CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

dask

Scales pandas, NumPy, and custom Python research workflows beyond memory or across clusters with Dask. Covers DataFrames, Arrays, Bags, Futures, chunking, schedulers, and distributed diagnostics. Use for partitioned file processing, scientific array computation, or parallel tasks whose memory and dependency structure require Dask.

Quality

83%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

65%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 well-structured overview of a five-component library: concrete code, clear component selection guidance, and exemplary progressive disclosure to seven real one-level-deep reference files. Weaker spots are version/date detail living in the main body instead of the review reference, some duplication between the global and per-component When-to-Use lists, and missing explicit validation checkpoints for batch write workflows.

Suggestions

Move time-sensitive version/date statements ('Reviewed release: 2026.8.0 (September 30, 2026 review)', 'Since 2025.1.0...') into references/review.md and keep only the pinned install command in the body, reserving a short 'old patterns' note for the removed query-planning toggle.

Add explicit validation checkpoints to the batch write workflows, e.g. after summary.to_parquet or da.to_zarr, verify the output (read back a sample partition or check row/partition counts) before proceeding.

Collapse the 'When to Use This Skill' section into a pointer to the per-component When-to-Use lists (or vice versa) to remove the duplicated trigger lists and tighten the token budget.

DimensionReasoningScore

Conciseness

Mostly efficient with quick examples and wrong/correct pairs, but it carries time-sensitive details in the main body ('Reviewed release: dask/distributed 2026.8.0 (September 30, 2026 review)', 'Since 2025.1.0...') rather than confining them to a deprecated/old-patterns or review section, and the 'When to Use This Skill' section largely duplicates the per-component When-to-Use lists. Not 2 because there is no conceptual padding of things Claude already knows — the excess is localized and trimmable.

3 / 5

Actionability

Concrete guidance throughout: executable install commands, runnable quick examples for every component, wrong/correct code pairs, and named troubleshooting steps. Not 5 because several examples rely on placeholder names ('python_function', 'computation', 'large_dataset', 'problematic_computation') that prevent copy-paste readiness, though the body disclaims these as illustrative.

4 / 5

Workflow Clarity

The iterative development workflow (synchronous → threads on a sample → distributed) provides a clear sequence, but its checkpoints are implicit ('Test logic, then scale to full dataset') and the batch write operations shown (to_parquet, to_zarr with mode='w-') include no output validation steps. The rubric's cap for batch operations without validation applies, so it cannot exceed 3; not 2 because sequences and error-recovery guidance (Common Issues) are present.

3 / 5

Progressive Disclosure

The body is an overview with quick examples per component, and every deep-dive is well signaled ('refer to references/X.md which includes:' followed by a contents list), plus a consolidated Reference Files index. All seven referenced files exist and are one level deep with no chained references. This matches the 5 anchor; not 4 because there are no organization gaps — navigation is easy and the split is appropriate.

5 / 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.

A strong, third-person description that states concrete capabilities, names the library's five components, and gives an explicit 'Use for...' trigger clause that gates on Dask actually being required. The only gap is modest synonym/extension coverage in trigger terms. Suggestions are not warranted.

DimensionReasoningScore

Specificity

The description lists multiple concrete actions — 'Scales pandas, NumPy, and custom Python research workflows beyond memory or across clusters', 'partitioned file processing, scientific array computation' — and enumerates coverage of all five components ('DataFrames, Arrays, Bags, Futures, chunking, schedulers, and distributed diagnostics'). This matches the comprehensive-coverage anchor; it is not 4 because there are no meaningful gaps in the action inventory.

5 / 5

Completeness

Both questions are answered explicitly: what ('Scales pandas, NumPy, and custom Python research workflows... Covers DataFrames, Arrays, Bags, Futures...') and when ('Use for partitioned file processing, scientific array computation, or parallel tasks whose memory and dependency structure require Dask'). This mirrors the 5 anchor verbatim in structure; not 4 because the when-clause is concrete, not merely present.

5 / 5

Trigger Term Quality

Good natural keyword coverage: 'pandas', 'NumPy', 'clusters', 'parallel tasks', 'partitioned file processing'. Falls short of the 5 anchor because common variations users would say are missing — e.g. 'out-of-core', 'parallelize', and file extensions like '.parquet' or '.csv'.

4 / 5

Distinctiveness Conflict Risk

The description names a specific library (Dask) and gates usage explicitly ('parallel tasks whose memory and dependency structure require Dask'), giving it a clear niche with minimal conflict risk against adjacent pandas/NumPy/parallelism skills. It is not 4 because the explicit Dask gating removes the overlap risk that a 4 anchor would tolerate.

5 / 5

Total

19

/

20

Passed

Validation

100%

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

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository

K-Dense-AI/scientific-agent-skills

Path

skills/dask/SKILL.md

Reviewed

Table of Contents

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.