Skip to content

jev-skip

Browser extension that scores YouTube caption segments for sponsor probability with Jev and skips likely sponsor reads.

Browser extensionbetaMIT

Overview

jev-skip is a Chrome and Firefox extension that skips sponsor segments on YouTube videos that have not been crowd-labelled yet. It reads the caption track of the video you are watching, sends the transcript to Jev once, and gets back a probability per 30-second segment across five categories. Slices are painted on the seek bar tinted by confidence, and those over the threshold are auto-skipped.

It is for viewers who find SponsorBlock too late on fresh uploads. The README reports catching 77% of the sponsor seconds SponsorBlock's crowd marked across 23 videos, at 34 seconds of false skips per hour and $0.0008 per video, with answers about 0.9 seconds after the request. Requests are split when segment or token budgets are exceeded. Its stated limits: it reads text, not audio, so videos without captions get no opinion, and caption access depends on a proof-of-origin token trick that YouTube could change.

What the decision model decides

TypeQuestionAnswers
ScoreHow likely is this 30-second caption segment to be a sponsor (or other category) segment?(inferred)—

How it works

  1. Extension captures the signed caption URL and fetches captions as json3
  2. Transcript segments sent to Jev in one request (split if over budget)
  3. Per-segment probabilities painted on the seek bar by confidence
  4. Segments above the threshold are auto-skipped

Strengths

  • Works on videos nobody has labelled yet
  • README reports $0.0008 per video and about 0.9 s answer latency
  • Visible confidence: borderline slices are faint, confident ones solid
  • Only youtube.com and api.typesafe.ai are contacted

Caveats

  • No captions means no detection; it does not read audio
  • Caption fetching relies on YouTube internals that may change
  • Reported numbers were measured through a gateway shim and should be re-measured on the direct API

Getting started

git clone https://github.com/valentynkit/jev-skip && cd jev-skip && npm install && npm run build, then load dist/chrome-mv3 unpacked and paste your TypeSafe key.

At a glance

Type
Browser extension
Status
beta
Language
TypeScript
Runtime
Browser extension
Has code
yes
License
MIT

Use cases and integrations

How curated lists describe it

  • Media: browser extension that reads the YouTube caption track and scores each segment's sponsor probability on the seek bar before the intro ends, reporting 77% of SponsorBlock's sponsor seconds caught over 23 videos at $0.0008 a video.

    yibie/awesome-jev
  • Browser extension that reads the YouTube caption track and paints a per-segment sponsor probability on the seek bar before the intro ends, with no crowd database, reporting catching 77% of SponsorBlock's sponsor seconds across 23 videos at $0.0008 a video.

    AnotiaWang/awesome-decision-models
  • A browser extension that classifies YouTube caption segments with Jev, splitting requests when the segment or token budget is exceeded, paints five categories on the seek bar, and auto-skips those painted slices over the threshold.

    logicrw/awesome-jev-projects
  • Browser extension that reads the YouTube caption track and paints a per-segment sponsor probability on the seek bar before the intro ends, with no crowd database; reports catching 77% of SponsorBlock's sponsor seconds across 23 videos at $0.0008 a video.

    cobanov/awesome-jev
  • Browser extension that reads the YouTube caption track and paints a per-segment sponsor probability on the seek bar before the intro ends, with no crowd database.

    hellogumbo/awesome-jev
  • Browser extension that reads the caption track and paints a sponsor-probability overlay on the YouTube seek bar before the intro ends, with no crowd database. The author reports catching 77% of SponsorBlock's sponsor seconds across 23 videos at $0.0008 per video.

    MrJev/awesome-jev

Frequently asked questions

How is it different from SponsorBlock?

It does not wait for crowd submissions; it classifies the caption text itself so it works on brand-new videos.

How accurate is it?

The README reports 77% of SponsorBlock-marked sponsor seconds caught across 23 videos with 34 seconds of false skips per hour.

What does it cost to run?

You supply a TypeSafe key; the author reports about $0.0008 per video.

Related projects

  • Self-hosted podcast ad remover that transcribes episodes and uses an LLM or System One to detect and cut ad segments.

    Servicebeta
  • Chrome extension and web app that find YouTube sponsor reads with Jev from transcripts or audio and skip them.

    PluginbetaJavaScript
  • Designer's X demo where a natural-language phrase narrows a large icon set to matching icons, with Jev deciding.

    Discussiondiscussion
  • Privacy-first X filtering extension with local keyword and account rules, plus an optional Jev second pass on replies no rule matched.

    Pluginproduction
  • Desktop app that cuts long videos into captioned short clips, with Jev reviewing candidate cuts for strength, context and completeness.

    Appproduction
  • Chrome extension that asks Jev, sentence by sentence, whether text is what the reader came for and strikes out the filler.

    Browser extensionproductionJavaScript