Can I use this with local LLMs?

Yes, the skill includes specific configurations for using Ollama as a provider for both embedding models and generation models.

What is the primary benefit of using RAG?

RAG grounds LLM responses in specific, external context, which significantly reduces hallucinations and allows the model to answer questions based on your private or updated documents.

Which vector databases does this skill cover?

The skill provides implementation patterns for Pandas DataFrames (for simple use cases), ChromaDB (for persistence), and FAISS (for high-performance similarity search).

How do I fix poor retrieval results?

The skill suggests several strategies including increasing chunk overlap, reducing chunk size, trying different embedding models, or increasing the 'k' value in your retriever.

RAG Implementation Guide

Name: RAG Implementation Guide
Author: atrawog

byatrawog

Data Science & ML

Implements Retrieval-Augmented Generation workflows to ground AI responses with external knowledge and private documents.

About

This skill provides a comprehensive framework for building Retrieval-Augmented Generation (RAG) pipelines within Claude Code, enabling developers to ground LLM responses in domain-specific data. It covers the complete technical stack including document chunking strategies, embedding generation, vector store management with Pandas, ChromaDB, or FAISS, and the construction of conversational RAG chains using LangChain. It is essential for building accurate Q&A systems, reducing hallucinations, and allowing AI models to interact with up-to-date information not present in their training data.

Key Features

Conversational RAG pipelines with memory and multi-turn support
Performance troubleshooting for retrieval quality and latency
Embedding generation using OpenAI and Ollama APIs
Vector storage implementation via Pandas, ChromaDB, and FAISS
Advanced document chunking (fixed-size, sentence-based, and overlapping)
0 GitHub stars

Use Cases

Creating a factual Q&A system for internal company wikis and PDFs
Developing research tools that summarize and query large sets of external papers
Building a documentation assistant for private or proprietary codebases

About

Key Features

Conversational RAG pipelines with memory and multi-turn support
Performance troubleshooting for retrieval quality and latency
Embedding generation using OpenAI and Ollama APIs
Vector storage implementation via Pandas, ChromaDB, and FAISS
Advanced document chunking (fixed-size, sentence-based, and overlapping)
0 GitHub stars

Use Cases

Creating a factual Q&A system for internal company wikis and PDFs
Developing research tools that summarize and query large sets of external papers
Building a documentation assistant for private or proprietary codebases