🌌 Project Tavern: AI-Driven Roleplay Interface

A high-fidelity, multimodal roleplay interface designed for the next generation of LLM-driven storytelling. Create complex narratives, manage detailed character profiles, and engage in immersive conversations within a premium glass-morphic environment.

---

✨ Key Features

🏺 Character Library & Profile Management

• Persistent Storage: Powered by IndexedDB, the Tavern bypasses the 5MB limits of standard local storage. Manage extensive character libraries and infinite message histories with stability.
• Spec V2 Character Support: Seamlessly import characters from Tavern V2 .png cards or .json files.
• Profile Editor: Design high-fidelity AI Characters (NPCs) or User Personas.
• Auto-Identity Alignment: The Tavern automatically applies your most recently used User Persona when a new conversation begins.

🖼️ High-Fidelity Image Generation

• ComfyUI Integration: Includes a built-in proxy for ComfyUI. Generate stable, high-fidelity character portraits and scene imagery directly within the chat.
• Automated Prompting: LLM-driven generation of visual character descriptions to ensure structural consistency across all portraits.
• Persistent Image Storage: Generated images are automatically saved to your local database as Base64 assets.

🗣️ Voice & Message Engine

• Universal API Proxy: Robust support for OpenAI-compatible APIs (Z.ai, OpenRouter, local Llama.cpp).
• Text-to-Speech (TTS): High-fidelity voice output via Kokoro-TTS, ElevenLabs, or standard WebSpeech APIs.
• Dynamic Styling: Immersive storytelling with automatic italicized narration and bold-face dialogue.

---

🛠️ Installation Guide

1. Prerequisites

Ensure your local machine has the following requirements:

• Node.js (v18.0 or higher)
• npm (comes with Node)
• Recommended: A local LLM server (Llama.cpp, KoboldCPP) or a ComfyUI installation for image generation.

2. Clone the Repository

git clone https://github.com/your-username/project-tavern.git
cd project-tavern

3. Install Dependencies

npm install

4. Run the Application

npm run dev

Access the application at http://localhost:3000.

---

📦 Containerization (Docker)

For those who wish to run Project Tavern within an isolated container, Docker compose and standalone options are available.

Deploy via Docker Compose (Recommended)

This is the simplest method to setup Project Tavern and its dependencies.

1. Start Deployment:

   docker compose up -d --build

2. Access the Interface: The application will be available at http://localhost:3000.

Build Standalone Docker Image

If you prefer to build the container manually:

1. Build the Image:

   docker build -t project-tavern .

2. Run the Container:

   docker run -p 3000:3000 --name project-tavern project-tavern

---

⚙️ Configuration

1. Settings: Configure your AI model providers (OpenAI, OpenRouter, etc.) and ComfyUI endpoints.
2. Persistence: All character profiles and chat histories (IndexedDB) are saved to your browser's persistent storage.
3. Voice Settings: WebSpeech is available natively; Kokoro-TTS and other providers require external API nodes (configurable in Settings).

---

📜 Technical Stack

• Framework: Next.js 16 (App Router & Turbopack)
• Language: TypeScript
• Database: IndexedDB
• Styling: Vanilla CSS (Glassmorphism & Theme Support)
• Icons: Lucide React

---

⚖️ License

Distributed under the MIT License. Use responsibly.

---

🍵 Support the Project

If Project Tavern has added value to your workflow, consider supporting the project. Your contributions ensure the continued development of this interface.

BTC Address: bc1qpexehwqnm4huxztxn0kk33a2ps8gxj3rlef7xf

---

"Great stories start with a single prompt. Every message a memory, every silence a scene."