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 content is a well-structured instruction-only skill: clear multi-mode workflow, concrete report/fix procedures, and sound progressive disclosure into a real references bundle. Weaknesses are moderate verbosity and redundancy (overlapping workflow sections, repeated report-delivery instructions, a stray 'codex' reference and a 'describe anyways' typo), plus the absence of a concrete severity taxonomy or report example.
Suggestions
Merge the 'Workflow' section and 'Workflow Decision Tree' into one pass, and deduplicate the report-delivery instructions ('After you write the report file, summarize... Also tell the user where...').
Add a concrete severity taxonomy (e.g. Critical/High/Medium/Low with one-line definitions) and a short example report skeleton so findings and sections are formatted consistently.
Enumerate or link the available reference files (or at least the supported language list with its per-language general files), fix the 'describe anyways' typo, and remove the stale 'codex' reference in the TLS section.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient and does not explain concepts Claude already knows, but it has noticeable padding: the 'Workflow' section and 'Workflow Decision Tree' partially repeat each other, report instructions restate delivery steps twice ('After you write the report file, summarize the findings to the user. Also tell the user where the final report was written to'), and phrases like 'This skill provides a description of how to identify...' and 'From there it can operate in a few ways' could be trimmed. This matches anchor 3 ('mostly efficient but includes some unnecessary explanation or could be tightened') better than anchor 4's 'minor instances'. | 3 / 5 |
Actionability | As an instruction-only skill it gives concrete, executable guidance: the reference filename format '<language>-<framework>-<stack>-security.md' (which matches the actual bundle files), a specific report filename 'security_best_practices_report.md', explicit report structure requirements (executive summary, severity sections, numeric finding IDs, line numbers), and per-finding fix instructions. This fits anchor 4 ('mostly executable guidance; concrete... with minor gaps'); it is not a 5 because no example report skeleton or concrete severity taxonomy (e.g. Critical/High/Medium/Low) is provided. | 4 / 5 |
Workflow Clarity | The sequence is clear (identify languages/frameworks -> load matching references -> operate in one of three modes -> report -> fixes) with a decision tree for ambiguous cases, and user-confirmation checkpoints are present (let the user read the report before fixes, notify before fixing passive findings, follow the user's testing/commit flows). Minor validation gaps keep it at anchor 4 rather than 5: testing flows are conditional ('if any') and there are no explicit verify/re-check loops after applying fixes. | 4 / 5 |
Progressive Disclosure | The SKILL.md body is an overview that appropriately pushes the bulk of detail into ten real, one-level-deep reference files, with clear instructions to check the references directory and a naming convention that matches the actual bundle filenames. It falls short of anchor 5 because the references are never enumerated or linked (discovery relies solely on the format string), heading levels are inconsistent (h1 for '# Overrides'/'# Report Format'/'# Fixes' vs h2 elsewhere), and the two inline 'General Security Advice' items are not clearly signaled as cross-cutting rather than language-specific. | 4 / 5 |
Total | 15 / 20 Passed |