SanityTV

Filter YouTube videos exploiting clickbait, rage-bait, brainrot or sensationalism. 100% local. No tracking. Free.

SanityTV gives you a calmer YouTube by filtering out videos that exploit your attention through clickbait, rage-bait, sensationalism, brainrot patterns, or content harmful to children. WHAT IT DOES SanityTV scores each video on the YouTube home, search, and sidebar using a transparent set of heuristics. By default, any video that crosses the threshold disappears from your feed — you simply see fewer videos. Prefer a softer experience? Turn off "Hide all flagged" in the popup and borderline matches will be greyed-out with a warning badge instead, still clickable. You can also adjust the sensitivity slider, or turn the filter off entirely with one click. WHAT IT DETECTS • Clickbait titles (uppercase shouting, listicles, "you won't believe", emoji spam, repeated punctuation) • Rage-bait (combat verbs like "DESTROYS", culture-war keywords, confrontation framings) • Sensationalism (mystery patterns, conspiracy keywords, hidden- truth narratives, morbid keywords for tabloid-style coverage) • Brainrot signals (Shorts duration, emoji spam) • Harmful kid content (Elsagate-style co-occurrences, named dangerous challenges like Tide Pod or Blackout) It works in English and French out of the box. WHAT IT DOES NOT DO • No data leaves your browser. Ever. • No third-party servers, no telemetry, no analytics. • No API key required. • Does not modify the YouTube video player itself. • Is not a parental control. Use YouTube Kids for that. FOR THE CURIOUS The full source code is on GitHub. Every detection rule is a few dozen lines of TypeScript you can read. The architecture decisions are documented as ADRs in the repo. The extension passes 130 unit tests and 16/18 empirical tests against real YouTube searches before each release. PRIVACY SanityTV does not collect, transmit, or sell any data. See the full privacy policy at the link in this listing. SOURCE https://github.com/Bist0uille/sanitytv REPORT BUGS / ASK FOR FEATURES Open an issue on the GitHub repository above.