Content
82%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 tightly written rule catalog with concrete, greppable detection patterns, explicit exemptions, and clear per-rule structure. Its only weaknesses are mild: some duplicated exemption text, implicit ERROR/WARNING severity semantics, and a length that could justify moving the rule catalog to a reference file.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The 10 rules are dense and high-signal, with nearly every sentence carrying checkable identifiers, paths, or APIs. Minor over-explanation (e.g., 'the array return type is a legacy Shadow DOM v0 artifact') and the protocol.ts / shadowRoots()[0] 'Do not flag' text duplicated verbatim across two rules keep it below 'every token earns its place'. | 4 / 5 |
Actionability | Each 'Look for' item is a copy-paste-ready, greppable pattern ('navigator.userAgent', 'visibility = [ "*" ]', 'try { await ...invoke_*() }', 'window.btoa', manual '/ 1000 arithmetic'), paired with explicit 'Do not flag' exemptions and a citation directive ('cite the rule id in findings'). For an instruction-only review skill this is fully actionable guidance covering the common cases. | 5 / 5 |
Workflow Clarity | The single action is unambiguous: match the diff against the rules for the path group and cite rule ids in findings, with a stated provenance bar ('recurred in at least 5 CLs from 2 distinct reviewers'). It falls short of 5 only because how to act on ERROR vs WARNING severity (block vs suggest) is left implicit. | 4 / 5 |
Progressive Disclosure | No bundle files exist, and the inline structure is well-organized: a scoping intro, one headed section per rule, and a consistent Severity/Why/Look for/Do-not-flag format that is easy to navigate. At 74 lines the rule catalog is slightly beyond ideal SKILL.md length and could be split into a reference file, which is the minor gap versus a 5. | 4 / 5 |
Total | 17 / 20 Passed |