Afrexai Rag Engineering

# RAG Engineering — Complete Retrieval-Augmented Generation System

> Build production RAG systems that actually work. From chunking strategy to evaluation — the complete methodology.

You are an expert RAG engineer. When the user needs to build, optimize, or debug a RAG system, follow this complete methodology.

---

## Phase 1: RAG Architecture Assessment

### Quick Health Check (Existing Systems)

| Signal | Healthy | Warning | Critical |
|--------|---------|---------|----------|
| Answer relevance | >85% users satisfied | 60-85% | <60% |
| Retrieval precision@5 | >70% relevant chunks | 40-70% | <40% |
| Hallucination rate | <5% | 5-15% | >15% |
| Latency (P95) | <3s | 3-8s | >8s |
| Context utilization | >60% of retrieved used | 30-60% | <30% |
| Cost per query | <$0.05 | $0.05-0.20 | >$0.20 |

### RAG Project Brief

```yaml
rag_brief:
  project: "[name]"
  date: "YYYY-MM-DD"

  # What problem are we solving?
  use_case: "[customer support / code search / document Q&A / research / legal / medical]"
  user_persona: "[who asks questions]"
  query_types:
    - factual: "[percentage] — direct fact lookup"
    - analytical: "[percentage] — synthesis across documents"
    - procedural: "[percentage] — how-to, step-by-step"
    - comparative: "[percentage] — compare X vs Y"
    - conversational: "[percentage] — multi-turn follow-ups"

  # What data do we have?
  corpus:
    total_documents: "[count]"
    total_size: "[GB/TB]"
    document_types:
      - type: "[PDF/HTML/markdown/code/JSON/CSV]"
        count: "[count]"
        avg_length: "[pages/tokens]"
    update_frequency: "[static / daily / real-time]"
    languages: ["en", "..."]
    quality: "[curated / mixed / noisy]"

  # Requirements
  accuracy_target: "[% — start with 85%]"
  latency_target: "[ms P95]"
    max_cost_per_query: "[$]"
  scale: "[queries/day]"
  multi_turn: "[yes/no]"
  citations_required: "[yes/no]"

  # Constraints
  deployment: "[cloud / on-prem / hybrid]"
  data_sensitivity: "[public / inter...

# RAG Engineering — Complete Retrieval-Augmented Generation System

> Build production RAG systems that actually work. From chunking strategy to evaluation — the complete methodology.

## What This Skill Does

Turns your AI agent into a RAG engineering expert. Covers the entire pipeline:

- **Architecture design** — Pattern selection, decision trees, project brief templates
- **Data ingestion** — Extraction, cleaning, metadata enrichment for every document type
- **Chunking strategy** — The #1 cause of bad RAG, solved with decision frameworks
- **Embedding selection** — Model comparison, benchmarking, cost optimization
- **Vector store setup** — Database selection, indexing strategy, hybrid search
- **Retrieval optimization** — Multi-query, HyDE, reranking, context assembly
- **Generation & prompting** — Grounding, citations, hallucination prevention
- **Evaluation framework** — Golden test sets, LLM-as-judge prompts, automated CI
- **Production deployment** — Security, caching, scaling, monitoring
- **Advanced patterns** — Agentic RAG, Graph RAG, CRAG, Self-RAG, multimodal

## Install

```bash
clawhub install afrexai-rag-engineering
```

## Quick Start

Tell your agent:
- "Design a RAG system for customer support documentation"
- "My RAG results are bad — help me diagnose"
- "Which embedding model should I use for code search?"
- "Help me set up RAG evaluation"

## What Makes This Different

- **37KB of pure methodology** — no API wrappers, no dependencies
- **Decision frameworks** for every choice (chunking, embeddings, vector DB, reranking)
- **Production-tested patterns** with specific thresholds and benchmarks
- **Complete evaluation system** with LLM-as-judge prompts and golden test set design
- **Diagnostic decision tree** for when things go wrong
- **Cost modeling** at every scale tier
- **100-point quality scoring rubric**

## ⚡ Level Up

Want industry-specific AI agent context? Our $47 context packs include complete agent configurations for your vertical:...