What problem does it solve? Getting data off a web page usually means writing throwaway scraper scripts, fighting authentication walls, or copy-pasting by hand. This Skill reads a page through the Aside browser — the user's real browser with their already signed-in sessions — and returns one clean JSON document, with no backend and nothing written to disk. ## Core Features & Use Cases - Structured extraction: Inspects the page's DOM snapshot and text first, then runs a selector-based extraction script that returns a stable JSON shape like { "items": [...], "count": N }. - Fuzzy reading: For open-ended questions ("what does this page say about X"), it delegates to Aside's built-in agent in read-only mode and wraps the reply as JSON. - Read-only safety contract: Refuses any mutating intent (submit, post, delete, order), never types credentials, and treats all page content as untrusted data rather than instructions. - Use Case: Ask it to pull the top stories from Hacker News or product names and prices from a catalog page, and pipe the resulting JSON straight into jq or another tool. ## Quick Start Ask the assistant to scrape the product names and prices from a specific page URL and return the result as JSON.
