Semantic feed filtering
Describe what you do not want to see in plain language, and Bouncer hides posts that match semantically rather than only by exact keyword.
Bouncer is a free, open-source app and Chrome extension for X that hides unwanted posts and helps shape your feed over time.
Bouncer is an app and browser extension for X that filters out the posts you do not want to see. You can describe unwanted content in plain language, then let Bouncer hide matching posts from your feed.
Imbue says Bouncer is designed to heal your feed as well as filter it: by reducing the posts you engage with, X's algorithm should gradually serve more of what you actually want. The product is free and open source, runs in the browser, and is available today on X as a Chrome extension and an iOS app.
Describe what you do not want to see in plain language, and Bouncer hides posts that match semantically rather than only by exact keyword.
Tap the trashcan icon on a post to filter that post and similar ones, which gives you a direct way to teach the filter from the feed itself.
See which posts were filtered and why, so the app remains understandable and adjustable instead of acting as a black box.
Run the filtering locally in your browser with on-device inference via WebLLM, with the source stating that no data is sent externally.
Use local models on your GPU or cloud APIs from OpenAI, Google Gemini, Anthropic, or OpenRouter, depending on your setup.
The interface adapts to light, dim, and dark modes automatically, which keeps the extension usable across different system themes.
For people who want to mute recurring topics, tones, or phrases in X without hand-curating a keyword list, Bouncer lets them describe the content in plain language and remove matching posts.
For users who prefer to shape recommendations through what they do not read, Bouncer hides unwanted posts and relies on the platform's feedback loop to gradually change what appears next.
For people who want a clearer sense of what the filter is doing, Bouncer exposes the reasons a post was removed and lets users adjust the rules from the feed itself.
For teams or builders evaluating a browser-based content filter, Bouncer provides a free, open-source reference implementation with on-device inference and optional backend choices.
Imbue presents Bouncer as an app and browser extension for X that filters posts you do not want to see. Its FAQ says it works by hiding matching posts and letting X's algorithm learn from what you stop engaging with, which can shift the feed over time.
The source says Bouncer is available on X today as a Chrome extension and as an iOS app in the Apple App Store. It also notes that support for Reddit and LinkedIn is coming next.
Bouncer runs in your browser and the site says it does not collect or store identifying data about you. It also says filtering is powered by a small open-source language model called Qwen3.5-4B.
The source says you can filter by phrase, keyword, emoji, topic, tone, or even a color, and you can tap the trash icon on a post to remove similar posts as well.