This repository provides ready-to-use Docker Compose configurations for orchestrating AI agents and LLM runtimes. It includes examples for multiple agent frameworks such as A2A multi-agent systems, Agno, and Vercel AI-SDK, combined with local open-source models (llama3.2, qwen3) served via Docker Model Runner or llama.cpp. The setup integrates MCP (Model Context Protocol) tools and supports running agent workflows either locally with GPU or via cloud services.