MAR 12, 2024
The Architecture of Tab Dedup: Under the Hood
A deep dive into how we use semantic analysis to manage browser state.
Tab Dedup started as a simple itch: too many identical tabs cluttering the browser. But naive URL matching fails the moment query strings, fragments, or redirects enter the picture.
Instead of comparing raw URLs, we normalize each tab into a semantic fingerprint — stripping tracking parameters, resolving canonical links, and hashing the result. Two tabs collapse into one only when their fingerprints agree.
The extension runs entirely on-device. No page content ever leaves the browser, which keeps the tool fast and preserves user privacy by default.