Content
81%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 highly actionable, well-sequenced workflow skill: every step names concrete files, thresholds, formulas, and output formats, with explicit handling of missing-data edge cases. The main costs are verbose rationale embedded in the data-loading steps and inlined report templates that slightly bloat the single file.
Suggestions
Trim the justificatory parentheticals in Steps 2 and 3 (e.g. why the rank_score floor exists, why a blank fit_rating must not read as 0) down to the operative rule — skip/count/report — since the reasoning behind each rule is not needed to execute it.
Move the full report template and example entry formats (Steps 6 and 8) into a reference file under references/ and link to it, keeping SKILL.md as a leaner overview of the 8-step workflow.
Consolidate the repeated skip-count-report handling for missing gaps and blank fit_ratings into a single stated pattern in Step 2 and reference it from Step 3 instead of restating the treatment and its rationale twice.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient, actionable instruction, but Steps 2-3 carry dense nested parenthetical rationale — e.g. "below that, a job is Weak/Poor Fit and would otherwise dominate the heatmap with jobs the user shouldn't chase" and "that reads as weight 1.0, the maximum, and lets the one job the framework knows nothing about dominate the heatmap" — that explains reasoning Claude does not need. It fits the 'mostly efficient but could be tightened' anchor better than the 'several padded sections' anchor because the bulk is genuine procedural guidance. | 3 / 5 |
Actionability | Fully executable guidance throughout: exact file paths and column schemas, concrete thresholds ("rank_score >= 45"), a precise weighting formula ("(100 - fit_rating) / 100"), verbatim terminal messages, example WebSearch queries, copy-paste report templates, and filename slug rules. Specific examples cover the common aggregate and targeted cases. | 5 / 5 |
Workflow Clarity | Eight clearly sequenced steps with mode detection up front, an explicit intermediate checkpoint ("Print this table to the terminal... before continuing"), and error-recovery loops for the batch aggregate path: dead source URLs are skipped, missing gaps fields are "skipped, counted, and reported", and blank fit_ratings get the same treatment — plus a 9-rule closing checklist. The batch operation includes validation and reporting loops, so the missing-validation cap does not apply. | 5 / 5 |
Progressive Disclosure | Well-sectioned single file with clear step headers, and external references (".claude/skills/job-application-assistant/09-web-research.md", "01-candidate-profile.md", "04-job-evaluation.md") are one level deep and clearly signaled. At ~250 lines with full report templates inlined, some content (report format examples) could live in a reference file, which keeps it below the lean-overview 5 anchor. | 4 / 5 |
Total | 17 / 20 Passed |