DingyangLyu

DingyangLyu / MatClaw

Public

MatClaw: an open materials-science agent that turns natural-language tasks into reproducible simulation workflows.

github.combjzgcai ai-agents ai4materials material-science
23
2
100% credibility
Found Mar 10, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

MatClaw is an AI agent that autonomously runs materials science simulations like DFT, molecular dynamics, and Monte Carlo from natural language descriptions in chat apps.

How It Works

1
🔍 Discover MatClaw

You hear about an AI helper that turns everyday questions about materials into real simulation results, shared on GitHub.

2
📦 Get it ready

Download the ready-to-use package that has all the science tools built in, so you can start helping right away.

3
💬 Link your chat

Connect it to your favorite messaging app like Feishu or DingTalk, so you can talk naturally from your phone or computer.

4
🧑‍🔬 Ask your first question

Just describe a materials challenge like 'find the density of water' and watch it plan, calculate, and create charts automatically.

5
⏱️ Check progress anytime

Use simple chat commands to peek at what's happening, pause, or switch conversations without losing your place.

6
📊 See amazing results

Get back plots, data summaries, and insights delivered right to your chat, saving you hours of manual work.

🚀 Transform your work

Now you tackle complex simulations effortlessly, speeding up discoveries in batteries, catalysts, and more.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 23 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is MatClaw?

MatClaw is a TypeScript-based AI agent for materials science that turns natural-language tasks—like "calculate silicon's band gap"—into reproducible simulation workflows. You chat via messaging apps (Feishu, DingTalk, WhatsApp, etc.) or Docker stdin, and it autonomously writes scripts, runs DFT/MD/MC in isolated containers with pre-installed QE, LAMMPS, RASPA3, and ML potentials, then delivers plots and analysis. A web dashboard at localhost:3210 shows live logs and transcripts.

Why is it gaining traction?

It packs a full materials-science stack into one GPU-ready Docker image, with 221 built-in skills covering everything from phonons to catalysis, plus VASP integration via SSH. Chat commands like /watch, /status, and /resume make monitoring effortless, and it retries errors autonomously for reliable results. Multi-channel access and extensible skills beat manual scripting or fragmented tools.

Who should use this?

Materials scientists prototyping electronic structures, defects, or adsorption isotherms without writing input files. Computational chemists screening catalysts or batteries via natural language. Researchers bridging wet lab ideas to simulations fast, especially those on HPC with VASP.

Verdict

Try it if you're in materials science—impressive scope and reproducibility for an early project (17 stars, 1.0% credibility). Docs are thorough with examples matching QUASAR benchmarks, but watch for maturity as adoption grows.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.