ChemLint: Conversational Cheminformatics with Large Language Models.

We introduce ChemLint, an open-source Model Context Protocol (MCP) server that connects any MCP-compatible large language model to a curated suite of local cheminformatics and machine learning tools, enabling rigorous molecular data handling through a conversational interface. Molecular machine learning studies are frequently undermined by inconsistent data preprocessing, including invalid SMILES, unresolved duplicates, and train-test leakage, yet existing large language model (LLM)-based chemistry tools do not address these data-centric challenges. ChemLint provides tools for data exploration and diagnostics, molecule standardization, and machine learning modeling. All operations are executed deterministically by established libraries and recorded in a project manifest that traces every action, supporting reproducibility and making curation choices explicit. We demonstrate through several examples how ChemLint can be used to identify common data quality issues, evaluate splitting strategies, and execute complete modeling pipelines from raw data to evaluation. ChemLint is freely available at https://github.com/molML/ChemLint.