Content
88%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.
An excellent patterns skill: highly actionable code, a genuinely diagnostic debug workflow with validation checkpoints, and tight non-obvious content throughout. The only real gap is structural — everything lives in one 340-line file with no reference bundle, and a couple of sections (the DataFusion struct gotcha) sit at the edge of the skill's stated scope.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with non-obvious, experience-derived information ("Every materialization of a dataset.reader(...) DataFrame is one cloud round-trip") and assumes Claude's competence, delegating DataFrame basics to the datafusion-python skill. It is not a 5 because a few sections could be trimmed or moved — the struct-null gotcha section self-describes as "A DataFusion gotcha" that partially belongs in the companion skill, and the debug recipe recapitulates patterns already covered — but it is well above the 'mostly efficient with some unnecessary explanation' anchor. | 4 / 5 |
Actionability | Fully executable, copy-paste-ready code covers the common cases: connecting via CatalogClient, the cache() pattern, cross-segment grouping by rerun_segment_id, per-entity is_not_null() filters, dataset.info() and schema() probing, and the null-guard projection, plus a concrete alternatives table for count(). This matches the top anchor; there are only minor gaps, not the missing key details of the 4 anchor. | 5 / 5 |
Workflow Clarity | The 'Common debug recipe' is a clearly sequenced 6-step diagnostic workflow with explicit checkpoints (wrap materializations in time.perf_counter(), split build vs materialize timing, measure tbl.nbytes) and interpretation feedback at each step ("If 'build' takes seconds, something inside is materializing eagerly"). The skill is read-only querying, so the destructive/batch validation cap does not apply. | 5 / 5 |
Progressive Disclosure | Good structure: well-titled self-contained sections, deliberate one-level delegation to the datafusion-python skill ('The DataFusion side of the stack is covered by the datafusion-python skill'), and no nested references. No bundle files exist, so all ~340 lines of pattern detail live in SKILL.md; a reference file for the longer pattern sections would be a minor improvement toward the 5 anchor, but this sits above the 'content that should be separate is inline with weak signaling' anchor. | 4 / 5 |
Total | 18 / 20 Passed |