Elementary Introduction
Goal
In this section, you’ll gain practical experience with two key features of the Llama Stack platform: Retrieval-Augmented Generation (RAG) and tool integration using the MCP Python SDK. These modules teach you how to combine external knowledge and custom logic to build intelligent, agent-driven workflows.
Overview
This section is divided into two focused modules:
-
Retrieval-Augmented Generation with Llama Stack
Learn how to set up a basic RAG pipeline using an in-memory vector database. You’ll ingest documents, perform semantic retrieval, and enable your Llama Stack agents to reference external content for more informed responses. Swap out the simple database for a production-ready solution with minimal code changes. -
Tool Integration with MCP Weather server
Run a simple weather MCP server and expose it as a tool using the MCP protocol. This module demonstrates how to run MCP Servers and integrate them into your Llama Stack workflows to extend agent capabilities. -
ReACT example with MCP
This module demonstrates how use ReACT reasoning with MCP Servers and integrate them into your Llama Stack.
Next Step
Start with Retrieval-Augmented Generation with Llama Stack to enhance your agents with dynamic access to external knowledge!