User repository on GitHub
Telegram channel X-ray: Jev judges a year of posts, you get a card. One Cloudflare Worker.
- Forks
- 2
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.Homepage
Jev directory
CommunityJev Tools & CLIs
Jev Wrapped — gaborishka
Telegram channel X-ray that judges up to 1,500 public posts from the last year with one Jev Choice over ten post kinds and three Noul checks, then draws a shareable monthly mix card.
- Stars
- 3
- Language
- JavaScript
- License
- MIT
- Reading time
- 4 min
- Published by
- Community
- Author
- gaborishka
- Last verified
- Sep 2026
About this project
Paste the name of a public Telegram channel and get a card: what the channel posts, how much of it is ads, clickbait and emotional pressure, and how the mix changed over the last twelve months. Every post is judged separately by Jev, a model that answers typed questions with probabilities and generates no text. A channel with about 1,500 posts takes 35 to 85 seconds, depending on the way to Jev, and costs about $0.10 at list price. The whole thing is one Cloudflare Worker and fits the free plan.
Quoted from the project's README. Read the full README.
Works forJev for Classification
Used inJev for Software & DevTools
Highlights
- Each post with text gets its own Jev request with four typed questions: kind (one of ten), paid ad, clickbait and emotional pressure.
- A 1,500-post channel takes 35 to 85 seconds and costs about $0.10 at list price; posts without text are counted as media only.
- Reaches Jev through TypeSafe's API (2 pages at a time) or OpenRouter's alpha Decisions API (4 at a time, about 35 s).
- Posts come from Telegram's public t.me/s preview with no login or bot API; the id from a year ago is found by bisection.
- The card is one SVG string and the link preview is drawn pixel by pixel into a palette PNG with a 5x7 bitmap font, about 6 KB.
Quickstart
bash
git clone https://github.com/gaborishka/jev-wrapped.git
cd jev-wrapped
npm install
cp .env.example .env
npm run devWatch out
MIT-licensed. Needs Node 22+, a TypeSafe or OpenRouter key and a Cloudflare account with D1; every fresh run spends your key, and daily run and budget ceilings are set with env vars.
Questions about this entry
Short answers drawn from the linked project or page.
No. Posts come from Telegram's own public web preview endpoint, which requires no login, no bot, and no API key. A page holds about 20 posts and a before parameter pages backwards, while the id from a year ago is found by bisection over consecutive integer ids in roughly a dozen requests.
Not by choosing requests. The browser only asks for pages the Worker planned itself, post text always comes from Telegram rather than the browser, and the card is computed from rows the Worker wrote, so fabricated numbers cannot be posted. What is shared are the daily ceilings: fresh runs per address per day and a total budget for all visitors.
Treat them as a reason to look closer, not a verdict. Jev gives no explanation for an answer, and the README admits it marked two partner promotions as ads on a channel with no paid advertising at all. In its favour, the card surfaces the highest-scoring posts on clickbait, pressure, and advertising so a reader can check the extremes.
Mostly, with two ceilings. A 1,500-post channel is about 80 Worker requests and 80 D1 rows against a daily allowance of 100,000 of each, and about $0.10 at list price per run. The 10 ms CPU limit is the one you cannot measure locally: pages failing with error 1102 means the Workers Paid plan. Two visitors at once will get throttled and the run slows down.
Posts come from Telegram's own web preview at https://t.me/s/<channel>: no login, no bot and no API key. A page holds about 20 posts and ?before=<id> gives the 20 before it, and because post ids are consecutive integers it finds the id from a year ago by bisection.
shared/aggregate.js turns answers into shares, the monthly mix, median views per kind and the most telling posts. Clickbait and pressure count from a 50% probability; a post counts as an ad from 70% on the yes/no question, or 40% when the kind question also said ad.
Three: WRAPPED_DAILY_LIMIT caps fresh runs one address may start per day, WRAPPED_DAILY_BUDGET_USD caps all visitors together from the D1 rows, and the provider key's own spending limit holds even if the code is wrong. Every fresh run costs about $0.05 per 1,000 posts.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Search and tagging on keep.md
Reddit: Jev is on Workers AI as typesafe/jev
Other listings of this project
Forks and copies of this project in the directory. This entry is the canonical one.
A Telegram channel X-ray that pulls up to 1,500 public posts from the past year and has Jev rate each one, tallying news, ads, jokes, and clickbait into a shareable card.
Jev CookbooksPlayground#community#playground#demos
Community
More Jev resources like this
Unofficial Chrome Manifest V3 port of browser-use/jev-ultrafast: Jev picks the operation and DOM element in one request, a small text model writes typed values, and the run happens in your own tab with your logins. Ships a 17-task headless-Chromium suite.
Jev Tools & CLIs#community#typescript#chrome-extension
Community
chy4pro/jev-for-chromeTypeScript
chy4pro/jev-for-chromeTypeScriptA grep that finds lines by what they mean: TypeSafe Jev answers typed questions about each line, so one query matches across Japanese, French, Russian and Chinese, and meanings combine with AND, OR and NOT.
Jev Tools & CLIs#community#javascript#cli
Community
uehaj/sys1grepJavaScript
uehaj/sys1grepJavaScriptGrep by meaning instead of regex: Jev scores every line, meanings combine with AND, OR, and NOT, and one query matches across Japanese, French, German, Spanish, and more.
Jev Tools & CLIs#community#javascript#cli
Communityuehaj/jev-semgrepJavaScript
uehaj/jev-semgrepJavaScriptA structured decision plugin for agent harnesses: Jev recommends among currently available and permitted tools, skills, and agents, and evaluates output quality, returning choices, probabilities, and confidence while the host agent keeps planning and execution.
Jev Tools & CLIs#community#javascript#agents
A Claude Code plugin from alexei-led that picks the model and effort tier, micro to high, for each turn, advised by Jev, Cloudflare's Clef family, OpenAI, or a local Ollama classifier. It changes only model and effort, and it never proxies Anthropic traffic.
Jev Tools & CLIs#community#github#claude-code
Community
alexei-led/claude-routerJavaScript
alexei-led/claude-routerJavaScriptA little 3D chat friend with 17 outfits that dresses for whatever you ask: keyword rules pick looks by default, and supplying a Jev key lets Jev choose the outfit, mood and reply length. The live demo runs without keys on local rules and written replies.
Jev Tools & CLIs#community#github#javascript
Community
mortspace/momoJavaScript
mortspace/momoJavaScriptFrom the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.