Matlantis Agent Skills
Japanese version / 日本語版
A plugin marketplace providing Matlantis / PFP atomistic simulation skills for Claude Code.
Overview
Matlantis Agent Skills is a collection of skills that enables Claude Code to assist with atomistic simulations on the Matlantis platform. 16 skill modules guide Python code generation for structure optimization, molecular dynamics, property calculations, and more — all powered by PFP (Preferred Potential) neural network potentials.
Each skill systematically covers implementation patterns, best practices, and common errors with solutions, all built on ASE (Atomic Simulation Environment).
Skills Catalog
| Category | Skill | Summary |
|---|
| Foundation | matlantis | Platform overview, PFP calculation mode selection, library architecture |
| Foundation | setup | Calculator initialization, calc_mode configuration, structure file I/O |
| Foundation | ase-basics | ASE fundamentals, Atoms object, molecule & crystal construction |
| Structure Building | modeling | Surface slab generation, SMILES→3D conversion, substitution/doping, liquids & amorphous |
| Structure Building | external-data | Structure retrieval from PubChem, Materials Project, etc.; file format conversion |
| Computation | optimization | Fixed-cell & variable-cell relaxation, symmetry preservation, multi-stage optimization |
| Computation | dynamics | NVE/NVT/NPT molecular dynamics, metadynamics, transport properties |
| Computation | reaction | NEB reaction path search, transition state search, activation energy |
| Computation | property-analysis | Elastic constants, phonons, QHA, diffusion coefficients, viscosity, thermal conductivity |
| Workflow | integrated-workflows | Structure screening, adsorption campaigns, defect analysis pipelines |
| Workflow | background-job | Background execution of long-running calculations, checkpoint management |
| Acceleration | light-pfp | Light-PFP custom models (15–40x faster) training & inference |
| Utility | pfcc-extras | Visualization, structure processing, and job execution control utilities |
| Utility | visualization | Atomic structure & trajectory visualization, nglview, POV-Ray |
| Infrastructure | storage | Group Drive persistent storage, team data sharing |
| Infrastructure | ssh | SSH connection setup, VSCode Remote-SSH, WinSCP integration |
Getting Started
Prerequisites
- Claude Code CLI installed
Installation
Run the following commands in the Claude Code REPL:
# 1. Register the marketplace
/plugin marketplace add https://github.com/matlantis/matlantis-agent-skills
Configuration
# 2. Open the plugin manager
/plugin
# 3. Enable the matlantis plugin from the "Marketplaces" tab
Once enabled, Claude Code will automatically reference the appropriate skills when you work with Matlantis-related code or questions.
Skills Details and Usage Examples
matlantis
Platform overview skill. Covers PFP calculation mode selection guidelines (PBE, R2SCAN, WB97XD, etc.), library hierarchy, and critical constraints.
"I want to calculate the surface energy of copper. Which calc_mode should I use?"
setup
Calculator initialization and environment setup. Covers Estimator/ASECalculator configuration patterns, structure file I/O, and batch execution settings.
"Write code to initialize the PFP Calculator in R2SCAN mode"
ase-basics
ASE fundamentals. Covers Atoms object creation/editing, molecule & crystal construction, periodic boundary conditions, and file I/O.
"Load a crystal structure from a CIF file and create a 2x2x2 supercell"
modeling
Structure modeling. Covers surface slab generation (ASE builders / pymatgen), 3D molecule generation from SMILES (RDKit), atomic substitution/doping, and construction of liquids, amorphous materials, and high-entropy alloys.
"Create a 4-layer FCC copper (111) surface slab with 15 Å of vacuum"
external-data
Structure retrieval from external databases. Covers integration with PubChem (organic molecules), Materials Project (inorganic crystals), AFLOW, COD, and file format conversion.
"Retrieve the BaTiO3 crystal structure from Materials Project and optimize it"
optimization
Structure optimization (relaxation). Covers fixed-cell optimization (atomic positions only) and variable-cell optimization (including lattice vectors), LBFGS/BFGS/FIRE optimizers, and symmetry-preserving optimization.
