datachain-core

Use ONLY for abstract DataChain SDK questions — API usage, method signatures, or code patterns — when no specific dataset or bucket is referenced. If the request mentions creating, saving, listing, exploring datasets or buckets, use datachain-knowledge instead.

Quality

83%

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

78%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-organized SDK reference with strong executable guidance and clear pre-flight validation. Its main weakness is progressive disclosure: the large API and func reference sections are inlined rather than split into referenced files.

Suggestions

Move the Section 5 (Core API Reference), Section 6 (Type System), and Section 7 (func Module) reference material into a bundled REFERENCE.md and replace the inline blocks with one-level-deep pointers (e.g., 'See references/REFERENCE.md').

Add an explicit post-run verification step to the pipeline templates (e.g., after .save(), assert row count or print chain.count()) to close the validate→fix→retry loop for batch UDF pipelines.

Tighten the discursive rationale in Section 1 (e.g., the problem-specific vs data-quality filter definitions) into terser rule statements so every token earns its place.

Dimension	Reasoning	Score
Conciseness	Dense and reference-style, assuming Claude's competence and conveying domain-specific doctrine (anon probe semantics, UDF output-type resolution) Claude would not already know; a few discursive passages in Section 1 could be trimmed without loss.	4 / 5
Actionability	Pervasive executable code with ✓/✗ anti-pattern markers, concrete API signatures, and copy-paste-ready pipeline templates covering the common cases.	5 / 5
Workflow Clarity	Clear sequencing via the Pre-Generation Checklist, a numbered dataset-reuse workflow, and multi-stage templates that chain via read_dataset; minor gaps in post-execution verification feedback loops for batch pipelines.	4 / 5
Progressive Disclosure	Well-sectioned but monolithic: ~150 lines of API/func reference (Sections 5–7) and templates (Section 8) are inlined in a 770-line SKILL.md with no one-level-deep references to bundle files, matching the anchor where reference content that should be separate stays inline.	3 / 5
	Total	16 / 20 Passed

Description

87%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, well-scoped description that clearly states what it does, when to use it, and when to defer to a sibling skill. Specificity and trigger-term coverage are good rather than exhaustive.

Dimension	Reasoning	Score
Specificity	Names the domain and concrete question categories — 'API usage, method signatures, or code patterns' — which are specific actions, but coverage of the SDK's full surface is implied rather than enumerated, leaving minor gaps.	4 / 5
Completeness	Explicitly answers both 'what' (abstract DataChain SDK questions) and 'when' ('Use ONLY ... when no specific dataset or bucket is referenced'), with concrete trigger phrases and an explicit negative-routing clause.	5 / 5
Trigger Term Quality	Includes natural terms a user would say ('DataChain SDK questions', 'API usage', 'method signatures', 'code patterns', 'creating, saving, listing, exploring datasets or buckets') with good coverage; file extensions do not apply to an SDK skill, so it stops short of comprehensive.	4 / 5
Distinctiveness Conflict Risk	Carves out a clear niche (abstract questions, no specific dataset/bucket) and explicitly routes the overlapping case to datachain-knowledge, minimizing conflict risk.	5 / 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
skill_md_line_count	SKILL.md is long (776 lines); consider splitting into references/ and linking	Warning

	Total	15 / 16 Passed

Repository: datachain-ai/datachain
Path: src/datachain/skill/core/SKILL.md
Commit: bc3e2c8

Reviewed: ~19 hours 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.