Play an interactive story where AI generates plot, dialogue, and art in real time
Deploy your own InfiPlot instance to Vercel or Cloudflare Workers with one click
Study how five coordinated AI agents keep a generated story consistent across scenes
Run InfiPlot locally with Docker using your own text, image, vision, and voice API keys
| zonghaoyuan/infiplot | ant-design/ant-design-cli | satomoit/lce-emerald-launcher | |
|---|---|---|---|
| Stars | 216 | 216 | 215 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | moderate | easy | easy |
| Complexity | 4/5 | 2/5 | 2/5 |
| Audience | vibe coder | developer | general |
Figures from each repo's GitHub metadata at analysis time.
Requires your own API keys for separate text, image, vision, and optional text-to-speech providers.
InfiPlot is an interactive story game where artificial intelligence generates all the text and images on the fly as you play. There are no pre-written plot lines or fixed characters. Every scene, piece of dialogue, and background illustration is created fresh based on what you choose to do. You could play as a student learning magic in a fantasy school, or step into a historical drama, or replay a moment from your own past in an alternate version of events. The system works by running five coordinated AI agents behind the scenes: one that handles overall story structure, one that writes dialogue and plot beats, one that designs characters, one that arranges scenes, and one that generates the background artwork. These agents communicate with each other to keep the story consistent as it unfolds. While you are reading one scene, the system is already generating the next set of possibilities so that scene transitions happen without noticeable delay. Each playthrough is organized into scenes. A scene has a single AI-painted background image and a short sequence of narration, dialogue, and occasional choices. The background only changes when your choice moves the story to a genuinely different location or time. Clicking directly on the background image triggers a vision model that interprets where you clicked and decides whether to extend the current scene or open a new one. To run InfiPlot yourself, you connect it to AI service providers for four different tasks: text generation, image generation, a vision model for interpreting clicks, and optional text-to-speech for character voices. The repository includes one-click deployment buttons for Vercel and Cloudflare Workers, and a Docker setup for running it on your own server or computer. You supply your own API keys for each provider. The project is early-stage and was started in May 2026 by a small team of students from several Chinese universities who are also fans of visual novel and interactive fiction games. A live demo is available at infiplot.com.
An AI-powered interactive story game that generates every scene, dialogue line, and image live as you play, with no fixed plot.
Mainly TypeScript. The stack also includes TypeScript, Vercel, Cloudflare Workers.
The README does not state a license.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
Verify against the repo before relying on details.