Does it handle large document sets?

Yes, it includes context window management logic to truncate and prioritize documents based on relevance and token budgets to fit within model limits.

Does this support hybrid search?

Yes, it includes Reciprocal Rank Fusion (RRF) patterns to combine the strengths of semantic vector search and traditional keyword search.

What is a RAG retrieval skill for Claude Code?

It is a set of specialized patterns that allow Claude to fetch external documents and use them as context, ensuring responses are grounded in specific data rather than just training knowledge.

How does it prevent AI hallucinations?

The skill utilizes a 'Sufficiency Check' pattern that validates if retrieved context is enough to answer a query before the LLM generates a response, allowing the model to abstain when data is missing.

Can I use this for production AI applications?

Absolutely. The skill includes patterns for FastAPI, structured outputs, and token management designed for production-grade AI pipelines.

RAG Retrieval Patterns

Name: RAG Retrieval Patterns
Author: yonatangross

byyonatangross

•

Data Science & ML

Implements robust Retrieval-Augmented Generation (RAG) patterns to ground LLM responses with accurate, cited, and validated external data.

This skill provides production-ready patterns for building advanced RAG pipelines within Claude Code, focusing on factual accuracy and hallucination prevention. It includes implementations for hybrid search—combining semantic and keyword results—alongside automated citation generation and context window optimization. A key highlight is the context sufficiency check, which evaluates whether retrieved information is adequate before generation begins, significantly reducing model hallucinations and improving response reliability in data-heavy environments.

Key Features

01Sufficiency check patterns to prevent LLM hallucinations

02Context window management and token budget optimization

0369 GitHub stars

04Hybrid search implementation using Reciprocal Rank Fusion (RRF)

05Automated inline citation tracking and source attribution

06Production-ready templates for basic and advanced RAG workflows

Use Cases

01Implementing search systems that combine keyword precision with semantic understanding

02Building enterprise chatbots that require grounded answers from internal documentation

03Creating automated research assistants that provide verifiable source citations

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add yonatangross/orchestkit rag-retrieval

For use in Claude.ai and ChatGPT

Download Skill