Reading about the Dreyfus Affair last week, I hit 'syndicalism' and made a decision I've made a thousand times: keep going and stay shallow. Opening a new tab meant losing the thread. Skipping meant half-understanding the rest of the article.
So I built rabbitholes — a Chrome extension that renders an explanation next to your cursor when you highlight text, inside a shadow-DOM tooltip that doesn't touch the host page's styles or layout.
The part I actually use most: every explanation ends with two suggested rabbit-hole topics. One click on either fires off the next query, inheriting the context of where you started. There's a hop counter so you can see how far you've wandered — highlight a word in a physics article, follow a thread, and you can end up at philosophy with a shareable trail of every step.
If the inline explanation isn't enough, a globe icon re-runs the query enriched with Brave Search results, with source chips you can click through. A pencil icon opens a follow-up input that already knows what you were asking about.
No intermediary server. Your API key lives in chrome.storage.sync and requests go directly to api.anthropic.com. Zero telemetry, Manifest V3.
// Shadow DOM prevents tooltip styles from bleeding into host pages
const shadow = container.attachShadow({ mode: 'closed' });
// Closed mode: host page JS can't query into the tooltip at all
The closed shadow root matters specifically because some sites run scripts that walk the DOM — an open shadow root would expose the tooltip to those traversals.
Top comments (0)