Rag

This handles the full RAG pipeline from document ingestion through retrieval, covering chunking strategies, embedding generation, vector storage, and query optimization. You get practical guidance on picking vector databases (Pinecone for production, Chroma for local dev) and embedding models based on your constraints. The examples show real LangChain4j code for common patterns like metadata filtering and multi-source retrieval. What's helpful here is the chunking advice, 500-1000 tokens with 10-20% overlap, which actually matters for quality. The security warnings about prompt injection in retrieved documents are worth reading. If you're building a chatbot over company docs or a semantic search system, this covers the decisions you need to make and includes enough error handling patterns to avoid the obvious mistakes.

Install to Claude Code

npx -y skills add giuseppe-trisciuoglio/developer-kit --skill rag --agent claude-code