Skill UI

This skill provides comprehensive web browser automation capabilities for agents. It utilizes AI-optimized snapshots, reducing context size by 93% by using element references instead of full DOM. Functions include navigation, interactive element extraction, form filling, data scraping, and state management across multiple isolated sessions. Ideal for complex web testing and data extraction workflows.

Automates the process of filling out web forms using a structured field map (field_name → value). This tool supports using stored templates for known forms, or falling back to analyzing the page accessibility snapshot if no template exists. It includes mandatory PII screening (aidefence_has_pii) and handles form submission, making it ideal for automating sign-ups, checkouts, and data entry tasks.

This tool enables complex web interactions using plain natural language commands. Instead of relying on unreliable selectors, it utilizes a powerful page-agent and LLM loop to interpret user intent, performing actions like filling forms, clicking buttons, and extracting data from dynamic websites. Ideal for testing and automation in pages with ambiguous or changing markup.

This powerful utility automates the initial authentication process against a target website. It captures the resulting session cookies, sanitizes them rigorously using AIDefence to protect raw secrets, and then securely vaults a reusable cookie handle. This enables subsequent automation skills to seamlessly continue operations using the authenticated session context without needing to re-run the entire login flow.

A comprehensive skill to extract and transcribe subtitles from any YouTube video URL. It utilizes a two-stage approach: first, attempting fast extraction via the yt-dlp CLI tool; failing that, it falls back to advanced browser automation using Chrome DevTools MCP to scrape the transcript panel content directly from the webpage DOM. The output is always a clean, timestamped plain text file.

This SDK allows asynchronous Python programming to control a live web browser, simulating user interactions for web automation, testing, and data extraction. Users can navigate pages, locate elements, fill forms, click buttons, and capture snapshots of the current page's content. It enforces a 'perceive → act → verify' cycle, making it ideal for building robust web scrapers and end-to-end testing tools.