User repository on GitHub
- Forks
- 3
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Demos & Experiments
DOOM-JEV
An autonomous ViZDoom agent that decouples Doom's simulation from Jev SystemOne inference: the game runs at 35 ticks/sec while Jev is queried asynchronously at about 10 Hz for macro goals, targets, movement and firing decisions.
- Stars
- 8
- Language
- Python
- Reading time
- 2 min
- Published by
- Community
- Author
- amoghcreator
- Last verified
- Oct 2026
Used inJev for Gaming
Highlights
- Two decoupled loops: ViZDoom ticks at its native 35 ticks/sec while Jev SystemOne is queried asynchronously at about 10 Hz.
- Carry-hold actuation replays and adjusts actions between inference batches, so an 80-120 ms roundtrip never freezes the player.
- Depth-buffer median sampling plus 2D sector raycasting detects walls; a rendered label buffer keeps fire on visible hostiles.
- Jev guides macro strategy while trigonometry aims the crosshair and a trigger lock fires inside 15 degrees or on firing confidence.
- Rich terminal HUD shows health, ammo, threats, the decision DAG with confidences, and live roundtrip latency.
Watch out
Grounded on a truncated README and GitHub metadata captured 2026-10-15; no license is listed and the README excerpt carries no install or run commands, so setup is not covered here. The 35 ticks/sec, 10 Hz and 80-120 ms figures are the project's own description of its architecture.
Questions about this entry
Short answers drawn from the linked project or page.
An autonomous Doom agent powered by TypeSafe's Jev SystemOne model, pairing the ViZDoom engine with a decoupled asynchronous control loop, geometric raycasting and an interactive Rich terminal HUD.
So the agent does not suffer network-induced stutter: the simulation runs at Doom's native 35 ticks/sec with smooth rendering and continuous physics, while inference runs asynchronously at about 10 Hz, querying structured decision trees from Jev.
Structured questions covering macro goal, target, movement, rotation, jump and fire; answers come back as choices with calibrated confidences, while low-level aiming and actuation are handled geometrically between calls.
Carry-hold actuation smoothly replays and adjusts the action space using real-time geometric tracking. A trigonometric crosshair bearing steers onto enemies with zero delay, and the trigger locks when an enemy is within 15 degrees or firing confidence passes threshold.
Line-of-sight filtering uses the rasterizer's rendered label buffer so the agent only targets hostiles actually visible on screen. Depth-median analysis and 2D sector raycast intersections detect walls ahead, left and right.
The Rich terminal HUD shows the situation report (health, armor, weapons, ammo, visible threats, obstacles), the decision DAG with colour-coded confidence levels, the 8-channel actuation indicators and live API latency. Dynamic standing orders can be entered from stdin.
The digest's GitHub metadata lists it as Python, created and last pushed 2026-09-17, with no license and no archived flag. The README excerpt does not include setup or run instructions or reported scores.
More Jev resources like this
A pygame Flappy Bird that hands the bird to Jev: each new pipe becomes one multiple-choice question about simulated flight paths, and Jev answers with a typed choice plus a probability distribution.
Jev Demos & Experiments#community#catalog-import#logicrw
Community
leftspace89/jevbirdPython
leftspace89/jevbirdPythonWatch Jev play Freedoom in a local dashboard: structured engine observations go in, typed decisions come out, and the UI shows the game, choices, probabilities, navigation, latency and cost under a $15 default cap.
Jev Demos & Experiments#community#catalog-import#logicrw
Community
olivier-motium/jev-doomPython
olivier-motium/jev-doomPythonA local proof of concept where Jev picks which of two conflicting signal phases gets the green light at a real Hong Kong intersection, using near real-time keyless open-data traffic sensors. A phase-timing state machine enforces minimum and maximum green, amber and all-red around the model's answer.
Jev Demos & Experiments#community#catalog-import#logicrw
TypeSafe's Jev plays the real Steam game Vampire Survivors - it picks the character, stage, every level-up and the walking direction four times a second, with a live decision dashboard. A BepInEx plugin and Python brain; native Linux only.
Jev Demos & Experiments#community#catalog-import#logicrw
A demo that walks a real Pokémon FireRed ROM: the code reads the overworld out of RAM, works out every reachable destination, and Jev picks where to go — and how to fight wild encounters — through typed decisions, not screenshots.
Jev Demos & Experiments#community#catalog-import#logicrw
Community
daniel4x/jevemonHTML
daniel4x/jevemonHTMLTypeSafe Jev plays NetHack 5.0 with no LLM in the loop: each turn the code lists the legal moves, Jev picks one as a choice question, and a small motor translates the answer into keystrokes, with a live dashboard and Hardfought support.
Jev Demos & Experiments#community#catalog-import#logicrw
Community
statico/jev-nethackPython
statico/jev-nethackPythonFrom 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.