SKILL.md
536 lines · 2,123 words
RAG Architect
Overview
Retrieval-Augmented Generation (RAG) enhances LLM outputs by grounding them in external knowledge. A well-architected RAG system requires careful decisions across ingestion, indexing, retrieval, and generation stages. This skill covers production-grade RAG design from document processing through quality assessment.
RAG Architecture Layers
+-----------------+ +------------------+ +----------------+
| Ingestion | --> | Indexing | --> | Retrieval |
| - Parse docs | | - Embed chunks | | - Query embed |
| - Chunk text | | - Store vectors | | - Vector search|
| - Extract meta | | - Build indices | | - Re-rank |
+-----------------+ +------------------+ +----------------+
|
v
+----------------+
| Generation |
| - Pack context |
| - Prompt LLM |
| - Cite sources |
+----------------+