Tools/LangChain
LangChain
Framework for chaining LLM calls with tools, retrieval, and memory — prompt templates, document loaders, and pluggable vector-store/retriever abstractions (Qdrant, etc.). Predates the current "agent harness" framing (see Agent_Harness); most production teams now use it as the retrieval/tooling layer underneath a more structured orchestration layer like LangGraph rather than for the whole agent loop.