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.
A well-built router skill: highly actionable symptom-to-reference routing, clear decision logic, and strong progressive disclosure with per-topic summaries that tell the model what each reference contains without inlining it. The main cost is redundancy — the decision tree appears twice and Example Invocations restates the Quick Reference — and a handful of unverifiable or path-less external references.
Suggestions
Drop either the dot-graph decision tree or the numbered list — they state the identical mapping, and the numbered list alone saves ~20 lines with no loss.
Trim or cut the Example Invocations section, which restates the Quick Reference mapping a third time; a few representative examples would preserve the natural-phrasing signal at a fraction of the tokens.
Give external skill references a consistent, actionable form (e.g. the '/skill <name>' syntax used for axiom-shipping) instead of bare 'See axiom-build' / 'See axiom-payments suite'.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense routing information with no explanations of concepts Claude already knows, but the same symptom-to-file mapping is rendered four times: the dot graph exactly duplicates the numbered decision tree, and the ~40-line Example Invocations section largely duplicates the Quick Reference table. This is 'mostly efficient but could be tightened' (anchor 3) rather than clearly padded enough for anchor 2. | 3 / 5 |
Actionability | Concrete, executable routing throughout — an exact reference file per symptom, real CLI commands ("security find-identity", "codesign -d --entitlements"), and a scanner invocation. Minor gaps: some destinations are vague ("See axiom-payments suite", "See axiom-build" with no path or skill invocation syntax), keeping it below anchor 5. | 4 / 5 |
Workflow Clarity | The task-to-reference selection sequence is clear and unambiguous via the numbered decision tree and the conflict-resolution if/then rules with concrete error-code signals (CODESIGN, ITMS-90xxx, errSec). No validation checkpoints are needed since the skill routes rather than performs destructive operations, so anchor 4 fits; anchor 5's feedback loops are not applicable to a router. | 4 / 5 |
Progressive Disclosure | Good structure with well-signaled, one-level-deep references organized by topic and a quick-reference table up front. Minor gaps: no bundle files are present to verify the skills/*.md paths resolve, several references point to external skills without paths (axiom-build, axiom-shipping, axiom-payments), and the inline Codex-porting comment block adds noise — anchor 4 rather than 5. | 4 / 5 |
Total | 15 / 20 Passed |