nemo-curator

GPU-accelerated data curation for LLM training. Supports text/image/video/audio. Features fuzzy deduplication (16× faster), quality filtering (30+ heuristics), semantic deduplication, PII redaction, NSFW detection. Scales across GPUs with RAPIDS. Use for preparing high-quality training datasets, cleaning web data, or deduplicating large corpora.

Quality

80%

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

68%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 content is highly actionable with executable examples throughout, but it is padded with duplicated benchmark data and a cost-comparison section, and its batch pipeline lacks the validation/verification checkpoints the rubric requires for destructive operations.

Suggestions

Add explicit validation/verification steps to the pipeline (e.g., check row counts before/after each filter, sample-inspect dropped documents) so the batch workflow has feedback checkpoints.

Remove the duplicated performance figures — keep a single benchmark section and delete the restated 16×/120h/7.5h numbers from the 'GPU acceleration' table, or move all benchmarks to a reference file.

Move the full multimodal curation code and cost-comparison into reference files, keeping SKILL.md as a lean overview with one-level-deep links.

Dimension	Reasoning	Score
Conciseness	Mostly efficient code but includes redundant material — the 'GPU vs CPU' table and a separate 'Performance benchmarks' section restate the same 16×/120h/7.5h figures, and the cost-comparison section pads the body.	3 / 5
Actionability	Provides copy-paste-ready, executable code with concrete class names, parameters, and install commands for every pipeline stage and modality.	5 / 5
Workflow Clarity	Stages 1–4 are clearly sequenced, but this batch/destructive pipeline (filtering and dedup drop data) has no validation or verification steps, capping workflow clarity at 3 per the rubric.	3 / 5
Progressive Disclosure	Good structure with two real one-level-deep reference files clearly signaled at the end, though substantial detail (full multimodal curation code, benchmarks, cost tables) that could live in references is inlined.	4 / 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 comprehensive and well-structured, explicitly answering both what the skill does and when to use it with concrete actions and natural trigger phrases. Minor room for more synonyms or file extensions, but otherwise strong.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'fuzzy deduplication (16× faster)', 'quality filtering (30+ heuristics)', 'semantic deduplication', 'PII redaction', 'NSFW detection' — covering the domain comprehensively.	5 / 5
Completeness	Clearly states both what it does (curation features) and when to use it via the explicit 'Use for preparing high-quality training datasets, cleaning web data, or deduplicating large corpora' trigger clause.	5 / 5
Trigger Term Quality	Includes natural user phrases ('preparing high-quality training datasets', 'cleaning web data', 'deduplicating large corpora') and domain terms, but omits some synonyms and concrete file formats.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (GPU-accelerated LLM data curation) with distinct triggers that are unlikely to fire for unrelated skills.	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: OpenLAIR/dr-claw
Path: skills/data-processing/nemo-curator/SKILL.md
Commit: 4b35756

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.