Automate design, construction, and evaluation of RAG systems.
npx skills add https://github.com/asfbay-bit/opchain-skills --skill oc-rag-forge
Please help me install this Agent Skill. Skill: oc-rag-forge Source: https://github.com/asfbay-bit/opchain-skills/tree/main/skills/oc-rag-forge Command: npx skills add https://github.com/asfbay-bit/opchain-skills --skill oc-rag-forge
This Skill provides a comprehensive solution for designing and building Retrieval-Augmented Generation (RAG) systems, addressing the complexities of vector database selection, embedding models, chunking strategies, hybrid search, and retrieval evaluation.
To start using the oc-rag-forge skill, initiate the /oc-rag command to design, build, and evaluate a RAG system.
High-intent search queries and answers about installing and using this skill.
To design and evaluate a RAG system with hybrid search, you need to integrate dense and BM25 search modes, select an appropriate embedding model, and apply a chunking strategy to optimize retrieval performance and measure quality.
Choosing a vector database for a semantic search system involves comparing options like pgvector, Turbopuffer, Pinecone, and Supabase Vectors based on your specific scalability needs, existing infrastructure, and retrieval evaluation metrics.
Chunking strategy affects retrieval performance in RAG applications by determining how source documents are segmented, which directly impacts the precision of the dense and BM25 hybrid search results returned to the embedding model.
You can use pgvector and Pinecone for building a knowledge base, but the choice depends on your architecture; pgvector integrates with existing PostgreSQL environments while Pinecone offers a managed, specialized vector search experience.
You need an embedding model from providers like Cohere, Voyage, or OpenAI to build a RAG pipeline, as these models convert text chunks into vector representations required for the underlying semantic search retrieval process.
You should use hybrid search instead of dense search alone in a RAG system when your knowledge base requires matching specific keywords alongside semantic context, a process measurable through dedicated retrieval evaluation techniques.
lucifertrjClean, tokenize, and normalize English text with NLTK for NLP pipelines.
h4vzzOptimize raw context chunks for AI models within a token budget.
PheonixCodderEnables semantic and hybrid search in PostgreSQL using vector embeddings and SQL functions.
zechenzhangAGIManage a serverless vector database with hybrid search and metadata filtering.
