About
The Agent Orchestrator is a professional-grade Python framework implementing the Model Context Protocol (MCP). It establishes a structured, decoupled client-server architecture over Standard Input/Output (StdIO), allowing LLMs to seamlessly integrate with diverse external tools and knowledge sources. This includes real-time data fetching, such as weather information via external APIs, and an advanced Retrieval-Augmented Generation (RAG) system capable of ingesting PDF and TXT documents, utilizing ChromaDB for persistent storage, and optimizing searches with Maximal Marginal Relevance (MMR). Its intelligent client bridge handles asynchronous lifecycle management, automatic tool schema conversion for OpenAI-compatible function calling, and persistent conversation states for multi-turn reasoning.
