Materials Simulation Skills plugin
24 skills for numerical methods and materials science: the agent checks stability, meshes, convergence and cluster runs.
- 83 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- Skills run Python scripts on your machine and read simulation input files
- Generated SLURM scripts submit jobs to a cluster only after you launch them
Install
- Works in
Manual install
npx skills add HeshamFS/materials-simulation-skillsInstalls the skills into `~/.agents/skills/`, which many agents read.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Skills run Python scripts on your machine and read simulation input files
- Generated SLURM scripts submit jobs to a cluster only after you launch them
This is third-party code. Review the repository files before installing.
What it does
The set packages a simulation engineer expertise into skills with tested scripts. The agent finds the right one by task, runs the script and explains the result. For a time step stability question, for example, it calls `cfl_checker.py` and reports the CFL number and the allowed step. Skills fall into eight groups: numerical foundations (stability, time stepping, meshing, convergence, integration, differentiation, solvers), the simulation workflow, SLURM script generation for clusters, verification and validation, FAIR packaging, cross-code failure triage and the CMSO and ASMO ontologies. A separate `skill-evaluator` skill evaluates any other skill in any agent CLI. The repository also ships its own `mss` command to list, install bundles, run scripts and checks.
Who it is for. Simulation engineers and materials researchers who run models with an agent and do not want to explain CFL, convergence and cluster quotas each time.
Good fit when
- You must know whether an explicit step `dt=0.001` is stable for an advection problem at 2 m/s with mesh step 0.01
- A run on three meshes gave stress 98.5, 99.6 and 99.9 MPa and you must check whether the mesh reached the asymptotic range
- A problem needs a SLURM script with reasonable resources and a diagnosis of why the last run failed
- Results should be packaged for reproducibility under FAIR with metadata
Not a fit when
- You need to run LAMMPS or DFT solvers themselves: interfaces to them are on the author roadmap and not ready
- You need a physical conclusion from a result: script checks do not prove the model is right, and limits are in `docs/VALIDATION_SCOPE.md`
- You need to work without Python: the scripts target Python 3.10 to 3.12
Example request
Check whether dt=0.001 is stable for an advection problem with v=2.0 m/s and dx=0.01 and give the largest allowed step.Limitations
The scripts are written for Python 3.10 to 3.12. The author stresses that deterministic checks confirm the accuracy of script output and that an improvement in agent scientific correctness with these skills is not proven. Before treating a passing check as evidence about a physical model, read `docs/VALIDATION_SCOPE.md`. Interfaces to LAMMPS and DFT are still on the roadmap. SLURM script generation prepares files; running them and cluster quotas stay with you.
How to disable. Delete the installed skill folders from `~/.agents/skills/` or your agent directory. For the Claude Code plugin run `/plugin uninstall` and remove the marketplace if it is no longer needed.
FAQ
How do I install the set?
With `npx skills add HeshamFS/materials-simulation-skills` for any agent, with a pinned version via `gh skill install HeshamFS/materials-simulation-skills --pin v1.0.0`, or in Claude Code through `/plugin marketplace add HeshamFS/materials-simulation-skills`.
What does the mss command do?
`mss list` shows skills, `mss bundles` and `mss install --agent claude --bundle <name>` install a bundle into the chosen agent, `mss run` runs a script and `mss eval` runs the deterministic checks.
How many checks cover the set?
In the 20 October 2026 evaluation 115 cases, 131 checks and 725 assertions passed. The author notes that this verifies scripts and is no proof that agent answers improve.
Does the set fit Cursor or Codex?
Yes. The skills follow the open Agent Skills standard and install to `~/.agents/skills/`, where Codex, Cursor, Copilot and other agents read them.
What is skill-evaluator?
It is a skill from the Meta group: it evaluates any other skill in any agent CLI by comparing runs with and without it.
Discussion
No questions or reviews yet.
Related
PluginsPlugin with 130-plus skills for atomistic materials research: relaxation, dynamics, DFT and structure databases through MCP.
Skills for computational chemistry: ASE, DFT codes, LAMMPS, DeePMD, RDKit, Open Babel and more, built for OpenClaw
Around 200 skills for Claude Code on bioinformatics, chemistry and biostatistics, from RNA-seq analysis to molecular docking
17 skills for computational biology: RNA-seq, single-cell, protein structure, docking, literature and figures
A skill for math modeling contests CUMCM, MCM/ICM and the Electrical Cup: 10 stages, three roles, five gates and a LaTeX paper
Eight skills for math modeling contests that carry a task from the problem to a paper with figures and checked references
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Materials Simulation Skills (this card) | no | 83 | ||
| StatsPAI: causal inference and econometrics for agents | no | not required | 374 | the MCP server and skill of StatsPAI for econometrics rather than numerical simulation methods |
| Feynman | yes | 10,116 | a terminal agent for scientific literature reviews, with no calculation scripts |
README in short
The description explains why an agent needs numerical methods knowledge and shows a dialogue checking a CFL number. It then lists quality checks: script tests, full deterministic eval coverage and explicit validation limits. Install, use, the table of eight categories, contributing and documentation follow.