Content
17%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is a templated stub: generic trigger sections, platitude-level instructions, and a persona/purpose paragraph that duplicate the description. It contains no concrete, executable data science guidance, and all eleven referenced files (the playbook and ten sub-skills) are missing from the bundle, so the promised depth is inaccessible.
Suggestions
Replace the generic Instructions with concrete, domain-specific guidance — e.g., specific EDA steps, model selection/validation commands, or a worked code example for a common task like regression or classification.
Add a real workflow sequence with validation checkpoints (e.g., check data quality, validate model on held-out set, only then summarize insights) instead of the current clarify/apply/provide template.
Create the referenced sub-skills/ and resources/ files (or remove the links) — all eleven referenced paths currently do not exist, making the skill's detailed content unreachable.
Trim the circular 'Use this skill when'/'Do not use this skill when' bullets and the Purpose paragraph, which restate the frontmatter description without adding information.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Noticeably padded for its length: circular trigger bullets ('Working on data scientist tasks or workflows', 'The task is unrelated to data scientist'), a Purpose paragraph that restates the description with more buzzwords ('Masters the complete data science workflow'), and platitudes like 'Apply relevant best practices and validate outcomes' that add nothing Claude doesn't already know. Not a 1 because the body is short overall and doesn't explain textbook concepts. | 2 / 5 |
Actionability | The Instructions are entirely abstract — 'Clarify goals, constraints, and required inputs', 'Provide actionable steps and verification' — with no code, commands, or domain-specific steps anywhere, matching the anchor for entirely vague direction. The only concrete pointer ('open resources/implementation-playbook.md') targets a file that does not exist. | 1 / 5 |
Workflow Clarity | A rough generic template exists (clarify goals, apply best practices, provide steps) but the steps are poorly defined and not a real data science sequence, and validation appears only as a bare mention with no checkpoint. Fits anchor 2 rather than 1 because some sequence is present. | 2 / 5 |
Progressive Disclosure | References are one level deep and clearly signaled, but every referenced path is dangling — resources/implementation-playbook.md and all ten ./sub-skills/*.md links point to files that do not exist in the bundle, so progressive disclosure fails in practice. The bare numbered H3 link list also gives no indication of when to open which module. Not a 1 because the body itself is well-sectioned and not a monolith or nested chain. | 2 / 5 |
Total | 7 / 20 Passed |