Content
75%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.
A well-organized, high-signal reference body: concrete code patterns, honest labeling of template examples, thorough validation and scientific-pitfall notes, and correctly deferred bulk API detail in a one-level-deep reference bundle. The main improvement levers are tightening the redundant bullet lists and moving some inlined API detail into the existing reference file.
Suggestions
Trim or collapse the 'When to Use This Skill' bullet list and per-section 'Key capabilities' bullets, which largely restate the description and the code that immediately follows.
Move version-migration detail (renames, removed APIs like the SSW wrapper) into a consolidated 'Deprecated/renamed APIs' section in references/api_reference.md, keeping only active patterns in SKILL.md.
Give the placeholder-based snippets (nj(distance_matrix), permanova(dm, grouping)) minimal self-contained constructions so the most common workflows are fully copy-paste executable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is code-heavy and dense with non-obvious version-specific facts ('renamed from otu_ids in 0.6.0', 'replaces the removed SSW wrapper', 'default 10 fails when n < 10 in 0.7.4'), with deprecation guidance properly quarantined in 'Important notes' rather than free-roaming. Minor trimming is possible — the 10-bullet 'When to Use' list restates the description and per-section 'Key capabilities' bullets partially duplicate the code — which fits the score-4 anchor (minor instances of over-explanation) rather than the lean score-5 anchor. | 4 / 5 |
Actionability | Most snippets are executable with real signatures, literal sequences, seeds, and named parameters (e.g., the pair_align and ProteinEmbedding examples), but several use undefined placeholders such as 'nj(distance_matrix)', 'permanova(distance_matrix, grouping)', and 'counts_matrix' without construction. The skill transparently flags these as 'workflow templates'; that honest framing plus mostly concrete code fits the score-4 anchor, short of the uniformly copy-paste-ready score-5 anchor. | 4 / 5 |
Workflow Clarity | Four explicit step chains in 'Common Workflows' (e.g., 'Read BIOM table → Calculate alpha/beta diversity → Ordination (PCoA) → Statistical testing (PERMANOVA)') plus pervasive validation guidance ('Reject empty samples explicitly', 'Align sample IDs first', 'Record seed and versions', 'Inspect warnings and retained variance'). Checkpoints are distributed in notes rather than embedded as gated validate-then-proceed steps in the workflows, so it matches the score-4 anchor (clear sequence, most checkpoints present) rather than the score-5 anchor with explicit feedback loops. | 4 / 5 |
Progressive Disclosure | Both bundle references are real, clearly signaled, and one level deep: a dedicated 'Reference Documentation' section enumerates what references/api_reference.md (778 lines) covers, and the Overview links references/review.md for evidence and boundaries. The 470-line body is well-sectioned but still carries per-section API detail that could migrate to the reference file, placing it at the score-4 anchor ('good structure; most content appropriately placed; minor organization gaps') rather than the clean-overview score-5 anchor. | 4 / 5 |
Total | 16 / 20 Passed |