Content
92%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, high-density audit skill: every section carries specific, non-obvious domain guidance with executable commands and grep patterns, a clearly sequenced verify-then-record workflow with explicit validation gates, and a clean one-level-deep reference structure backed by real bundle files. The only room for improvement is trimming minor rhetorical and redundant passages to tighten token efficiency.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with non-obvious domain facts (exact OWASP parameter floors, per-language library footguns like 'Cipher.getInstance("AES")' defaulting to ECB) and avoids explaining concepts Claude already knows, but contains minor trimmable material: the second intro paragraph partly overlaps the first, the rhetorical 'There are < 50 people on Earth' line, and a long output template. Not 5 because of these minor padding instances; not 3 because nearly every line carries a specific, actionable fact. | 4 / 5 |
Actionability | Guidance is copy-paste ready throughout: per-section grep patterns ('AES.*ECB', 'pbkdf2_hmac (check iterations arg)'), executable commands ('testssl.sh --jsonfile', 'aws kms get-key-policy --key-id <id> --policy-name default', 'node references/validate-findings.mjs findings.json'), concrete vulnerable-code examples, and specific parameter floors (PBKDF2 ≥ 600,000 iterations, scrypt N=2^17, bcrypt cost ≥ 12, 128-bit token entropy). | 5 / 5 |
Workflow Clarity | The multi-step process is explicitly sequenced with validation checkpoints: audit checklist → refutation pass ('Before anything is confirmed, run the refutation pass') → verdict classification (confirmed/needs_validation/rejected/hardening) → write and validate findings.json → render report, with a runtime fix-verification section covering round-trip testing and migration. Feedback loops are explicit and the destructive/batch-sensitive audit output is schema-validated. | 5 / 5 |
Progressive Disclosure | SKILL.md serves as the operational overview while detail lives one level deep in references/ — findings format, verification refutation pass, severity calibration, validator, SARIF exporter, and schema — each well-signaled with purpose ('run the refutation pass in [references/verification.md]', 'Rate severity with [references/severity-calibration.md]') and all six referenced files verified to exist as real, substantive content in the bundle. | 5 / 5 |
Total | 19 / 20 Passed |