Brief
Describe the player fantasy, genre, camera, controls, win condition, and one visual rule.
OpenGame Studio workflow
AI Game Agent that ships playable drafts, not just ideas
Use a focused prompt to let the agent create a browser-game draft, preview the result, iterate with version evidence, and export only after the core loop works in play.
Agent run card
Prompt → preview → export
One core loop per run
Browser preview before expansion
Version notes kept with each build
Studio export only after play evidence
The public search page explains the workflow. The actual generation workspace stays in Studio, where account state, credits, previews, and downloadable artifacts are handled safely.
How it works
A compact agent workflow for browser-game prototypes
Build
Let the agent create the first playable browser-game draft instead of stopping at a design note.
Preview
Open the generated build, check input, first-screen readability, feedback, and failure state.
Iterate
Keep the useful version, fix one visible issue, and export only when the loop survives play.
Review gate
The agent is useful only when the game can be played
A good AI Game Agent run produces a visible loop: input works, feedback is readable, the objective is clear, and the build can be shared or exported without guessing what happened.
First screen
The player can understand what to do.
Controls
Keyboard, pointer, or touch input changes the world.
Outcome
Win, fail, restart, or progress feedback is visible.
Handoff
The useful version has notes and an export path.
Start with one playable loop
Open Studio, write the smallest game brief, and let the agent produce a browser draft you can judge in play.
Open AI Game Agent