First-Class Local AI & Embeddings
Connect local LLMs via Ollama, LM Studio, or vLLM, run pure-local ONNX embeddings for semantic memory, or play instantly with zero-GPU built-ins that require no setup.
Local-First · AI-Powered · Custom Worlds & Rules
LocalRPG
Craft original worlds with AI-assisted generation, build custom tabletop rules with automated test scenarios, and embark on solo campaigns with local AI as a first-class citizen. LocalRPG brings your adventures to life with multi-voice narration, speech-to-text dictation, style-pack scene imagery, and shareable visual novel exports.
- Local AI First-class local & ONNX embeddings
- Voices Narration, NPC speech & STT dictation
- Theatre Replay & export web/video bundles
Capabilities
Everything your campaign needs to come alive
From custom mechanics to living lore, LocalRPG gives you complete creative freedom. Bring your own local AI models, immerse yourself in spoken dialogue and generated imagery, and capture your story forever.
AI World Generation & Living Lore
Author factions, locations, and starter lore by hand, or generate complete, cohesive settings with schema-first AI pipelines, batch entity expansion, and draft review.
Custom Systems & Scenario Testing
Design dice mechanics, outcome tiers, and character creation prompts with worked examples in Systems Studio, and prove system balance with automated test scenarios.
Dynamic Scene Art & Style Packs
Bring scenes to life with curated art style packs and procedural vector art, preserving permanent world aesthetics while reflecting dynamic character appearance changes.
TTS Narration & Voice Dictation
Hear prose narrated aloud while NPCs speak with distinct voice profiles and pitch. Take your turns hands-free using click-to-talk speech-to-text dictation.
Story Theatre, Packages & Export
Relive adventures in visual novel Theatre Mode, share worlds and rules via standalone content packages, and export complete campaigns as web bundles or rendered video.
The look
Glassmorphic panels, translucent and quiet
A tour of the studios, the theatre and the codex.
Launcher Hub
Campaigns, worlds, and systems dock with quick-start creation wizards, ready for immediate play.
Worlds Studio
Author factions, lore, and entity trees by hand or generate complete settings with AI pipelines.
Systems Studio
Craft dice mechanics and action modes with worked examples, and verify balance with test scenarios.
Settings Studio
Configure local LLMs, ONNX embeddings, multi-voice audio, and media engines across structured tabs.
Codex Drawer
Inspect wikilinked entity lore, track dynamic character appearances, and customize voice profiles.
Story Theatre
Replay campaign scenes as an interactive visual novel, spoken in each character's distinct voice.
Get started
From a clone to a campaign
Launch the desktop app
# Fetch dependencies and build
mise run setup
mise run build
# Launch the desktop app
bin/localrpg guiConfigure your preferred local AI models in Settings Studio, or jump straight in with the built-in zero-GPU storytelling engine.
Create a world and play
# Or export an existing campaign as a standalone web bundle
bin/localrpg export web my-campaign --out dist/
# Or render your campaign into a shareable video
bin/localrpg export video my-campaign --out campaign.webmDesign custom worlds and mechanics in the built-in studios, start a new campaign, and export your journey when complete.
Documentation
The same guide the app ships
These articles are read straight out of the repository and embedded into the binary, so what you read here is what the Help panel shows in-game.
Start Here
Core Concepts
- Architecture & Core Concepts Schema-agnostic design, three-tier separation, and how AI agents drive tabletop adventures.
- Worlds & Lore Anatomy of a world, starting location resolution, and system overrides.
- Systems & Mechanics Rules prompts, action modes, dice expressions, and the sandboxed JavaScript mechanics engine.
- Campaigns & Turns The turn lifecycle, canonical history log, non-destructive rewinding, and living-world progression.
Configuration & Providers
- AI & Media Providers Configuring LLMs, voice synthesis, speech recognition, image generation, and spend tracking.
- AI Agents & Roles Assigning models to GM, Narrator, and Extractor roles, plus token budgets and fallback chains.
- Storage & Directories Standard XDG directories, workspace mode, and disposable SQLite caching.
- Usage, Cost & Pricing Reading the spend ledger and setting the provider prices that turn usage into a cost figure.
- Provider & Model Catalogue Every registered provider ID, its presets and models, and the ledger key that prices match against.
- Configuration Reference Every key LocalRPG reads from config.yaml, with the accepted values for the enumerated ones.
- Style Packs Theme the procedural art and the app chrome with a pack, or keep the built-in look.
- Trace & Telemetry Record what a turn did in a local JSONL trace, or export spans and metrics to an OpenTelemetry collector.
Studio Guides
- Building Worlds Studio Guide Walkthrough for creating custom worlds, defining lore prompts, and authoring starter entities.
- Building Custom Systems Guide Authoring system rules prompts, coding mechanics.js hooks, and testing dice expressions.
- AI World Generation Draft a world from a sentence, a folder, or a set of pages, then review it before it is committed.
Codex & Content Reference
Local AI & Self-Hosting
- Setting Up Ollama for Local LLMs Hardware requirements, native/Docker setup, recommended models for tabletop RPGs, and connecting Ollama to LocalRPG.
- Setting Up Kokoro-FastAPI for Neural TTS Running Kokoro-FastAPI via Docker (CPU/GPU), voice profile selection, and connecting to LocalRPG.
- Setting Up Fish Audio S2 with vLLM-Omni GPU requirements, Docker/vLLM-Omni container setup, voice cloning, and how emotion tags steer the performance.
- Setting Up Faster-Whisper for Speech Recognition OpenAI-compatible speech-to-text server setup via Docker, hardware considerations, and LocalRPG voice input configuration.
- Setting Up ComfyUI for Local Image Generation ComfyUI API configuration, model setup (SDXL/Flux/SD1.5), and connecting LocalRPG for scene and entity illustration.
- Fully Local AI Stack with Docker Compose One Docker Compose project that runs Ollama, Kokoro, Faster-Whisper, and ComfyUI together, with recommended models, system requirements, and the matching LocalRPG configuration.
- Local Semantic Search with a Built-in Encoder How LocalRPG turns entities and turns into searchable vectors, why the default projection ranks unrelated words together, and how to switch on the built-in BGE encoder that separates them.
- Local-First and Offline Play What runs locally in LocalRPG, what hardware each tier requires, and how to configure a zero-network tabletop stack.