Running Computational Chemistry Workflows with Agent Skills

Where this fits

A researcher using OpenClaw to automate molecular simulations often gets stuck on repetitive steps: preparing input files, parsing output logs, and converting between formats. The Computational Chemistry Agent Skills collection packages 34 ready-to-use skills that handle these tasks, so you can focus on the science instead of the plumbing.

Why agents benefit

• Execute multi-step workflows (e.g., geometry optimization → frequency calculation → property extraction) without manual intervention.
• Parse output files from common quantum chemistry packages (ORCA, Gaussian, etc.) into structured data.
• Generate input files with correct basis sets, functionals, and convergence criteria.
• Handle file conversions (e.g., .xyz to .sdf) and coordinate manipulations.
• Provide reproducible, versioned skills with a DOI for citation in publications.

Practical scenarios

Automated geometry optimization and property prediction

An agent can take a SMILES string, generate a 3D structure, run a geometry optimization, then compute HOMO-LUMO gaps and dipole moments—all in one session.

High-throughput screening of candidate molecules

For a library of 1000 molecules, the agent can loop through each, run a quick semi-empirical calculation, and rank results by a target property, saving days of manual work.

Reproducible reaction path analysis

An agent can set up a transition state search, run the calculation, and extract activation energies, ensuring the same parameters are used every time.

Add it to your agent workflow

Install with:

npx skills add jinzhezenggroup/computational-chemistry-agent-skills

Then in OpenClaw, you can invoke a skill like run_geometry_optimization with a molecule file:

from skills import geometry_optimization
result = geometry_optimization.run("input.xyz", method="B3LYP", basis="6-31G*")
print(result.energy)

The skills are documented on the project site and include a DOI for citation.

Compare before adopting

Before committing, compare this collection with other chemistry agent skills on: maintenance freshness (last commit, open issues), quality signals (stars, DOI, documentation), and workflow fit (does it support your specific quantum chemistry package?). Also check if the skills are modular enough to extend to your own pipelines.

Why it is worth tracking

This repository stands out because it has a DOI (10.1021/acs.jctc.6c00622) and 128 GitHub stars, indicating peer-reviewed backing and community adoption. The skills are actively maintained and cover a wide range of computational chemistry tasks. If you're building a chemistry-aware agent, this is a solid foundation to evaluate—and you can find more context on OpenAgentSkill for a deeper dive.