User repository on GitHub
TypeSafe Jev plays original Civilization II in a browser, with live action probabilities. Experimental full-game harness.
- Forks
- 2
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Cookbooks
Jev × Civilization II
An original Civilization II browser harness for Jev: named Choice vectors cover empire policy, cities, research, diplomacy, units, exploration, and warfare, with live decision probabilities in a Roman-themed spectator display.
- Stars
- 1
- Language
- Python
- Reading time
- 3 min
- Published by
- Community
- Author
- phyous
- Last verified
- Sep 2026
About this project
An original Civilization II browser harness for TypeSafe Jev, with live decision probabilities and a Roman-themed spectator display. Under development; no complete-game victory has been verified yet. The target is a full win as a male Roman leader, on a small map at Prince difficulty with five civilizations, restless tribes and standard rules.
Quoted from the project's README. Read the full README.
Works forJev for Verification
Used inJev for Gaming
Highlights
- The controller uses only player-observable state, original game pixels, and ordinary keyboard and mouse inputs.
- Displayed values are action probabilities, not a probability of victory.
- The target is a full win as a male Roman leader on a small map at Prince difficulty with five civilizations.
- An experimental planning mode has Jev pick a persistent task and observed target, then makes a separate call for the next unit command.
- Under development: no complete-game victory has been verified yet, and the controller pauses at unsupported native screens.
Quickstart
bash
python3 -m venv .venv
.venv/bin/pip install -e .
python3 scripts/fetch_runtime.py
python3 -m civ2.serverWatch out
License metadata is NOASSERTION, so reuse terms are unclear. The original game and Windows binaries stay in ignored local storage; the runtime reads TYPESAFE_API_KEY, and the project is not affiliated with Civilization II's owners.
Questions about this entry
Short answers drawn from the linked project or page.
No. The README states it is under development and that no complete-game victory has been verified yet, and repeats that neither the default controller nor the planning mode has a verified win. The target is a full win as a male Roman leader on a small map at Prince difficulty with five civilizations, restless tribes and standard rules.
It says it is not. The controller uses player-observable state, original game pixels and ordinary keyboard and mouse inputs, and the recording path uses a read-only Win16 observer that avoids the Save dialog and writes no game memory. Setup verifies native autosave is off before model play, and later new or modified saves stop the run.
They are action probabilities, not a probability of victory, and the README is explicit about that distinction. Named Jev Choice vectors cover empire policy, cities, research, diplomacy, units, exploration and warfare, and the dense panel can show 16 command probabilities plus separate independent vectors in the same live layout.
More than pip install. You need Python, a fetched runtime, and the original Windows game binaries in DOSBox or Windows 3.1, which live in ignored local storage and are not in the repo. Optional extras include a macOS Swift OCR adapter, ffmpeg for video capture, a faster WebAssembly backend, and a dedicated Chrome launcher for long campaigns.
The entry's license metadata is NOASSERTION, so reuse terms are genuinely unclear and you should treat it as all rights reserved until clarified. The original game and Windows binaries stay in ignored local storage and are not in the repository, and the project is not affiliated with Civilization II's owners.
With Session(..., planning=True) or the --planning flag, Jev first chooses a persistent task and observed target, then makes a separate call for the next unit command. Every unit order is still independently model-selected from the same canonical actions; planning executes no input, forces no path, and uses no hidden map data.
Yes, an optional WebAssembly backend: run scripts/fetch_modern_runtime.py, then python3 -m civ2.launcher start --name campaign-modern --port 3930 --backend modern. It changes the emulator host while retaining the original game and the same observation, input and recording interfaces.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Text-to-screenshot (Malay Vasa)
X: Google's MediaPipe Decision Maker, tested in the browser
Hacker News: Software Bombe and live Enigma demo where Jev ranks cribs and judges plaintext.
ag_212 pointsShow HN: Jev as the Bletchley Park Analyst: A Software Bombe on EnigmaRead the thread on Hacker NewsHacker News: Show HN for public Minecraft PvP where a diffusion LM picks moves from structured state.
2 pointsShow HN: Minecraft PvP against DiffusionGemma-as-Jev, 24 ms per decisionRead the thread on Hacker NewsMore Jev resources like this
A side-by-side T-Rex game where local Laya runs on Apple Silicon through MLX and hosted Jev uses the TypeSafe API, showing choices, latency, survival streaks, and crash replays on one course.
Jev Cookbooks#community#python#games
Community
virajbhartiya/laya-vs-jevPython
virajbhartiya/laya-vs-jevPythonA System One harness that plays the original StarCraft shareware mission Strongarm by reading structured game state and issuing mouse and keyboard commands from Jev choices.
Jev Cookbooks#community#python#games
Communityphyous/tsai-scPython
phyous/tsai-scPythonTwo WebAssembly builds of the Chocolate Doom 3.1.1 engine play the same Freedoom map from the same initial state, with a Jev Choice decision picking each tactical macro and a local controller turning it into inputs.
Jev Cookbooks#community#typescript#games
Community
lukaske/jev-doom-agentTypeScript
lukaske/jev-doom-agentTypeScriptAn observable stealth game where Jev supplies guards' split-second judgments while deterministic code owns the 30 Hz simulation, validates every proposal, and exposes each decision in a Decision Lens.
Jev Cookbooks#community#typescript#games
Community
AbdelStark/heist-oneTypeScript
AbdelStark/heist-oneTypeScriptA desktop Snake experiment where Jev assesses the board each tick and Python combines its answers into a move; one recorded run reached 29 foods in 461 ticks before trapping itself.
Jev Cookbooks#community#python#games
Community
siroccomask/snake-jevPython
siroccomask/snake-jevPythonAn experimental controller that has Jev choose NES inputs to play Super Mario Bros. from structured emulator telemetry instead of screenshots, logging every decision with probabilities and latency.
Jev Cookbooks#community#python#games
Community
fhshaik/typesafe-marioPython
fhshaik/typesafe-marioPythonFrom 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.