Overview

ChemGraph

ChemGraph is an agentic framework that can automate molecular simulation workflows using large language models (LLMs). Built on top of LangGraph and ASE, ChemGraph allows users to perform complex computational chemistry tasks, from structure generation to thermochemistry calculations, with a natural language interface.

ChemGraph

ChemGraph supports diverse simulation backends, including ab initio quantum chemistry methods (e.g. coupled-cluster, DFT via NWChem, ORCA), semi-empirical methods (e.g., XTB via TBLite), and machine learning potentials (e.g, MACE, UMA) through a modular integration with ASE.