Skip to content

Jev directory

JEV Tetris — thelau

A playable Tetris where Jev never sees the board: code enumerates every legal placement as a sentence and Jev returns a probability for each in about 380 ms. The field glows with the whole distribution before the piece drops.
Stars
6
Language
JavaScript
License
MIT
Reading time
3 min
Published by
Community
Author
thelau
Last verified
Oct 2026

thelau/jev-tetris

User repository on GitHub

View on GitHub

A Tetris that a judgment model plays. The code finds every way the piece can land and writes each one as a sentence; JEV reads them and points at one. The board glows with its whole distribution before the piece falls.

Forks
0
Last pushed
Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

This is one of several Jev listings for the same project. The canonical entry is Jev Tetris — trungdq88.

Highlights

  • Every frame the code searches all placements the piece can reach — tucks and spins included — and describes each in a sentence; Jev answers in about 380 ms.
  • The board shows the whole distribution before the drop, every cell lit by preference and a white ring on the chosen placement.
  • Honesty check: shuffling the model's own numbers to the wrong options drops play to the random line — 0 lines and 18–24 pieces.
  • Less flattering row: a keyword table over the same sentences reaches 118/73/118 lines across seeds, and El-Tetris clears 118/118/116.
  • No dependencies, Node 22 or later; a game costs a few cents and the server refuses to spend past 1,500 calls or $1.

Quickstart

bash
git clone https://github.com/thelau/jev-tetris && cd jev-tetris
cp .env.example .env         # put your TYPESAFE_API_KEY in it
npm run dev                  # -> http://localhost:5173

Watch out

Copy reviewed against the repo README on 2026-10-15. The benchmark table is the project's own three-seed run with a 300-piece cap, not independent testing — and it prints a less flattering row where a keyword table over the same sentences and the El-Tetris heuristic both beat the model. Two fonts load from Google Fonts at runtime.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

X: Jev picks which Apple HIG guides go into the agent's context. NatashaTheRobot built a small AskHIG CLI and…

X: Three decision models, one game of Snake

X: Diogo Almeida with a16z on Jev

X: Luna plans, Jev plays poker

From 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.

Follow @typesafeai