jev-gamepilot
Game agent claiming to combine local Laya reflexes with Jev strategy to play PC and Android games via screen capture and real inputs.
Overview
Jev-GamePilot describes itself as a universal autonomous gaming agent that captures gameplay at 60+ FPS, uses Laya for fast local decisions and TypeSafe Jev (Choice, Score and Noul) for higher-level strategy, and sends hardware inputs to Windows games and connected Android phones.
The README lists profiles for many games, including Balatro, Slay the Spire, Hearthstone and other card battlers, Windows Solitaire, aim trainers, Roblox, Minecraft, Trackmania and Chrome Dino, each with the inputs used and what the AI evaluates (hand values, energy and enemy intent, mana curves, obstacles). Performance claims are the author's and no measured results appear in the excerpt. Automating inputs in online games may conflict with those games' terms.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which game action should be taken now (play, discard, attack, move)?(inferred) | — |
| Score | How valuable is this hand or board state?(inferred) | — |
How it works
- Capture the game window or phone screen
- Extract game state
- Laya makes fast reflex decisions; Jev makes strategic choices
- Send keyboard, mouse or touch inputs
Strengths
- Two-tier design combining local and hosted models
- Many game profiles
Caveats
- No measured results in the excerpt
- Automating online games may breach their terms
- Claims not independently verified
Getting started
Run `pc_pilot.py` or `run-pc-pilot.bat` for PC games, per the README.At a glance
- Type
- Project
- Status
- prototype
- Language
- Python
- Runtime
- Windows desktop and Android via device connection
- Has code
- yes
Use cases and integrations
How curated lists describe it
\*\*Jev-GamePilot\*\* is a universal autonomous AI gaming agent powered by \*\*Laya (local sub-30ms System One inference)\*\* and \*\*TypeSafe's Jev System One\*\* (\`Choice\`, \`Score\`, \`Noul\`). It captures real-time gameplay at 60+ FPS, fuses instant local reflexes with high-level strategic reasoning, and executes physical hardware inputs across Windows PC games and connected Android phones.
logicrw/awesome-jev-projects
Frequently asked questions
What does Jev decide?
Strategic game moves, with Laya handling fast local reflexes.
Which games are supported?
The README lists card games, aim trainers, Roblox, Minecraft, Trackmania and others.
Are results measured?
Not in the README excerpt.
Related projects
Browser extension that lets Jev, local Laya or an OpenAI-compatible model take over a running ra2web (WannaFire) RTS match.
ProjectbetaTypeScriptBrowser arena where local Laya and hosted Jev race in Snake and fight in a Mortal-Kombat-style game, with every move a real decision.
ServiceprototypePythonChrome T-Rex game where local Laya on MLX and hosted Jev choose moves side by side, with latency charts and crash replays.
ServiceprototypePythonBrowser Pac-Man race comparing hosted Jev and self-hosted Laya choosing routes from the same board state, after code removes fatal routes.
ServiceprototypeJavaScriptRuns Intern-Decision-0.8B on Apple Silicon behind a local /v1/systemone endpoint that accepts screenshots for agent decisions.
ServicebetaPythonNative macOS computer-use loop: local OCR and accessibility find controls, Jev picks the next action from a finite list.
Projectbeta