@o-lang/semantic-doc-search

Semantic document retrieval engine for O-Lang workflows.

This package provides vector-based document search (RAG retrieval layer) that integrates with O-Lang kernel workflows. It handles document ingestion, chunking, embedding, and similarity search, returning LLM-ready context outputs.

Features

• Semantic vector search using embeddings
• Document ingestion from local filesystem (.txt, .md)
• Automatic text chunking for large documents
• Pluggable embedding providers (local, OpenAI, Groq, etc.)
• Multiple vector database support:
  • In-memory store
  • Redis (adapter)
  • PostgreSQL / pgvector (adapter)
  • Pinecone (adapter)
• Embedding cache support (embeddings.json)
• Normalized LLM-ready output format (text + matches)
• Designed for O-Lang .ol workflow integration

Installation

npm install @o-lang/semantic-doc-search