LangChain and LlamaIndex

@knolo/langchain@6.2.0 and @knolo/llamaindex@6.2.0 are retriever adapters. They depend on @knolo/core ^6.0.0. They do not call a chat model and they do not add a vector database.

@knolo/semantic-ollama@6.2.0 is an optional embedding provider for the V4 hybrid layer. It talks to a local Ollama server only when the application uses it. Semantic rerank still runs after lexical retrieval. See Semantic reranking.

Use these packages when an existing LangChain or LlamaIndex pipeline should retrieve from a mounted Knolo pack. Keep answer generation, tools, and credentials in the host. A retriever hit is not a KAR evaluation and it is not a Cortex trace.

Examples in the Knolo Core repository start at examples/langchain-basic.