Designs and implements production-grade RAG systems by chunking documents, generating embeddings, configuring vector stores, building hybrid search pipelines, applying reranking, and evaluating retrieval quality. Use when building RAG systems, vector databases, or knowledge-grounded AI applications requiring semantic search, document retrieval, context augmentation, similarity search, or embedding-based indexing.
Detected risks:
The rag-architect skill is designed to help teams design and implement production-grade retrieval-augmented generation (RAG) systems for knowledge-grounded AI applications. It focuses on building scalable semantic search and document retrieval pipelines by guiding users through core architectural decisions such as vector database selection, document chunking strategies, embedding generation, indexing workflows, and retrieval optimization. The skill addresses common challenges in RAG development, including retrieval accuracy, latency constraints, metadata management, deduplication, and retrieval quality evaluation. It provides a structured workflow that emphasizes validation checkpoints between stages to reduce implementation errors and improve reliability in production environments.
The skill includes guidance for configuring vector stores, selecting embedding models, designing chunking strategies, and implementing retrieval pipelines with hybrid search and reranking techniques. It also covers evaluation and iteration practices for measuring retrieval quality and debugging retrieval performance. Reference materials are organized by topic, including vector databases, embedding models, retrieval optimization, chunking strategies, and RAG evaluation methodologies. Implementation examples demonstrate practical workflows using tools such as LangChain, OpenAI embeddings, and Qdrant, including metadata enrichment, deterministic document IDs, and vector indexing operations. The workflow encourages domain-specific tuning rather than relying on default settings.
This skill is intended for AI architects, machine learning engineers, platform engineers, and developers building semantic search systems, enterprise knowledge bases, AI assistants, or retrieval-powered applications. It is particularly useful for teams implementing vector databases, embedding-based indexing systems, or hybrid retrieval pipelines that require scalable and maintainable architecture decisions. Organizations building internal knowledge retrieval systems, document search platforms, or context-aware AI solutions can use this skill to standardize RAG design practices and improve retrieval performance over time.
This skill is intended for retrieval-augmented generation systems, semantic search platforms, vector database architectures, and knowledge-grounded AI applications that rely on document retrieval and embeddings.
The skill references vector database technologies including Pinecone, Weaviate, Chroma, pgvector, and Qdrant as part of its vector database guidance.
Yes. The skill includes implementation examples for document chunking, embedding generation, vector indexing, metadata handling, and deduplication workflows using Python-based tooling.
The skill covers chunking strategies, embedding selection, hybrid retrieval pipelines, reranking, query optimization, indexing strategies, and retrieval evaluation practices.
Yes. The workflow includes checkpoints and assertions for validating metadata completeness, deduplication integrity, and retrieval pipeline correctness before progressing to the next stage.
Quick Setup:
.claude/skills/Repository
jeffallan/claude-skills