How do the three sages reach a single decision?

Each of the three sages judges the question independently through Jev, then the system takes a majority vote weighted by calibrated probabilities.

Should I start with the English or the Japanese guide?

Start with README.en.md for English or README.ja.md for Japanese. The top README only points at those guides and the live site.

Is the live site enough or do I need to run code?

The live site hosts the interactive recreation for direct trying. Read the language guides when you want to understand or reproduce the setup.

How is a three-sage vote different from one Jev question?

One question gives one typed judgment. Here three independent sage judgments are combined by majority vote instead of trusting a single call.

What language is it in, and how much setup detail exists?

JavaScript, and very little. The top README is short and mostly points at README.en.md, README.ja.md, and the live site, so the language-specific guides are where the actual usage detail lives. Budget for reading the guide in whichever language matches you rather than expecting the landing page to carry it.

Does it need a TypeSafe key to try?

Three independent Jev judgments per question means three calls, so yes, you need a key to see the real thing rather than a single decision. The live site hosts the interactive recreation for trying it without setting anything up locally.

Is this affiliated with TypeSafe or the Evangelion rights holders?

It is a fan recreation of the MAGI concept from Neon Genesis Evangelision, not an official project. What it borrows is the three-sage majority-vote idea; the implementation calls Jev three times per question and combines the calibrated probabilities rather than taking a raw label.

CommunityJev Tools & CLIs

Magi System On Jev

Evangelion-inspired MAGI demo that runs a question past three independent Jev sages and decides by majority vote over calibrated probabilities, with a live site and bilingual guides.

See highlights

Open github.com

Stars: 1
Language: JavaScript
Published by: Community
Author: hide-G
Last verified: Oct 2026

hide-G/magi-system-on-jev

User repository on GitHub

View on GitHub

MAGI system (Neon Genesis Evangelion) recreated with Jev, TypeSafe AI's System One model. 3 sages deliberate your question.

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

Recreate the MAGI system from Neon Genesis Evangelion with Jev, the "System One" model from TypeSafe AI. Three sages judge your question independently and decide by a majority vote over calibrated probabilities. 新世紀エヴァンゲリオンの MAGIシステムを、TypeSafe AI の System One モデル Jev で再現。3賢者が議題を独立に判断し、確率つき多数決で結論を出します。.

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

Tagscommunity github

Works forJev for Classification

Used inJev for Software & DevTools

Languages: JavaScript

Highlights

Three sages judge the same question independently before voting.
Decision uses majority vote over calibrated probabilities.
Live site hosts the interactive recreation for direct trying.
English guide in README.en.md and Japanese guide in README.ja.md.

Watch out

Fan recreation with little setup detail in the top README; see the language-specific guides for usage. 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

using-typesafe-and-pinecone

Example collection pairing Pinecone retrieval with Jev judgments, including a reranking demo that narrows full-text candidates by natural-language criteria.