cp2k-dft-preparation

star 9

Use this skill for source-grounded CP2K conventional DFT preparation in materials workflows: single-point, fixed-cell geometry optimization, cell optimization, frequency, and DOS-style stages.

q734738781 By q734738781 schedule Updated 6/9/2026

name: cp2k-dft-preparation description: "Use this skill for source-grounded CP2K conventional DFT preparation in materials workflows: single-point, fixed-cell geometry optimization, cell optimization, frequency, and DOS-style stages." allowed-tools: "cp2k_prepare remote_submission remote_submission_batch get_avail_remote_task"

cp2k-dft-preparation

Overview

Use this skill when the task is a conventional CP2K DFT calculation stage owned by materials_worker. AIMD belongs to dynamics_worker.

Quick Start

  1. Verify the source structure path.
  2. Use cp2k_prepare with recipe="sp", geo_opt, cell_opt, freq, or dos.
  3. Verify the prepared stage contains job.inp and manifest.json.
  4. Submit with remote_submission(task_name="cp2k_execute") or remote_submission_batch for first-level batch children.
  5. Parse accepted results with task-specific scripts or mature libraries; do not assume a universal CP2K analyzer.

Allowed tools

  • cp2k_prepare
  • remote_submission
  • remote_submission_batch
  • get_avail_remote_task

Workflow

1. Keep CP2K conventional DFT in materials_worker

  • Use this skill for single point, geometry optimization, cell optimization, frequency, DOS/PDOS, and population-analysis setup.
  • Do not use it for AIMD thermostat, barostat, trajectory, or restart workflows.

2. Prepare one clean stage per calculation

  • cp2k_prepare writes job.inp, input.xyz, and manifest.json.
  • recipe="sp" prepares an ENERGY_FORCE style calculation.
  • recipe="geo_opt" prepares a fixed-cell geometry optimization.
  • recipe="cell_opt" prepares a cell optimization and requests stress handling.
  • recipe="freq" prepares a vibrational-analysis stage.
  • recipe="dos" prepares a DOS-oriented energy stage with property output enabled.

3. Keep settings controlled

  • Use the settings object only for supported knobs such as xc, basis_set, potential, charge, multiplicity, cutoff, rel_cutoff, eps_scf, max_scf, kpoints, periodic, cell_abc, dispersion, convergence tolerances, and properties.
  • Project policy: do not insert arbitrary CP2K blocks through the LLM. If an unsupported section is required, write a reviewed local recipe or ask for confirmation.

4. Submit through the generic execute task

  • Use only task_name="cp2k_execute" for CP2K execution.
  • Do not invent per-recipe task names such as cp2k_geo_opt_execute.
  • For a batch root, every first-level child must already be a complete CP2K stage.

Method-critical defaults

  • Parameter priority: honor explicit user requirements first; otherwise choose CP2K settings overrides from the system class and task objective; if that judgment remains uncertain, run a narrow literature or official documentation check before finalizing the override.
  • Do not add CP2K settings overrides just to restate the tool baseline; only override when the user, system class, task objective, or a checked source justifies it.
  • Report the chosen XC functional, basis, potential, cutoff, charge, multiplicity, periodicity, and k-points when they affect comparisons.
  • Treat frequency, DOS/PDOS, and charge analysis as follow-up stages from an accepted structure, not as automatic analysis of every CP2K run.
  • For surfaces or low-dimensional systems, make periodic explicit rather than silently relying on automatic inference.

Output Contract

Return:

  • prepared CP2K stage path
  • selected recipe
  • manifest.json path
  • cp2k_execute receipt/context when submitted
  • any unsupported setting or parser limitation

References

Install via CLI
npx skills add https://github.com/q734738781/CatMaster --skill cp2k-dft-preparation
Repository Details
star Stars 9
call_split Forks 1
navigation Branch main
article Path SKILL.md
More from Creator