typesafe-chess (Dimesio)
Local app testing Jev as a chess decision-maker against Stockfish, with Stockfish grading every pick.
Overview
TypeSafe Chess is a local app for measuring how well Jev chooses chess moves. chess.js owns the rules and lists legal moves, Jev picks from them, and Stockfish grades every pick; Stockfish's evaluations never reach Jev except in opt-in lesson levels that feed Jev its own graded mistakes. The aim is to measure, not to make Jev look good or bad, and findings are recorded in FINDINGS.md.
You can set each side to yourself, Jev or Stockfish, adjust Stockfish strength and grading depth, and change Jev's setup: raw move labels or assisted with move facts, and one Choice over all moves or yes/no questions per move. It runs on Node 20 with no build step, starts in a mock mode without a key, and the README estimates live decisions at about $0.00007 each, so a game costs well under a cent.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which legal move should be played? | — |
| Yes / no | Is this move a good move? (per-move noul mode) | yes · no |
How it works
- chess.js lists legal moves, optionally annotated with facts.
- Jev picks a move via Choice or per-move Noul.
- Stockfish grades the pick.
- Results and mistakes are recorded for analysis.
Strengths
- Independent grading by Stockfish.
- Mock mode for exploring without a key.
- Key stays on the Node server.
- Very low reported cost per decision.
Caveats
- An experiment; strength results are in FINDINGS.md, not summarized here.
- Requires Stockfish and a TypeSafe key for live play.
Getting started
npm install && npm start, then open http://localhost:5173At a glance
- Type
- Project
- Status
- research
- Language
- JavaScript
- Runtime
- Node 20 local web app
- Has code
- yes
Use cases and integrations
How curated lists describe it
FUn little experiment with Typesafe AI Jev Model playing chess against stockfish :).
hellogumbo/awesome-jev
Frequently asked questions
What is being tested?
How well Jev picks chess moves, graded by Stockfish.
Does Stockfish help Jev?
No, except in opt-in lesson levels that show Jev its own graded mistakes.
What does a game cost?
The README estimates well under a cent at about $0.00007 per decision.
Related projects
Fifteen live arcade games where Jev and LLMs play each other, led by WikiRace in which Jev picks only real links.
ProjectbetaTypeScriptOne-page web app where Jev plays chess against OpenRouter LLMs, Stockfish or you, showing the moves it weighed and their probabilities.
AppbetaJavaScriptPac-Man where Jev or other System One models choose each move at junctions, with live odds, fallbacks and a leaderboard.
ProjectbetaTypeScriptBrowser chess lab where Jev, LLMs and people play, each Jev turn being one Choice over the legal moves, with Stockfish analysis.
ProjectbetaPythonNiceGUI desktop chess app for evaluating Jev's move choices against Stockfish, random players or itself, with PGN export.
ProjectprototypePythonBrowser 2D driving sim where Jev answers lane, speed, hazard and pedestrian-yield questions every 200 ms, with an LLM comparison mode.
ProjectprototypeJavaScript