diceframe

DiceFrame is a self-hostable AI tabletop RPG engine supporting D&D / CoC / custom rules with a multiplayer Web UI.

Automatically checked · GitHub checked 2026-11-13

diceframe

Overview

DiceFrame Logo

DiceFrame

English | 简体中文

GitHub Stars GitHub Release License

DiceFrame Web UI preview

DiceFrame is a self-hostable AI tabletop RPG engine supporting D&D / CoC / custom rules with a multiplayer Web UI.

A play round: type an action, reveal an automatic dice check, and receive the GM reply Real Web UI; demonstration game with scripted check results and GM replies.

Features

  • Play solo or with friends in a browser, with invitations and shared game state.
  • Describe actions naturally; rules handle automatic dice checks and character state.
  • Build worlds, lorebooks and adventures, or import scripts, PDFs and supported modules.
  • Reuse character cards and create worlds, rules and characters with AI assistance.
  • Keep long campaigns with summaries, optional memory recall and text-to-speech.
  • Add plugins and chat bots; use the owner-only DF Agent for reviewed operations.

Quick start

Windows portable

Download the installer DiceFrame-vX.Y.Z-windows-setup.exe (no administrator rights needed) or the portable DiceFrame-vX.Y.Z-windows-portable.zip (extract it and run DiceFrame.exe) from Releases. Add a provider under Manage → Settings → Model API, then assign the main model under Model Routing. Installation details · First game

Docker

From a cloned repository or extracted source release containing docker-compose.yml:

cp .env.example .env
# edit .env as needed
docker compose pull
docker compose up -d

Open http://localhost:9876, then configure your provider and model in Settings. Data stays in data/. Docker details and updates

From source

Requires Python 3.10+, Node.js 20.19+ or 22.12+, and an OpenAI-compatible Chat Completions endpoint.

git clone https://github.com/diceframe/diceframe.git
cd diceframe/frontend-v2
npm ci && npm run build
cd ..
pip install -r requirements.txt
python web_server.py

Open http://localhost:18000, then configure your provider and model in Settings. Source and standalone frontend details

Documentation

Documentation · Website · Issues · Contributing · AGPL-3.0