent to run JavaScript directly within the browser context to manipulate the DOM, extract data, or bypass lazy-loading mechanisms. While dynamic execution is a sensitive capability, it is the core functionality required for browser automation and is implemented here using standard, well-documented APIs.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the extraction of data from external websites using locators like page.textContent and page.$$eval. This establishes an attack surface where malicious instructions embedded in a web page's content could theoretically influence an agent processing the results. The skill provides technical extraction methods but does not include explicit sanitization examples for the extracted text.
[COMMAND_EXECUTION]: The documentation includes standard package installation commands for Playwright, Puppeteer, Selenium, and AI-powered automation tools. These commands target official registries and well-known, reputable packages.

browser-automation