Content
53%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 well-organized, jargon-consistent contract document that records concrete mechanisms (gate key, credential file, doctor probe) and a specific operational checklist. Its weaknesses are redundancy — the display-only contract and gate-off state are each restated multiple times — and the absence of validation checkpoints or navigable references, since the actionable detail is deferred to skills outside this bundle.
Suggestions
Merge "What this skill is not" into "What a suggestion is" — the not-a-verdict and not-an-availability-promise points largely restate the four prohibitions and the gate-off rule already given, cutting roughly a dozen lines of redundancy.
State the gate-off and catalogue-ownership facts once each instead of three times (opening section, rejected-mechanisms section, and closing section) to tighten conciseness.
Add a concrete validation step to the operational checklist (e.g., how a caller verifies the doctor readiness output before proceeding, or how to confirm a skill was filtered rather than merely suggested) so the sequence has an explicit checkpoint.
Reference the question-design skill and the product catalogue by path or linkable identifier rather than by name alone, so the external pointers are navigable instead of descriptive.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with no padding over concepts Claude already knows, but it repeats its core contract structurally: the not-selection/not-auto-load/not-judgment rules appear in "What a suggestion is" and again nearly verbatim in "What this skill is not" ("never close cards, approve merges, or mutate queues"), and both "the gate ships off" and "the catalogue owns the call path" are stated multiple times. It fits anchor 3 — mostly efficient but could be tightened by merging the redundant closing section. | 3 / 5 |
Actionability | There is genuinely concrete guidance (the gate key `workflow.jev.enabled`, the credentials file `~/.moai/.env.typesafe`, the TCP-only doctor probe, and a specific 6-item checklist covering logging, task-origin filtering, `disable-model-invocation` filtering, session caching, duplicate-injection guards, and name canonicalization). However, the bulk of the body is definitional contract and rejected-mechanism record, and the actionable question-design rules are explicitly deferred elsewhere ("Those rules live in the question-design reference skill"), leaving the guidance incomplete for anchor 4. | 3 / 5 |
Workflow Clarity | The gate-first flow has one explicit conditional checkpoint ("A caller that cannot confirm all three proceeds as if the gate were off") and the operational checklist is numbered, but the checklist items are conditional guidance with no validation or verification steps, and the sequence presumes a caller that the skill itself says does not exist. Steps are listed with checkpoints missing or implicit — anchor 3, and the under-50-line simple-skill exception does not apply to this ~92-line body. | 3 / 5 |
Progressive Disclosure | The body is well-organized into six clearly-headed sections with no inlined reference-dump content, and no bundle files exist to misplace. It misses anchor 5 because it is a monolithic single file over 50 lines whose only external pointers ("the question-design reference skill", "the product catalogue") are named without paths or links, giving no navigable reference structure. | 4 / 5 |
Total | 13 / 20 Passed |