Skills

Complete framework for an AI agent to architect, implement, and iteratively improve a production-grade Retrieval Augmented Generation pipeline from initial requirements through deployment and continuous optimization.

When to use

• Designing a new RAG system architecture from scratch
• Building a knowledge retrieval pipeline that feeds context into an LLM
• Creating or selecting a vector database and search system for document retrieval
• Implementing document ingestion, chunking, and embedding workflows
• Improving an existing RAG system that suffers from poor retrieval quality, hallucinations, or latency issues
• Reducing hallucinations in a knowledge-based AI system by grounding responses in retrieved evidence
• Integrating external or enterprise knowledge sources (documents, databases, APIs) into AI-generated responses
• Evaluating retrieval accuracy, ranking quality, or end-to-end answer correctness of a RAG pipeline
• Optimizing context window usage, retrieval speed, or embedding efficiency in a live RAG deployment
• Migrating a naive "stuff all documents into the prompt" approach to a scalable retrieval-based architecture

Instructions