Sharp
Browser extension that filters your X timeline by plain-language rules, with a decision model like Jev scoring every post by default.
Overview
Sharp is a Chrome and Firefox extension that filters your X timeline with a model you choose and pay for directly. You describe what to see or never see in your own words, such as 'hide engagement bait and crypto threads', add per-author rules and blocked words, and teach it by marking wrong calls or choosing 'hide posts like this'; recent corrections are sent to the model as examples. Optionally it tells X you are not interested in hidden posts. It also has YouTube tidying (hide Shorts and comments, blur thumbnails, no model involved) and greyscale modes.
By default a decision model classifies every post: fast, nearly free, and it reports how sure it is. Close calls stay behind a tinted banner with thresholds you set, and undecided posts wait behind a placeholder so the timeline does not jump; after three provider failures a post is shown rather than held. Jev can be reached via OpenRouter, Vercel AI Gateway or TypeSafe; any chat model can be used instead for written reasons. Images are described in words only when text leaves a post in doubt. There is no Sharp server; your key stays in the browser.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Should this post be hidden under the user's plain-language rules?(inferred) | — |
How it works
- Apply per-author rules and blocked words first
- Send undecided posts and recent corrections to the decision model
- Hide, show, or banner close calls by user thresholds
- Optionally signal 'not interested' to X
Strengths
- Nearly free per-post scoring with confidence
- No timeline jumps; posts wait until judged
- No Sharp server or account
Caveats
- Depends on X page structure
- Teach-X feature acts on your account
- Post text goes to your chosen provider
Getting started
Install from the Chrome Web Store or Firefox Add-ons, then add an OpenRouter, Vercel AI Gateway or TypeSafe key.At a glance
- Type
- Browser extension
- Status
- production
- Language
- TypeScript
- Runtime
- Browser extension
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Browser extension that filters your X timeline by plain-language rules, with Jev as the default classifier.
MrJev/awesome-jev
Frequently asked questions
What does the decision model decide?
Whether each X post matches your plain-language rules for hiding or keeping it.
Do I need TypeSafe?
No; Jev is reachable via OpenRouter or Vercel too, and chat models are an alternative.
Does Sharp see my timeline?
No; there is no Sharp server and your key talks to your provider directly.
Related projects
Privacy-first X filtering extension with local keyword and account rules, plus an optional Jev second pass on replies no rule matched.
PluginproductionChrome extension that labels Weibo, Threads and X posts by topic and expression style using Jev and your own label dictionary.
Browser extensionbetaTypeScriptChrome extension adding a second Twitch chat column that shows only messages Jev classifies as helpful, questions, funny or feedback.
Browser extensionbetaJavaScriptBrowser extension that scores X posts 0-100 with Jev on signal, actionability and originality and folds low-value posts.
Browser extensionbetaTypeScriptChrome extension that has Jev judge each LinkedIn post for engagement bait, self-promotion and relevance, hiding noise with explanations.
Browser extensionbetaTypeScriptChrome extension that covers YouTube comments while one Jev yes/no per comment checks for spoilers of the video or protected titles.
Browser extensionbetaTypeScript