What actually leaves my browser when this is on?

Enabled features send the draft or post text and your classifier definitions directly to TypeSafe, and only that: images and the surrounding conversation are not included. Your key stays in local Chrome storage and is not shared with website scripts, and the project ships no developer server, analytics or saved post history.

Can a script on Bluesky read what the extension decided?

The labels are not posted to Bluesky, but the README is explicit that scripts on the page can read them, which is why it says to grant access only to sites you trust. Other domains can be added under Sites, and some need custom CSS selectors under Site detection before labels appear at all.

Will this quietly run up a bill?

The Usage tab shows request counts and an estimated cost, and it is an estimate: you set token prices under Cost estimate yourself, with Clear stats to reset the counters, and the README points at TypeSafe for actual charges. Filters are off by default, so nothing gets hidden or labelled until you switch classifiers on.

Does it look at the images people post?

No, labels are based on text only and attached images and videos are not analyzed. It requires Chrome 130 or newer, is loaded unpacked with no build step, and needs your own TypeSafe API key, which TypeSafe charges for per call.

What tone dimensions does Jevibe Check label, and can I add my own?

Built-in labels cover warmth, constructiveness, tension, sarcasm, clarity and intent. You can add a custom classifier with your own question and 2 to 12 possible labels. Draft labels and labels on posts you read can be turned on or off separately.

Why are some labels hidden?

Results below 55% model confidence are hidden. Filters are off by default, and a filter needs a classifier, a label, a minimum confidence, and either Blur or Collapse. Filters only run when that classifier is also enabled, and you can reveal a filtered post at any time.

How does the extension track requests and cost?

The Usage screen shows request counts and an estimated cost. You set token prices under Cost estimate, and Clear stats resets the counters while keeping your key and settings. The README warns to check TypeSafe for actual charges rather than relying on the estimate.

CommunityJev Tools & CLIs

Jevibe Check

Chrome extension that labels Bluesky posts and drafts with tone classifiers as you read or type, hiding results the model is not confident about.

See highlights

Open github.com

Stars: 2
Language: JavaScript
License: MIT
Published by: Community
Author: sriganesh
Last verified: Oct 2026

sriganesh/jevibe-check

User repository on GitHub

View on GitHub

A live tone labeler for Bluesky posts and drafts, using TypeSafe's Jev API.

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

A live tone labeler for Bluesky posts and drafts, using TypeSafe's Jev API. See warmth, constructiveness, tension, sarcasm, clarity, and intent while browsing or typing. Add custom classifiers with your own questions and labels. Blur or collapse posts that match your filters. Track API requests, token usage, and estimated cost. Labels are based on text only. Attached images and videos are not analyzed. The rudeness labels shown above are from a custom classifier.

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

Tagscommunity jevbest catalog-import awesomejev

Works forJev for Classification

Used inJev for Software & DevTools

Languages: JavaScript

Highlights

Built-in classifiers cover warmth, constructiveness, tension, sarcasm, clarity and intent; custom ones take a question plus 2-12 labels.
Results below 55% model confidence are hidden rather than shown, so weak judgments do not render as confident ones.
Filters are off by default and act by blur or collapse on a chosen classifier, label and minimum confidence, with reveal on demand.
No build step: load the extension folder unpacked, enter a key in the popup, refresh Bluesky, and reload after every update.

Watch out

Text only, with attached images and videos never analyzed. Anything you type is sent to TypeSafe, and while labels are not posted to Bluesky they are readable by scripts on the page, so the README says to grant access only to sites you trust. The Usage tab estimates cost from token prices you set yourself rather than reading an invoice. It needs Chrome 130 or newer, an unpacked extension load, and your own TypeSafe key. 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

jev-reviewer

Browser app for systematic-review extraction that opens a trial report with supplements and returns verbatim quotes with file and place, stored locally as projects holding studies.