Content
75%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.
A well-structured, token-efficient workflow skill with clear sequencing, explicit guard procedures, and a properly split reference bundle. The main issues are small: duplicated reference sections, an orphaned bundle file, pseudocode usage examples, and a thin completion checklist.
Suggestions
Merge the "Refs" and "References" sections into one — they list the same three files twice.
Link references/troubleshoot-to-frame-problem-llm.md from the body (e.g. from the Cynefin routing note in the Usage Examples) so the handoff template is discoverable.
Replace the comment-only bash Usage Examples with the actual prompts/outputs as text, or drop them; and name the concrete learnings-file path and thinking-store environment variable instead of "a known location".
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is written in terse, imperative bullets with no explanation of concepts Claude already knows (e.g. "Wolf Fence (binary search)", "OODA" are named, not taught). It misses the 5 anchor because the "Refs" and "References" sections duplicate the same three files almost verbatim, and the Usage Examples blocks add bulk with limited instructional value. | 4 / 5 |
Actionability | Mostly concrete, executable guidance for an instruction skill: exact qualification questions ("Stack? Environment? Changed recently?"), an explicit search query pattern "[error] [stack] [framework]", enumerated required artifact sections, and a precise AskUserQuestion fallback procedure. Below the 5 anchor because the Usage Examples are comment-only pseudocode blocks, and the learnings file and thinking-store locations are vague ("a known location", "an environment variable"). | 4 / 5 |
Workflow Clarity | The workflow is a clearly numbered 0-6 sequence with explicit skip conditions ("Solution found → skip to 5.Learn"), entry conditions ("Only if diagnosis inconclusive"), an exit condition ("Exit when root cause confirmed and fix verified"), and an OODA feedback loop. Below the 5 anchor because the completion checklist contains only a single item and verification of the fix is named but not operationalized as a checkpoint step. | 4 / 5 |
Progressive Disclosure | The body is an appropriate overview pointing to three real, one-level-deep bundle files (references/protocols/diagnose.md, references/protocols/search-multi-source.md, references/reference.md), each of which exists and is well-scoped. Below the 5 anchor because the reference list is duplicated across two sections ("Refs" and "References"), and the bundle file references/troubleshoot-to-frame-problem-llm.md is never linked from the body despite the Usage Examples alluding to routing to another skill. | 4 / 5 |
Total | 16 / 20 Passed |