rag-agent

name	rag-agent
description	Provides a template for building Retrieval-Augmented Generation (RAG) agents. Use for: creating agents that can access and reason over external knowledge bases, semantic search, and contextual retrieval.

RAG Agent Skill

The Retrieval-Augmented Generation (RAG) Agent template is designed for building agents that can access and reason over external knowledge bases.

Key Features

• Semantic Search: Integration with vector databases (e.g., Pinecone, Milvus, Chroma) for efficient and relevant information retrieval.
• Contextual Retrieval: Dynamic retrieval of relevant documents based on user queries, ensuring the agent has the most pertinent information.
• Source Attribution: Mechanisms to ensure the agent cites its sources for transparency and verifiability.

Usage Instructions

To utilize this RAG agent skill, follow these steps:

1. Ingest Data: Load and chunk your documents into manageable pieces.
2. Generate Embeddings: Convert text chunks into vector representations suitable for semantic search.
3. Query Agent: The agent will automatically retrieve relevant context from the knowledge base before generating a response to a user query.

Example Implementation Sketch

class RAGAgent:
    def __init__(self, vector_db, llm):
        self.vector_db = vector_db
        self.llm = llm

    def query(self, question):
        context = self.vector_db.search(question)
        return self.llm.generate(question, context)

Resources

• Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks