version: "1.0.0" name: vasp-to-isaac description: Convert VASP DFT calculations (slab or bulk) to ISAAC AI-ready records (v1.05). Handles IrOx surface slabs and ternary oxide bulk DOS calculations on NERSC Perlmutter. license: internal metadata: version: "1.0"

VASP to ISAAC Converter

Converts VASP DFT calculation directories to ISAAC v1.05 AI-ready JSON records. Two scripts handle two distinct cases — choose based on the geometry of the calculation.

Which Script to Use?

Quick check: look at the INCAR.

• NSW > 0 and slab structure → use slab script
• NSW = 0 and bulk crystal → use bulk script

Slab Script

Script: vasp_slab_to_isaac.py (located in scripts/ directory of this skill) Reference: references/slab_workflow.md

# Activate your project virtual environment
source .venv/bin/activate
 
# Minimal — facet/coverage inferred from directory names
python3 vasp_slab_to_isaac.py <vasp_dir> [output.json]
 
# With explicit labels
python3 vasp_slab_to_isaac.py <vasp_dir> output.json \
  --electrode-type anode \
  --surface-facet 101 \
  --surface-coverage 4O

Extracts: energy trajectory, per-atom forces, stress, magnetic moment, Hubbard-U params, slab thickness, vacuum, all VASP assets with SHA256.

Bulk Script

Script: vasp_bulk_to_isaac.py (located in scripts/ directory of this skill) Reference: references/bulk_workflow.md

source .venv/bin/activate
 
# Pass the material directory (auto-detects dos/ subdir)
python3 vasp_bulk_to_isaac.py <material_dir> [output.json]

Extracts: space group, mp_id, PBE vs PBE+U (from actual LDAUU values), DOS settings, energy/forces/stress, all VASP assets including DOSCAR and CHGCAR.

ASE DB Pipeline (bulk only)

# Step 1: build ASE db from raw VASP dirs
python3 make_ase_db.py output.db <dataset_dir>/*/
 
# Step 2: convert ASE db to ISAAC records
python3 ase_db_to_isaac.py output.db ./isaac_records/

Note: the ASE route loses INCAR-level detail (smearing, EDIFF, per-element U params). See references/bulk_workflow.md for the full comparison.

Batch Conversion

# Slab — all configs in a facet
for config in <config1> <config2>; do
  python3 vasp_slab_to_isaac.py <facet>/$config isaac_ready/<facet>_${config}_isaac.json \
    --electrode-type anode
done
 
# Bulk — all materials in a dataset directory
for d in <dataset_dir>/*/; do
  name=$(basename $d)
  python3 vasp_bulk_to_isaac.py $d isaac_ready/${name}_isaac.json
done

References

• references/isaac_schema.md — ISAAC v1.05 schema blocks, required fields, DFT method spec, how to pull latest from GitHub and update this skill
• references/cathub_organize.md — prerequisite for slab ASE db: how to organize raw VASP slabs with cathub organize and cathub folder2db
• references/slab_workflow.md — slab directory structure, CLI args, extracted fields, known quirks
• references/bulk_workflow.md — bulk directory naming, mp_id parsing, PBE+U detection, ASE db pipeline, route comparison
• references/dependencies.md — pymatgen, numpy, ASE, cathub install commands; venv location; POTCAR quirk

Scripts

Scripts are in the scripts/ directory of this skill (symlinked to the working copies in your project):

• vasp_slab_to_isaac.py — slab converter
• vasp_bulk_to_isaac.py — bulk converter