hdf5-pde-data-loading

Patterns for loading PDE simulation datasets (PDEBench, PhiFlow, JAX-CFD) from HDF5 files. Handles layout detection (single tensor vs separate variables), spatial/temporal downsampling, multi-variable systems, HuggingFace and DaRUS data sources, and efficient PyTorch DataLoader creation. Use when preparing PDE data for neural operator training.

Quality

82%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

72%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The skill body is highly actionable with executable code throughout and is mostly token-efficient. Its main weaknesses are the absence of explicit validation feedback loops for batch downloads and the lack of any progressive disclosure into separate reference files despite a substantial length.

Suggestions

Add an explicit validate-and-retry checkpoint after DaRUS/HuggingFace downloads (e.g., verify file size or HDF5 integrity, re-download on failure) so batch download operations have a clear feedback loop.

Split the larger sections—DataLoader setup and the downsampling/pitfalls reference tables—into separate reference files (e.g., DATALOADER.md, DOWNSAMPLING.md) and point to them from SKILL.md for one-level-deep progressive disclosure.

Remove the leftover debug line ('print(f"Keys: {list(f.keys())}")') from the robust loading example to tighten conciseness.

Dimension	Reasoning	Score
Conciseness	The body is code-dense and assumes Claude's competence without explaining what HDF5 or DataLoader are, but a leftover debug line ('print(f"Keys: {list(f.keys())}")') and a few instructional asides prevent it from being fully lean.	4 / 5
Actionability	Provides fully executable, copy-paste-ready code across loading, HuggingFace download, aria2c DaRUS download, a Dataset subclass, and DataLoader setup, covering the common cases with concrete commands.	5 / 5
Workflow Clarity	The body is organized by topic rather than as a sequenced workflow, and batch/large-file download operations lack explicit validate-then-retry checkpoints (integrity is only mentioned in a pitfalls table), which caps workflow clarity at 3 per the destructive/batch guidance.	3 / 5
Progressive Disclosure	Content is well-sectioned with clear headers, but at ~163 body lines everything is inlined in a single SKILL.md with no one-level-deep references to separate files, so content that could be split (DataLoader setup, downsampling tables) remains inline rather than disclosed progressively.	3 / 5
	Total	15 / 20 Passed

Description

92%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is strong: it states concrete capabilities, names real data sources and benchmarks, and provides an explicit 'Use when' trigger. Its only gap is minor—lacking a few natural synonyms and the .h5 file extension that users might mention.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions—'layout detection (single tensor vs separate variables)', 'spatial/temporal downsampling', 'multi-variable systems', 'HuggingFace and DaRUS data sources', 'PyTorch DataLoader creation'—giving comprehensive coverage rather than minimal or generic actions.	5 / 5
Completeness	Explicitly answers 'what' (handles layout detection, downsampling, multi-variable systems, data sources, DataLoader creation) and 'when' with the concrete trigger clause 'Use when preparing PDE data for neural operator training.'	5 / 5
Trigger Term Quality	Includes natural terms users would say ('PDE simulation datasets', 'HDF5 files', 'neural operator training', named benchmarks PDEBench/PhiFlow/JAX-CFD) but omits common synonyms and file extensions like '.h5' or 'simulation data', so a few natural terms are missing.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche—loading PDE simulation datasets from HDF5 with named benchmarks—making it unlikely to trigger for unrelated skills and giving it minimal conflict risk.	5 / 5
	Total	19 / 20 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/data-engineering/hdf5-pde-data-loading/SKILL.md
Commit: 3a6c3a9

Reviewed: 1 day ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.