RAG Implementation Workflow

Overview

Specialized workflow for implementing RAG (Retrieval-Augmented Generation) systems including embedding model selection, vector database setup, chunking strategies, retrieval optimization, and evaluation.

When to Use This Workflow

Use this workflow when:

• Building RAG-powered applications
• Implementing semantic search
• Creating knowledge-grounded AI
• Setting up document Q&A systems
• Optimizing retrieval quality

Workflow Phases

Phase 1: Requirements Analysis

Skills to Invoke

• ai-product - AI product design
• rag-engineer - RAG engineering

Actions

1. Define use case
2. Identify data sources
3. Set accuracy requirements
4. Determine latency targets
5. Plan evaluation metrics

Copy-Paste Prompts

Use @ai-product to define RAG application requirements

Phase 2: Embedding Selection

Skills to Invoke

• embedding-strategies - Embedding selection
• rag-engineer - RAG patterns

Actions

1. Evaluate embedding models
2. Test domain relevance
3. Measure embedding quality
4. Consider cost/latency
5. Select model

Copy-Paste Prompts

Use @embedding-strategies to select optimal embedding model

Phase 3: Vector Database Setup

Skills to Invoke

• vector-database-engineer - Vector DB
• similarity-search-patterns - Similarity search

Actions

1. Choose vector database
2. Design schema
3. Configure indexes
4. Set up connection
5. Test queries

Copy-Paste Prompts

Use @vector-database-engineer to set up vector database

Phase 4: Chunking Strategy

Skills to Invoke

• rag-engineer - Chunking strategies
• rag-implementation - RAG implementation

Actions

1. Choose chunk size
2. Implement chunking
3. Add overlap handling
4. Create metadata
5. Test retrieval quality

Copy-Paste Prompts

Use @rag-engineer to implement chunking strategy

Phase 5: Retrieval Implementation

Skills to Invoke

• similarity-search-patterns - Similarity search
• hybrid-search-implementation - Hybrid search

Actions

1. Implement vector search
2. Add keyword search
3. Configure hybrid search
4. Set up reranking
5. Optimize latency

Copy-Paste Prompts

Use @similarity-search-patterns to implement retrieval

Use @hybrid-search-implementation to add hybrid search

Phase 6: LLM Integration

Skills to Invoke

• llm-application-dev-ai-assistant - LLM integration
• llm-application-dev-prompt-optimize - Prompt optimization

Actions

1. Select LLM provider
2. Design prompt template
3. Implement context injection
4. Add citation handling
5. Test generation quality

Copy-Paste Prompts

Use @llm-application-dev-ai-assistant to integrate LLM

Phase 7: Caching

Skills to Invoke

• prompt-caching - Prompt caching
• rag-engineer - RAG optimization

Actions

1. Implement response caching
2. Set up embedding cache
3. Configure TTL
4. Add cache invalidation
5. Monitor hit rates

Copy-Paste Prompts

Use @prompt-caching to implement RAG caching

Phase 8: Evaluation

Skills to Invoke

• llm-evaluation - LLM evaluation
• evaluation - AI evaluation

Actions

1. Define evaluation metrics
2. Create test dataset
3. Measure retrieval accuracy
4. Evaluate generation quality
5. Iterate on improvements

Copy-Paste Prompts

Use @llm-evaluation to evaluate RAG system

RAG Architecture

User Query -> Embedding -> Vector Search -> Retrieved Docs -> LLM -> Response
                |              |              |              |
            Model         Vector DB     Chunk Store    Prompt + Context

Quality Gates

• Embedding model selected
• Vector DB configured
• Chunking implemented
• Retrieval working
• LLM integrated
• Evaluation passing

Related Workflow Bundles

• ai-ml - AI/ML development
• ai-agent-development - AI agents
• database - Vector databases

Limitations

• Use this skill only when the task clearly matches the scope described above.
• Do not treat the output as a substitute for environment-specific validation, testing, or expert review.
• Stop and ask for clarification if required inputs, permissions, safety boundaries, or success criteria are missing.