Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is concise and the simple workflow is clear with a sensible load-once guard. Its main weaknesses are actionability — the real guidance lives in referenced files that aren't bundled — and progressive disclosure, since the referenced standards files do not exist within the skill.
Suggestions
Bundle the per-language CLAUDE.md files under references/ (or ship inline convention summaries) so the referenced paths resolve and the skill is self-contained.
Add a short inline example of an applied convention (e.g. a Go naming/import rule) so the body is actionable even before the external file is loaded.
Verify the '../../code-standards/<lang>/CLAUDE.md' paths are correct relative to the skill's base directory, or convert them to bundle-relative paths.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean — a detection table plus brief load/apply instructions — with no padding and no explanation of concepts Claude already knows, so every token earns its place. | 3 / 3 |
Actionability | It gives a concrete file-pattern→standards-file table and names the Read tool, but the actual conventions are deferred to external CLAUDE.md files that are not present in the bundle, leaving the actionable substance incomplete. | 2 / 3 |
Workflow Clarity | As a simple single-purpose skill, its detect→load-once→apply sequence is unambiguous and includes an explicit dedup guard ('if you have already loaded Go standards, do not re-read'), satisfying the simple-skill allowance for a top score. | 3 / 3 |
Progressive Disclosure | References are one level deep and clearly signaled via the table, but they point to files outside the bundle (../../code-standards/<lang>/CLAUDE.md) that do not exist in the workspace, so the disclosure structure does not actually resolve. | 2 / 3 |
Total | 10 / 12 Passed |