Skip to main content
Flower Engine Hero Light

​
What is Flower Engine?

Flower Engine is a modern, AI-powered roleplaying system that combines the depth of tabletop RPGs with cutting-edge technology. Built with a split architecture design, it delivers blazingly fast, real-time narrative experiences through a sophisticated Python backend and a beautiful Rust terminal interface.
Fully Vibecoded Application — Built at the speed of thought with AI-first engineering.

​
Key Features

Flower Engine is designed for hardcore tabletop-style roleplaying with modern AI capabilities.

Split Architecture

Decoupled Python “Brain” (FastAPI/LLM) and Rust “Face” (Ratatui TUI) communicating via WebSockets for maximum performance and modularity.

Live Token Streaming

Blazingly fast, real-time response rendering with typewriter animations. Watch the story unfold character by character.

Hardcore System Rules

Built-in logic for grounded reality, NPC autonomy, and consequence-driven storytelling. The world reacts to your choices.

Session Management

Full history persistence, session hot-swapping, and strict memory isolation. Never lose your progress.

RAG-Powered Memory

Vector-based retrieval with ChromaDB and local embeddings ensures contextually relevant responses without sending your data to the cloud.

Multi-Provider Support

Works with OpenRouter, Google Gemini, Groq, and DeepSeek. Switch models on the fly without restarting.

​
Architecture Overview

Flower Engine uses a split-brain architecture that separates concerns for optimal performance: 
    [ THE FACE ]             [ THE BRAIN ]
    (Rust / Ratatui)         (Python / FastAPI)
          |                         |
    TUI Interface <--- WebSocket ---> LLM Orchestrator
          |            (JSON V1)            |
    Async Input                     RAG (ChromaDB)
    Event Loop                      SQLite Persistence

​
The Brain (Backend)

  • Python 3.12+ with FastAPI for asynchronous orchestration
  • ChromaDB for vector storage and RAG (Retrieval-Augmented Generation)
  • SentenceTransformers (all-MiniLM-L6-v2) for local embeddings
  • SQLite for session and character persistence

​
The Face (Frontend)

  • Rust with Ratatui for a blazingly fast terminal UI
  • Tokio for multi-threaded async event handling
  • WebSocket connection for real-time streaming
The split architecture means you can run the backend on a powerful server while keeping the lightweight TUI on your local machine.

​
System Rules

The engine enforces a permanent ruleset to ensure immersive, consequence-driven gameplay:
  1. Grounded Reality: No outside intervention or prophecies
  2. Player Is Not Special: You must earn your influence through action
  3. Secrecy Is Absolute: NPCs do not have access to your character sheet
  4. Time Persists: The world moves forward even if you wait

​
Quick Start

Ready to begin your journey? Get up and running in minutes.

Quickstart

Get Flower Engine running in 5 minutes with our automated setup.

Installation

Detailed installation guide with system requirements and troubleshooting.

Architecture

Deep dive into the split-brain design and WebSocket protocol.

Creating Worlds

Learn how to build rich, immersive worlds with YAML configuration.

​
Use Cases

Flower Engine is perfect for:
  • Solo Tabletop RPG Sessions: Experience full campaigns with an AI narrator
  • Creative Writing: Generate and explore narrative branches interactively
  • World Building: Test and develop your RPG settings with real-time feedback
  • Game Master Tools: Use as a co-narrator for your tabletop games

​
Next Steps

1

Install Prerequisites

Set up Python 3.12+, Rust, and get your API keys ready.
2

Run the Setup Script

Our automated script handles virtual environments and dependencies.
3

Configure Your API Keys

Add your OpenRouter, Gemini, or DeepSeek keys to config.yaml.
4

Launch and Play

Run ./start.sh and enter your first narrative adventure.

Ready to start?

Follow our quickstart guide to get Flower Engine running in minutes.

Build docs developers (and LLMs) love

Get started for freeTalk to us