Harnessing AtomisticSkills for Agentic Atomistic Research

2026-05-18

AI Summary

Atomic-scale scientific research faces significant challenges due to its inherent complexity and fragmented tooling, rendering general-purpose large language models inadequate for tasks demanding high rigor. This work proposes AtomisticSkills, a novel hierarchical skill framework tailored for materials and chemical research. It decomposes the scientific workflow into over one hundred modular, extensible, and plug-and-play atomic-level skills, integrating capabilities such as database querying, thermodynamic and kinetic modeling, machine-learned interatomic potentials (MLIPs), density functional theory (DFT) simulations, and multimodal analysis. The framework enables AI agents to collaboratively orchestrate and autonomously conduct research. Its broad applicability and powerful compositional capability have been demonstrated across diverse tasks—including solid-state electrolyte design, MOF screening, MLIP optimization, virtual drug screening, XRD interpretation, and catalyst discovery—laying a foundational infrastructure for autonomous AI scientists.

Abstract

Computational materials science and chemistry span vast knowledge domains and fractured software ecosystems. Although large language models (LLMs) have demonstrated research capabilities, scaling monolithic agents to manage the rigor and complexity of atomistic research remains a challenge. Here, we introduce AtomisticSkills, an open-source harness framework that empowers general-purpose AI coding agents to conduct atomistic research across materials science, chemistry, and drug discovery. By hierarchically decomposing scientific workflows into agent skills and tools, AtomisticSkills provides agents with modular, extensible, and plug-and-play research capabilities. The framework integrates more than 100 human-curated multidisciplinary skills, including database access, thermodynamics and kinetics modeling, and diverse simulation engines employing machine learning interatomic potentials (MLIPs) and density functional theory (DFT). We validate its functional coverage against scientific literature and demonstrate robust orchestration capabilities across diverse scientific campaigns: generative design of Li-ion solid-state electrolytes, high-throughput screening of metal-organic frameworks for CO2 capture, autonomous MLIP benchmarking and fine-tuning, multi-stage structure-based virtual screening for drug design, multimodal X-ray diffraction pattern analysis, and screening of Fe-oxide catalysts for oxygen evolution reaction. AtomisticSkills provides a critical agent infrastructure towards building fully autonomous AI scientists.

Problem

• atomistic research
• AI agents
• computational materials science
• scientific workflows
• autonomous AI scientists

Innovation

• AtomisticSkills
• AI agents
• modular scientific workflows
• machine learning interatomic potentials
• autonomous research