rag-query

name	rag-query
description	Query a RAG (Retrieval-Augmented Generation) knowledge base to find relevant information from ingested documents. This skill should be used when the user wants to search within a loaded knowledge base or document collection.

RAG Query

Query the RAG knowledge base for semantically relevant documents.

Usage

python scripts/rag_query.py "<query>" [top_k]

Parameters

Parameter	Type	Required	Description
query	string	Yes	The search query
top_k	int	No	Number of results to return (default: 3)

Behavior

• Uses semantic similarity search against the FAISS vector store
• Returns matching document chunks with relevance scores
• Returns a message indicating the knowledge base is not initialized if no retriever is available
• Documents must be ingested first using the RAG pipeline

Notes

• The knowledge base must be loaded before querying
• Use the RAG retriever to ingest documents before running queries
• Embedding model: DashScope text-embedding-v3