ray-data

Scalable data processing for ML workloads. Streaming execution across CPU/GPU, supports Parquet/CSV/JSON/images. Integrates with Ray Train, PyTorch, TensorFlow. Scales from single machine to 100s of nodes. Use for batch inference, data preprocessing, multi-modal data loading, or distributed ETL pipelines.

Quality

82%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

76%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.

A highly actionable, well-structured reference-style skill with abundant executable examples and clean one-level-deep references. Its main weaknesses are absent validation checkpoints in batch workflows (capping workflow clarity) and duplication between the body and the reference files.

Suggestions

Add validation/verification steps to the batch-inference and preprocessing-pipeline examples (e.g., assert row counts before/after transforms, inspect output schema before writing) to lift workflow clarity above 3.

De-duplicate the body and reference files: replace the inlined GPU-transform and Ray Train integration sections with pointers to references/transformations.md and references/integration.md so each topic lives in one place.

Trim promotional padding—the "Performance benchmarks" and "Use cases" company name-drops and the redundant opening line—to improve token efficiency.

Dimension	Reasoning	Score
Conciseness	The body is mostly lean code examples that assume Claude's competence, but it carries minor trimmable padding: the redundant opening line, promotional "Performance benchmarks" and "Use cases" company name-drops, and content duplicated from the reference files.	4 / 5
Actionability	Extensive copy-paste-ready, executable code spans reading, transformations, GPU transforms, writing, batch inference, and PyTorch/TensorFlow integration, covering the common cases concretely.	5 / 5
Workflow Clarity	Sequences like the batch-inference and multi-step preprocessing pipelines are present, but these batch operations lack validation/verification checkpoints (e.g., verify output schema, check row counts), so the batch-operation cap of 3 applies.	3 / 5
Progressive Disclosure	Two real, clearly-signaled one-level-deep references (transformations.md, integration.md) exist and the body is well-sectioned, but the body duplicates material also found in those reference files (GPU transforms, Ray Train integration), a minor organization gap below 5.	4 / 5
	Total	16 / 20 Passed

Description

88%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.

A strong, third-person description that concretely states capabilities and gives an explicit "Use for..." trigger clause. Voice and completeness are exemplary; only trigger-term synonym coverage and minor overlap with general data-processing skills hold it back from a perfect score.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete capabilities and actions—"Streaming execution across CPU/GPU", "supports Parquet/CSV/JSON/images", "Integrates with Ray Train, PyTorch, TensorFlow", "batch inference, data preprocessing, multi-modal data loading, or distributed ETL pipelines"—giving comprehensive coverage; not below 5 since no meaningful action is missing.	5 / 5
Completeness	Clearly answers "what" (scalable streaming data processing across CPU/GPU with format/framework integrations) and explicitly answers "when" via the "Use for batch inference, data preprocessing, multi-modal data loading, or distributed ETL pipelines" trigger clause.	5 / 5
Trigger Term Quality	Natural user phrases like "batch inference", "data preprocessing", "multi-modal data loading", and "ETL pipelines" are present, but a few common synonyms/variations are missing, so it sits above the 3 anchor but below a fully comprehensive 5.	4 / 5
Distinctiveness Conflict Risk	The Ray-specific niche (Ray Train, CPU/GPU streaming) is clearly distinct, but generic triggers like "data preprocessing" and "distributed ETL pipelines" carry minor overlap risk with Pandas/Spark/Dask skills, keeping it just below a 5.	4 / 5
	Total	18 / 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: OpenLAIR/dr-claw
Path: skills/data-processing/ray-data/SKILL.md
Commit: 4b35756

Reviewed: 5 days 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.