A Toolset-First Paradigm Based on Large Language Model and Agent via Model Context Protocol for Intelligent Computation
- 7 August 2026
- journal article
- Published by American Chemical Society (ACS) in The Journal of Physical Chemistry Letters
- Vol. 17 (33) , 9557-9567
Abstract
Large language models (LLMs) and agents have lately propelled advances in artificial intelligence for science (AI4S). In this work, we propose a toolset-first paradigm for intelligent computation that decouples the tools from the agents via the model context protocol (MCP). We construct four MCP servers, HEA-MCP, VASP-MCP, LAMMPS-MCP, and OpenClaw-MCP. HEA-MCP provides theoretical calculations for high-entropy alloys (HEA). VASP-MCP and LAMMPS-MCP provide first-principles calculations and molecular dynamics simulations, respectively. OpenClaw-MCP contains an autonomous agent, OpenClaw, which renders the autonomous agent a callable tool. OpenClaw itself also supports the MCP to form a recursive agent architecture. By packaging the tools as MCP servers and connecting them to a general MCP-compatible agent, domain-specific capabilities can be obtained without building dedicated agents and can be reused across agentic frameworks.Keywords
Funding Information
- Natural Science Foundation of Shanghai Municipality (23ZR1418200)
- National Natural Science Foundation of China (21703158)
- National Natural Science Foundation of China (21873073)
- National Natural Science Foundation of China (22273023)
- National Natural Science Foundation of China (92477103)
- Science and Technology Commission of Shanghai Municipality (25511102400)
- Natural Science Foundation of Zhejiang Province (LMS26B030003)
- Natural Science Foundation of Chongqing Municipality (CSTB2023NSCQ-MSX0616)
This publication has 40 references indexed in Scilit:
- New Opportunities For the Integration of Artificial Intelligence With Materials Science: From Large Language Models to Embodied Large ModelsMaterials Genome Engineering Advances, 2026
- TopoMAS: Large Language Model Driven Topological Materials Multi‐Agent SystemMaterials Genome Engineering Advances, 2026
- Machine Learning Accelerated Diamine/Tertiary‐Amine Mixtures Design for CO 2 CaptureCarbon Neutralization, 2025
- T2MAT (text-to-material): A universal agent for generating material structures with goal properties from a single sentenceActa Physico-Chimica Sinica, 2025
- Bridging language models and computational materials science: A prompt‐driven framework for material property predictionMaterials Genome Engineering Advances, 2025
- Application of machine learning for material prediction and design in the environmental remediationChinese Chemical Letters, 2024
- Quantifying the advantage of domain-specific pre-training on named entity recognition tasks in materials sciencePatterns, 2022
- LAMMPS - a flexible simulation tool for particle-based materials modeling at the atomic, meso, and continuum scalesComputer Physics Communications, 2022
- A thermodynamic description of the Co-Cr-Fe-Ni-V system for high-entropy alloy designCalphad, 2019
- Computational Thermodynamics2007