qm-dft — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited qm-dft (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Quantum mechanics-based methods compute electronic structure explicitly — enabling bond breaking/forming, spectroscopic properties, and accurate energetics beyond force fields. Python ecosystem: ORCA (best free QM, subprocess), xTB/tblite (fast semi-empirical, Python API), PySCF (pure Python, scriptable).
| Method | Cost | Accuracy | Use case |
|---|---|---|---|
| GFN-FF | O(N²) | ~MM | Pre-screening, conformers |
| GFN2-xTB | O(N²·8) | Good | Conformers, pre-opt, pKa |
| r²SCAN-3c | O(N³) | Very good | Routine geometry opt |
| B3LYP-D3BJ/def2-SVP | O(N⁴) | Good | Drug-like molecules opt |
| B3LYP-D3BJ/def2-TZVP | O(N⁴) | Better | Single-point on opt geom |
| ωB97X-D/def2-TZVP | O(N⁴) | Very good | Reaction barriers, CT states |
| DLPNO-CCSD(T)/CBS | O(N⁵⁺) | Benchmark | High-accuracy energetics |
# xTB geometry optimization (fastest QM-level method)
import subprocess
result = subprocess.run(
['xtb', 'mol.xyz', '--opt', '--gfn', '2', '--alpb', 'water'],
capture_output=True, text=True, cwd='workdir/'
)
# Output: xtbopt.xyz (optimized), xtbopt.log
# Parse final energy
for line in result.stdout.split('\n'):
if 'TOTAL ENERGY' in line:
energy = float(line.split()[3]) # Hartree
print(f"E = {energy:.8f} Eh")# ORCA single-point DFT (via subprocess)
orca_input = """\
! B3LYP D3BJ def2-SVP TightSCF
%pal nprocs 4 end
%maxcore 2000
* xyzfile 0 1 mol.xyz
"""
with open('sp.inp', 'w') as f:
f.write(orca_input)
result = subprocess.run(['orca', 'sp.inp'], capture_output=True, text=True)
# Parse with chem_qm.py: python chem_qm.py --parse sp.out| Task | Reference |
|---|---|
| DFT functionals, basis sets, dispersion, Jacob's ladder | references/dft-theory.md |
| ORCA: input syntax, optimization, freq, NMR, TD-DFT, output parsing | references/orca-practical.md |
| xTB/GFN2: CLI, Python (tblite), CREST, solvation, pKa | references/xtb-semiempirical.md |
| PySCF: Python QM, HF/DFT/MP2/CCSD, NMR, ESP charges | references/pyscf-python.md |
| Standard recipes: opt→freq, RESP, UV-Vis, barriers, NBO | references/common-workflows.md |
| Tool | Version | Install | Role |
|---|---|---|---|
| ORCA | 6.0 | orca-forum.org (free) | General QM: DFT, MP2, CCSD(T), TD-DFT |
| xTB | 6.7 | conda install -c conda-forge xtb | Fast semi-empirical |
| tblite | 0.3 | pip install tblite | xTB Python API |
| CREST | 3.0 | conda install -c conda-forge crest | Conformer/ensemble search |
| PySCF | 2.7 | pip install pyscf | Python-native QM |
| Psi4 | 1.9 | conda install -c conda-forge psi4 | Python QM + MP2/CCSD |
ase — structure building, ASE-driven optimization with ORCA/xTB calculatorsforce-fields — pre-optimize with MM before QM; GAFF2 validationdocking — QM refinement of top docking poseschem_qm.py — ORCA/Gaussian input gen + output parsing (ALKYL native)scientific-skills:rowan — cloud QM (DFT, pKa, Chai-1) without local ORCA install~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.