llama-farm LlamaFarm RAG Pipeline Skill
Deep guidance for configuring RAG (Retrieval-Augmented Generation) pipelines.
When to Load
Activate this skill when:
- User asks about RAG configuration
- User needs help with chunking, embeddings, or retrieval
- User is troubleshooting RAG quality issues
- User wants to optimize RAG performance
RAG Pipeline Overview
Files → Parser → Chunks → Embedder → Vector Store → Retriever → LLM
↓
Extractors → Metadata
Quick Reference
Chunk Size Guidelines
| Document Type | Chunk Size | Overlap |
|---|---|---|
| Legal/Regulatory | 1000-1200 | 150 |
| Technical Docs | 600-800 | 80-100 |
| Code | 500-600 | 50 |
| Notes/Markdown | 400-500 | 40-50 |
| Large PDFs | 900-1100 | 140-160 |
Top-K Guidelines
| Use Case | Top-K |
|---|---|
| Quick Q&A | 3-5 |
| Research | 8-10 |
| Legal/Compliance | 10-15 |
| Code Search | 5-8 |
Progressive Disclosure
Load these files for detailed guidance:
- chunking.md - Chunking strategies and configuration
- embeddings.md - Embedding model selection and setup
- retrieval.md - Retrieval strategy comparison
- troubleshooting.md - Diagnosing and fixing RAG issues