mcp-dm-gemini

# TTRPG Assistant MCP Server This project is a Model Context Protocol (MCP) server designed to assist with Tabletop Role-Playing Games (TTRPGs). It provides an AI assistant with access to parsed TTRPG rulebooks and campaign management capabilities, accessible via a web interface, a command-line tool, and a Discord bot. ## Features * **Advanced Search:** Perform hybrid semantic and keyword searches with query suggestions and completion for rules, spells, monsters, and items. * **Campaign Management:** Store, retrieve, and manage campaign-specific data using ChromaDB's document storage. * **Adaptive PDF Processing:** Add new TTRPG rulebooks with intelligent content classification that learns patterns over time. * **LLM Personality Configuration:** Automatically extracts a "personality" from each rulebook to configure the LLM's voice and style. * **Character Generation:** Tools to help you create player characters and generate rich backstories. * **NPC Generation:** A flexible tool for generating non-player characters with stats appropriate to the player characters' level. * **Map Generation:** Generate simple SVG maps for combat encounters. * **Content Packs:** Create and share content packs with other users. * **Web Interface:** A user-friendly web UI for interacting with all the major features of the assistant. * **Interactive CLI:** Interactive sessions for character and NPC creation for a guided experience. * **Discord Bot:** Interact with the TTRPG Assistant through a Discord bot. * **MCP Integration:** Exposes a standardized MCP interface for AI assistants with full feature parity between FastMCP and MCP protocol servers. * **Enhanced Features:** Session management, search statistics, result explanations, and cross-platform configuration support. * **Export/Import:** Export and import campaign data for sharing and backup. ## Documentation For detailed instructions on how to install and use the TTRPG Assistant, please refer to our comprehensive documentation: * **[Installation Guide](docs/installation.md)** * **[Usage Guide](docs/usage.md)** ## Getting Started ### Option 1: Docker (Recommended) ```bash docker-compose up ``` ### Option 2: Manual Setup ```bash ./bootstrap.sh ``` After setup, you have multiple ways to use the TTRPG Assistant: - **For Claude Desktop integration:** `./mcp_stdio.sh` - **For direct testing:** `./run_main.sh` - **For web UI:** `python mcp_server.py` then visit [http://localhost:8000/ui](http://localhost:8000/ui) The system uses ChromaDB for local vector storage - no external database required! For more detailed installation instructions, please refer to the [Installation Guide](docs/installation.md). ## Troubleshooting **Connection refused error when running the CLI:** This usually means that the MCP server is not running. Make sure you have started the server by running `docker-compose up` or `./bootstrap.sh`. **ChromaDB connection error:** This means that there's an issue with the ChromaDB database. ChromaDB is an embedded database that doesn't require a separate server. If you encounter issues, ensure that the application has write permissions to the data directory (default: `./chroma_db`). ## Architecture Changes **🔄 Major Update: ChromaDB Migration** This version has migrated from Redis to ChromaDB for improved vector search performance and simplified deployment: - **No external database required** - ChromaDB runs embedded within the application - **Enhanced search capabilities** - Hybrid semantic + keyword search with query suggestions - **Adaptive PDF processing** - Learns content patterns for better classification over time - **Improved session management** - Uses ChromaDB's document storage for campaign data - **Cross-platform configuration** - Automatic config file discovery across Windows, Linux, and macOS ## New Features ### Advanced Search - **Hybrid Search**: Combines semantic similarity with keyword matching - **Query Suggestions**: Get completion suggestions while typing - **Search Explanations**: Understand why results were returned - **Search Statistics**: Monitor search performance and usage ### Enhanced PDF Processing - **Adaptive Learning**: System learns content type patterns from your rulebooks - **Pattern Caching**: Reuses learned patterns for faster processing - **Enhanced Metadata**: Better content classification and section detection ### MCP Protocol Support - **Full Feature Parity**: Both FastMCP and standard MCP servers support all features - **Cross-Platform Scripts**: `mcp_stdio.sh/.bat` for Claude Desktop integration - **Direct Testing**: `run_main.sh/.bat` for FastMCP server testing