Does it include advanced search patterns?

Yes, it covers sophisticated patterns such as hybrid search (combining dense and sparse retrieval), multi-query expansion, and contextual compression.

Can I use this for source citations?

Yes, the skill provides specific prompt engineering templates and retrieval logic to ensure the system returns and cites source documents for transparency.

What vector databases does this skill support?

The skill provides implementation guidance for several leading vector stores including Pinecone, Weaviate, Milvus, Chroma, Qdrant, and FAISS.

What chunking strategies are included?

The skill includes patterns for recursive character splitting, token-based splitting, semantic chunking, and markdown header-aware splitting to preserve document context.

How does this skill help reduce LLM hallucinations?

It implements Retrieval-Augmented Generation (RAG) which grounds LLM responses in specific, retrieved context from your own documents, ensuring the AI only answers based on provided facts.

RAG Implementation

Name: RAG Implementation
Author: HermeticOrmus

byHermeticOrmus

データサイエンスとML

Implements Retrieval-Augmented Generation (RAG) systems with vector databases and semantic search to build grounded, knowledge-aware AI applications.

概要

The RAG Implementation skill provides a comprehensive framework for building production-grade Retrieval-Augmented Generation systems. It streamlines the process of integrating LLMs with external knowledge bases by offering standardized patterns for document chunking, embedding generation, and vector database configuration. Beyond basic retrieval, this skill guides developers through advanced optimization techniques like hybrid search, reranking, and contextual compression, ensuring that AI applications deliver accurate, fact-based responses while minimizing hallucinations through rigorous grounding and source citation.

主な機能

Implements advanced retrieval patterns like hybrid search and multi-query retrieval
Features evaluation metrics for measuring groundedness and retrieval quality
Supports multiple vector stores including Pinecone, Chroma, Weaviate, and FAISS
Includes reranking workflows using Cross-Encoders and Cohere for improved accuracy
0 GitHub stars
Provides diverse chunking strategies including semantic and recursive character splitting

ユースケース

Creating research assistants that provide verifiable citations for every answer
Implementing semantic search engines for large-scale natural language document discovery
Building automated Q&A systems over proprietary internal documentation

概要

主な機能

Implements advanced retrieval patterns like hybrid search and multi-query retrieval
Features evaluation metrics for measuring groundedness and retrieval quality
Supports multiple vector stores including Pinecone, Chroma, Weaviate, and FAISS
Includes reranking workflows using Cross-Encoders and Cohere for improved accuracy
0 GitHub stars
Provides diverse chunking strategies including semantic and recursive character splitting

ユースケース

Creating research assistants that provide verifiable citations for every answer
Implementing semantic search engines for large-scale natural language document discovery
Building automated Q&A systems over proprietary internal documentation