Content
67%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 well-structured, largely actionable guide with a clear gated workflow that properly confirms before destructive file operations. Its main weaknesses are length/redundancy (the 'When to Use' section duplicates the description's job) and template placeholders that prevent the commands from being fully copy-paste ready.
Suggestions
Tighten or remove the 'When to Use This Skill' bullet list — trigger guidance belongs in the description, not duplicated in the body — to recover token budget.
Replace placeholder-only template lines (e.g. `[custom command for their setup]`, `[X]`, `[Y]`) with at least one concrete filled-in example so the summary block is executable as written.
Fix the duplicate-detection one-liner: `find ... -exec md5 {} \; | sort | uniq -d` compares whole 'hash filename' lines, so duplicates are rarely flagged — show a correct form (e.g. hash-only via `md5 -q` / `md5sum` then group) to avoid giving Claude a broken command.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly practical (commands, templates, lists) but runs ~250 lines with redundant framing — the 'When to Use This Skill' bullet list re-states triggers the description should carry, and several markdown template blocks could be trimmed — fitting 'mostly efficient but includes some unnecessary explanation or could be tightened' rather than the lean anchor above. | 3 / 5 |
Actionability | It supplies concrete, mostly executable bash commands (ls, find, du, md5, mkdir, mv) and ready-to-use plan/summary templates, with only minor gaps — placeholder tokens like [target_directory], [custom command], and [X]/[Y] keep it just short of fully copy-paste ready. | 4 / 5 |
Workflow Clarity | A clear seven-step sequence (scope → analyze → patterns → duplicates → propose plan → execute → summary) carries explicit validation checkpoints — 'Always ask for confirmation before deleting', 'Ready to proceed? (yes/no/modify)', 'Stop and ask if you encounter unexpected situations' — satisfying the destructive-operation feedback requirement, though undo/logging is described rather than scripted. | 4 / 5 |
Progressive Disclosure | No bundle files exist (references/, scripts/, assets/ are absent), and the single SKILL.md is well-organized into clearly headed sections (When to Use, What This Skill Does, Instructions, Best Practices); at ~250 lines some templates/commands could be split into reference files, which keeps it just below the ideal one-level-deep structure. | 4 / 5 |
Total | 15 / 20 Passed |