UdaPlay

•

Answers questions about video games by leveraging a local vector database and AI agentic workflow with tool-calling.

UdaPlay is an AI Research Agent designed to answer complex questions about video games. Built as an end-to-end project, it showcases an agentic workflow that combines a local vector database (ChromaDB) for game data, an OpenAI-compatible LLM for understanding and generation, and a state machine for orchestrating tool calls like semantic search and web search. It also includes a bonus MCP (Model Context Protocol) server and client, demonstrating an alternative approach to tool execution for enhanced flexibility.

主な機能

01Local Vector Database (ChromaDB) for efficient game knowledge retrieval
02OpenAI-compatible LLM integration with robust tool-calling capabilities
03Optional MCP server and client for external tool exposure and delegation
04Session memory for maintaining conversation context across interactions
050 GitHub stars
06AI Agent with State Machine orchestration for dynamic workflows

ユースケース

01Rapidly retrieve information and answer complex queries about video games
02Demonstrate and build agentic AI workflows for intelligent data retrieval
03Create an intelligent chatbot capable of utilizing external tools for research