The Wikipedia philosophy game works on any article: click a link, then click another, and within about seven hops you almost always land on Philosophy. I've spent more time doing that than reading the actual articles I opened.
rabbitholes started as a simpler itch — I'd be reading something and hit a term I half-knew, and the choice was always the same bad one: open a new tab and lose my place, or skip it and stay shallow. So I built a Chrome extension that renders an explanation inline via shadow DOM, right next to your cursor, without touching the host page's layout or styles.
The rabbit-hole counter came out of using it. When you click any word in an explanation to dig deeper — or drag across a phrase — the extension increments a hop counter. Keep clicking and you build a trail. Get deep enough into anything and you'll hit the same philosophical bedrock Wikipedia does. That trail is shareable.
A few other specific things it does:
Highlight → tooltip (Claude Haiku 4.5, shadow DOM)
Click word in response → explain that word (inherits context)
Globe icon → re-answers enriched with Brave Search, source chips included
Pencil icon → free-form follow-up, same context as background
Requests go directly from the browser to api.anthropic.com and api.search.brave.com — no intermediary server, no analytics, no telemetry. The Anthropic key lives in chrome.storage.sync. Manifest V3 throughout.
The thing I didn't expect: the suggested rabbit-hole topics at the end of each explanation are where most of the interesting detours start. Two options per answer, one click away. That's what keeps the counter moving.
Top comments (0)