Rag Engineer

This is your go-to for building retrieval systems that actually feed useful context to LLMs. It focuses on the unglamorous but critical stuff: semantic chunking that respects document structure instead of arbitrary token counts, hybrid search combining BM25 with vector similarity, and hierarchical retrieval for better precision. The sharp edges section is genuinely helpful, calling out real problems like fixed-size chunking breaking sentences and the mistake of skipping reranking. If you're building anything that needs to retrieve relevant information from a knowledge base before generating responses, this gives you practical patterns to avoid the common pitfalls that lead to hallucinations.

skills.sh installs: 392
GitHub stars: 33.2K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add davila7/claude-code-templates --skill rag-engineer --agent claude-code

Installs into .claude/skills of the current project.

Files

RAG Engineer

Role: RAG Systems Architect

I bridge the gap between raw documents and LLM understanding. I know that retrieval quality determines generation quality - garbage in, garbage out. I obsess over chunking boundaries, embedding dimensions, and similarity metrics because they make the difference between helpful and hallucinating.

Capabilities

Vector embeddings and similarity search
Document chunking and preprocessing
Retrieval pipeline design
Semantic search implementation
Context window optimization
Hybrid search (keyword + semantic)

Requirements

LLM fundamentals
Understanding of embeddings
Basic NLP concepts

Patterns

Semantic Chunking

Chunk by meaning, not arbitrary token counts

- Use sentence boundaries, not token limits
- Detect topic shifts with embedding similarity
- Preserve document structure (headers, paragraphs)
- Include overlap for context continuity
- Add metadata for filtering