Clean Twitter

Filters AI slop, rage bait, ads and media out of your X timeline using TypeSafe Jev.

Clean Twitter reads your timeline and hides the posts you would rather not see: AI-generated filler, engineered outrage, and advertising. Each filter has its own confidence slider, so you decide how aggressive it is. Filtered posts collapse to a single line with the reason and a Show button — nothing disappears without a trace. WHAT IT FILTERS • AI slop — thread-bait, hollow platitudes, machine-written filler • Rage bait — engineered outrage, dunks, manufactured moral panic • Ads and shilling — sales pitches, affiliate spam, crypto promos • Promoted posts, images, videos and GIFs — read from the page, no API call • Your own filters — anything you can phrase as a yes/no question about a post BRING YOUR OWN KEY Classification runs on TypeSafe's Jev model, reached directly or through OpenRouter. You supply your own API key and are billed by that provider — roughly three cents per thousand posts. There is no account to create and no server of ours in between. WHAT LEAVES YOUR BROWSER To classify a post, its text and author name are sent to the provider you chose. Direct messages are excluded outright. Images are never uploaded. Results are cached, so a post is only ever sent once. Your API key is stored locally and sent only to that provider. Open source: https://github.com/midplane/clean-twitter