Everything I read gets sent to TypeSafe — what exactly leaves my browser?

Extracted page text up to 20,000 characters, plus the page hostname, title and content type, sent directly from the extension to https://api.typesafe.ai/v1/systemone with your key. There is no separate Sloppy Jev's server or analytics endpoint, so nothing sits in between, but the text is still leaving the machine.

Will this blur the wrong things, and can I get the text back?

Both halves are real. The README states model scores are estimates that can produce false positives or missed content, so genuine writing can be hidden. Hovering over blurred content reveals it, and the AI and ad cutoffs adjust separately from 0 to 100 percent, with lower hiding more.

Do I have to click anything, or does it start scanning on its own?

It starts by itself. Supported pages are scanned on load and whenever their content changes, and new X feed posts are scanned automatically, all once your key is saved. The 'Jev is on' switch turns it off across every site, cancels in-flight requests and strips the overlays while keeping your settings.

Will it work on a banking login page or a private dashboard?

Partly, and that is the catch. Chrome internal pages and other browser-restricted pages cannot be scanned at all, and the README explicitly warns against scanning private content unless you intend to send it to TypeSafe. Anything that does load still has its text sent for classification.

What browser and manifest version does the extension use?

It is a Manifest V3 Chrome extension written in plain JavaScript and CSS with no build step, MIT licensed, and loaded unpacked. It requires access to all websites.

Where does the extension store my API key?

In Chrome local extension storage. There is no separate Sloppy Jev's server, so the key stays in the browser while page text goes directly to the TypeSafe API.

What happens to cached scores when I change a cutoff?

Feed scores cached in memory are re-checked against the new cutoff immediately, while page-level results are re-fetched. The AI and ad cutoffs are independent.

CommunityJev Demos & Experiments

Sloppy Jevs Extension

Sloppy Jev's is a Manifest V3 Chrome extension that blurs AI-generated prose and ads, classifying extracted page text through jev-latest on the TypeSafe API. Plain JavaScript and CSS, no build step, MIT licensed, loaded unpacked.

See highlights

Open github.com

Stars: 0
Language: JavaScript
License: MIT
Published by: Community
Author: neddes
Last verified: Oct 2026

neddes/sloppy-jevs-extension

User repository on GitHub

View on GitHub

Open-source Chrome extension that filters AI-generated prose and ads with Jev

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

An open-source Chrome extension that blurs AI-generated prose and ads with Jev by TypeSafe. Download this repository as a ZIP from Code → Download ZIP, then extract it. You can also clone it. Open chrome://extensions and enable Developer mode. Click Load unpacked and select the folder containing manifest.json. Open the extension popup and enter your TypeSafe API key. Choose the AI and ad filters. Filtering starts automatically once your key is saved.

Quoted from the project's README. Read the full README.

Tagscommunity jevbest catalog-import awesomejev

Works forJev for Classification Jev for Moderation

Used inJev for Software & DevTools

Languages: JavaScript

Highlights

AI prose and ads are two separate Noul questions, each with its own 0-100% cutoff defaulting to 50%.
Pages are scored whole first, then section by section if flagged; X feed posts are queued and cached in memory.
Requests carry up to 20,000 characters of extracted text plus hostname, title and content type; text under 40 characters is skipped.
A global 'Jev is on' switch stops scanning, cancels active requests and removes overlays while the key and cutoffs persist.

Watch out

Classification is Jev's estimate on every page it scans, so false positives hide real writing and false negatives leave AI text visible; the README says so plainly and offers no fallback beyond hovering to reveal. It reads text only, so images and video are never judged. It needs access to all websites and sends extracted page text plus hostname, title and content type to https://api.typesafe.ai/v1/systemone — there is no separate Sloppy Jev's server or analytics endpoint, but the text still leaves the machine, which matters on private pages and anything behind a login. Chrome internal and other browser-restricted pages cannot be scanned at all, so coverage is partial by design. Feed scores are cached in memory and re-checked against a new cutoff immediately, while page-level results are re-fetched. The development loop is a manual reload plus a manifest.json bump, and the key lives in Chrome local extension storage. Copy reviewed against the repository README on 2026-10-04, not from running it.

Questions about this entry

Short answers drawn from the linked project or page.

More Jev resources like this

last-exit

Last Exit is a short cyberpunk checkpoint game where you type a cover story, present a manifest and offer a bribe to an inspector who decides whether to let you through. Each reply makes two Jev calls: one turns your statements into typed observations, the other picks his next action.