Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a well-structured, actionable skill with clear multi-step workflows and concrete bash commands for each audit phase. Its main weakness is moderate verbosity from repeated explanations (especially about AGENTS.md equivalence) and some content that could be offloaded to reference files. The progressive disclosure structure is well-signaled but cannot be fully verified without bundle files.
Suggestions
Remove the repeated explanation that AGENTS.md is identical to CLAUDE.md — state it once in the Overview and trust Claude to apply it throughout.
Consider moving the detailed audit steps (Steps 3-5 especially) into a reference file to keep SKILL.md as a concise overview with the discovery workflow and high-level audit checklist.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Generally efficient but has some redundancy — the explanation that AGENTS.md is identical to CLAUDE.md is repeated three times (overview, discovery step 2 note, and implicitly throughout). Some sections like the freshness scoring table and structural verification could be tighter. However, it mostly avoids explaining things Claude already knows. | 2 / 3 |
Actionability | Provides concrete, executable bash commands for every discovery and audit step — glob testing, frontmatter extraction, file reference verification, and dead glob detection. The commands are copy-paste ready and specific to the task. | 3 / 3 |
Workflow Clarity | The workflow is clearly sequenced in two phases (Discovery 1-4, then Audit Steps 1-7) with explicit validation at each step. Each step has clear pass/fail criteria (e.g., dead globs, broken refs, <50% convention adherence) and the freshness scoring table provides a structured evaluation framework. The report generation step serves as a final validation checkpoint. | 3 / 3 |
Progressive Disclosure | The Reference Navigation table clearly signals four reference files with one-level-deep links, which is good structure. However, since no bundle files were provided, we cannot verify these references exist. Additionally, the main SKILL.md itself is fairly long (~120 lines of content) and some of the detailed audit steps (like convention drift detection) could potentially be offloaded to reference files to keep the overview leaner. | 2 / 3 |
Total | 10 / 12 Passed |