rag-vector-db

name	rag-vector-db
description	Retrieval-Augmented Generation implementation using vector databases, embedding models, and query routines.
allowed-tools	Read Write Edit Bash
license	MIT license
metadata	{"skill-author":"Lord1Egypt"}

Rag Vector Db

Overview

RAG enhances LLM outputs by retrieving relevant documents from an external vector index based on query embeddings.

When to Use This Skill

Use to build custom Q&A pipelines over proprietary documentation files or PDF sets.

Quick Start (with runnable code examples)

# Simple cosine similarity matching using numpy (conceptual vector lookup)
import numpy as np

def cosine_similarity(v1, v2):
    return np.dot(v1, v2) / (np.linalg.norm(v1) * np.linalg.norm(v2))

v_query = np.array([0.1, 0.2, 0.9])
v_doc = np.array([0.11, 0.19, 0.88])
print("Match Score:", cosine_similarity(v_query, v_doc))

Advanced Usage

Integrate ChromaDB/Pinecone clients, slice texts using recursive chunking algorithms, and build reranking layers.

Key References

• LlamaIndex Documentation

Dependencies

• numpy>=1.20.0