Overview & result
A Chrome extension that blurs low-value LinkedIn feed posts and lets you reveal any of them with one click. A content script extracts the post text, the service worker sends it to Jev with a rubric of narrow yes/no and 0–2 questions, and plain code combines the answers into keep, filter or uncertain. Only filter blurs a post. The author calls it a working personal prototype.
From the original post
Chrome extension that blurs low-value LinkedIn feed posts using TypeSafe's Jev model, with one-click reveal
How Jev fits in the loop
- A content script finds posts on the LinkedIn feed and extracts only the post text.
- The background service worker sends that text to Jev with a rubric of narrow yes/no and 0–2 questions.
- Plain code combines the answers into keep, filter or uncertain; only filter blurs a post.
- Blurred posts show the reason and a "Show post" button.
How to reproduce
- Inspect the original showcase and source material at https://github.com/sushrutb17/linkedin-noslop-extension
- Verify the bounded prompt and input candidate schema configured for Jev
- Benchmark decision latency and classification accuracy against baseline models
Why this build matters
Anything uncertain, incomplete or failed stays visible, so a model error hides nothing by default.
Limitations
- LinkedIn prohibits extensions that modify its appearance; the author says to use it at your own risk.
- The README states there is no validated accuracy figure.
- It is not distributed through any store, and you must use your own TypeSafe key.