User repository on GitHub
答案之书jev
- Forks
- 1
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Tools & CLIs
Answer Book (jev-to-answer)
A self-hosted answer book for everyday choices: an OpenAI-compatible LLM lists candidate options, Jev picks among them with a probability distribution, and answers can be favorited, restyled, or followed up.
- Stars
- 21
- Language
- JavaScript
- Reading time
- 2 min
- Published by
- Community
- Author
- csskrtao
- Last verified
- Oct 2026
Works forJev for Classification
Used inJev for Software & DevTools
- Languages
- JavaScript
- Platforms
- Vercel AI Gateway
- Capabilities
- Open sourceSelf-hostedRequires API key
Highlights
- Ask a question, get candidates from an LLM, then Jev evaluates them and returns a choice plus a probability distribution.
- Your own leaning (or 'hesitate') is recorded separately and never sent to Jev; Jev's answer is revealed on demand.
- Six reply styles (gentle, roast, concise, humorous, rational, healing) with a rewrite endpoint that does not re-call Jev.
- Runs on Node.js 22+ at localhost:9001 with an OpenAI-compatible LLM key and a TypeSafe or Vercel AI Gateway Jev key.
- Quotas default to 12 requests per minute and 60 per day per IP, plus a 3,000/day global cap reset at UTC midnight.
Quickstart
bash
npm ci
npm startWatch out
Copy reviewed against the repo README on 2026-10-15; the README is in Chinese and the manifest snapshot lists no license for the repository. Probabilities are documented as relative candidate preference, not real-world success rates, and quota counters run in memory and reset with the local server.
Questions about this entry
Short answers drawn from the linked project or page.
A self-hosted web app for everyday decisions, inspired by jev_demo: write your dilemma down, an OpenAI-compatible LLM lists candidate options, and Jev evaluates them and returns a choice with a probability distribution.
No. You pick your own leaning first (or choose 'hesitate') and that selection is never sent to Jev. Jev evaluates independently in the background, and the page reveals its choice and probability distribution only after you confirm, so its answer is not anchored to yours.
Node.js 22 or newer: run npm ci, copy .env.example to .env, fill in an OpenAI-compatible LLM endpoint plus a TypeSafe or Vercel AI Gateway Jev key, then npm start, which serves http://localhost:9001.
One decision typically spends two business requests and three upstream model calls; per-IP defaults are 12 requests per minute and 60 per day with a 3,000-per-day global cap. Counters run in memory and reset when the local server restarts.
Gentle (the default), roast, concise, humorous, rational, and healing. Switching style changes only the wording of direct answers and follow-ups; decision candidates stay neutral, and a rewrite never re-calls Jev or changes an existing choice.
In your browser: up to 100 saved answers, each keeping its recent 20 follow-up rounds. Clearing site data removes them, there is no sync between devices, and there is no login, so visitors share the server's model keys and quota rules.
It answers non-decision questions directly without generating options, has no web search for factual answers, and presents probabilities as relative candidate preference rather than real success rates. Jev is a decision model and cannot be configured as the chat model.
More Jev resources like this
Retro Neo-Brutalist web app for everyday micro-decisions in Chinese: whether to buy it, spend it, use it, what to eat, or whether to go, answered with a direct verdict and an optional TypeSafe Jev route.
Jev Tools & CLIs#community#catalog-import#logicrw
Community
kuhung/ask-jevTypeScript
kuhung/ask-jevTypeScriptToolkit, knowledge base and local dashboard for building with Jev via OpenRouter Alpha Decisions or the TypeSafe Direct API: a zero-dependency SQLite request log, a localhost REST server and Python CLI demos.
Jev Tools & CLIs#community#catalog-import#logicrw
Community
rileycarney/jevtoolsHTML
rileycarney/jevtoolsHTMLAn npm CLI that scans TypeScript, JavaScript and Python codebases for LLM calls that are really decisions, converts them into Jev modules with confidence-gated fallbacks, and shadow-tests them on recorded traffic before you swap.
Jev Tools & CLIs#community#catalog-import#logicrw
Community
0xjba/jev-swapHTML
0xjba/jev-swapHTMLA programming language where judgment is a runtime primitive: programs ask Jev narrow typed questions (noul, choice, score) as they run, and the Hocket engine coalesces judgments so two asks over one state become a single API request.
Jev Tools & CLIs#community#catalog-import#logicrw
Community
amberwhitehead/jevscriptJavaScript
amberwhitehead/jevscriptJavaScriptA read-only triage dashboard that finds likely duplicate GitHub issues: lexical retrieval shortlists candidates, Jev answers four Choice questions per pair, and selected passages are shown side by side. It never closes issues.
Jev Tools & CLIs#community#catalog-import#logicrw
A consensus engine for YouTube comments that replaces conversational summarization: Gemini Flash designs a per-video rubric once, Jev evaluates every comment against it in parallel batches, and code computes agreement deterministically. Live at jevpulse.vercel.app.
Jev Tools & CLIs#community#catalog-import#logicrw
Community
jaygajera17/jevpulseJavaScript
jaygajera17/jevpulseJavaScriptFrom 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.