Content
78%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 router skill: comprehensive, concrete routing with clean one-level-deep reference splitting and explicit disambiguation logic. Its main cost is redundancy — the same routing table is encoded four ways (lookup table, trigger lists, decision tree plus simplified list, example invocations), inflating the token budget without adding new information.
Suggestions
Collapse the DOT decision tree and the 'Simplified:' numbered list into one format — they encode identical routing, and the DOT version already lacks the production-triage and app-clips nodes the list has.
Trim Example Invocations from ~40 entries to the 10–15 most distinct queries; most mappings already appear verbatim in the Quick Reference table and Routing Logic trigger lists.
Merge each Quick Reference table row with its corresponding Routing Logic section's 'Reference' line so every symptom→skill mapping is stated exactly once.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Phrases like the Quick Reference table, the Routing Logic trigger lists, the 'Simplified:' decision list, and ~40 'User: ... → See ...' Example Invocations all restate the same routing mappings three to four times over. No tokens are spent explaining concepts Claude already knows — every line carries task-specific routing data — but the structural redundancy means it 'could be tightened', matching the mostly-efficient anchor rather than the noticeably-verbose one. | 3 / 5 |
Actionability | Routing is fully executable for an instruction skill: exact pointers ('See `skills/app-store-diag.md`'), agent launches ('Launch `triage-analyzer` agent (or `/axiom:triage sentry`)'), and concrete specifics like 'ITMS-90035 Invalid Signature', 'Guideline 2.1', and '10/15/100 MB' size tiers. Per the rubric's instruction-only note, absence of code is not penalized; the specific examples cover the common cases. | 5 / 5 |
Workflow Clarity | The decision tree sequences routing with explicit fallbacks ('Signing error?' -> 'skills/app-store-submission.md' [label="no, general prep"]) and a simplified ordered list. Minor gaps: the DOT tree omits the production-triage and app-clips nodes that the simplified list includes, and section numbering is inconsistent ('10a' followed by '10'). | 4 / 5 |
Progressive Disclosure | The body is a pure overview router: every topic is split out to a one-level-deep per-topic file ('skills/app-store-ref.md', 'skills/testflight-triage.md', 'axiom-security (skills/code-signing-diag.md)'), clearly signaled via the Quick Reference table, per-section '**Reference**:' lines, and the routing list — matching the clear-overview anchor. No bundle directories (references/, scripts/, assets/) ship with this evaluation copy, so the referenced plugin-relative paths cannot be verified on disk, but the written structure has no nested-reference chains. | 5 / 5 |
Total | 17 / 20 Passed |