skills.rest
Tgoldi
TgoldiCOMMUNITY
Explore all skills by this author →
GitHub

rag-engineer

Designs retrieval-augmented generation pipelines with chunking, embeddings, and hybrid search strategies.

1|Updated May 10, 2026
Search Tags:#semantic search#embeddings#llm#rag#vector search#hybrid search#chunking
One-click install
npx skills add https://github.com/Tgoldi/claude-skills --skill rag-engineer-tgoldi
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: rag-engineer
Source: https://github.com/Tgoldi/claude-skills/tree/main/rag-engineer
Command: npx skills add https://github.com/Tgoldi/claude-skills --skill rag-engineer-tgoldi

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Building RAG systems that return relevant context instead of hallucinations is hard: poor chunking, wrong embedding models, and missing reranking silently degrade answer quality. This Skill provides expert guidance on designing retrieval pipelines where retrieval quality drives generation quality. ## Core Features & Use Cases - Semantic Chunking Strategies: Chunk documents by meaning and structure rather than fixed token counts, preserving context across boundaries. - Hierarchical & Hybrid Retrieval: Combine multi-level indexing, BM25 keyword search, vector similarity, and Reciprocal Rank Fusion for better precision. - Sharp Edges Reference: A severity-ranked table of common RAG failures (fixed-size chunking, missing reranking, stale embeddings) with concrete solutions. - Use Case: You are building a document Q&A feature over a company knowledge base. Use this Skill to choose chunking boundaries, select embedding models per content type, add a reranking step, and set up retrieval evaluation separate from generation. ## Quick Start Ask the AI to design a RAG pipeline for your document search application, including chunking strategy, embedding model selection, and hybrid retrieval with reranking.

Frequently Asked Questions about rag-engineer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build a RAG pipeline for document search?▼

Start with semantic chunking that respects document structure, generate embeddings, and index them in a vector database. Add hybrid search combining BM25 keyword matching with vector similarity, then rerank results before passing them to the LLM.

What chunking strategy works best for RAG systems?▼

Semantic chunking outperforms fixed token-count splitting because it respects sentence boundaries, paragraphs, and topic shifts. Detect topic changes with embedding similarity, preserve headers, and add overlap for context continuity.

Should I use hybrid search or pure vector search for retrieval?▼

Hybrid search is recommended for most workloads. Combine BM25 or TF-IDF keyword matching with vector similarity, merge scores using Reciprocal Rank Fusion, and tune weights based on query type.

Why does my RAG system return irrelevant context?▼

Common causes include fixed-size chunking breaking sentences, missing metadata pre-filtering, no reranking step, and stale embeddings after source documents change. Measure retrieval quality separately from generation to isolate the failure.

When should I not use the same embedding model for all content?▼

Different content types such as code, prose, and tables often embed poorly under a single model. Evaluate embedding models per content type and refresh embeddings whenever source documents change.

Related Skills

Explore More Software Engineering Skills→
ShipFailShipFail
1

Memory Manager

Chunk text and documents into semantic pieces with metadata for Vector.Store.

Official
Intermediate
HouseGarofaloHouseGarofalo
3

langchain

Develop LLM applications with LangChain for chaining calls, prompts, agents, and retrieval.

Official
Advanced
agentscope-aiagentscope-ai
809

rag-eval

Diagnose RAG pipelines by separating retrieval quality from generation faithfulness.

Official
Intermediate
ricablericable

peft-fine-tuning

Fine-tune large language models with LoRA and QLoRA adapters.

Community
Advanced