Content
92%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.
An exemplary extension skill: it layers two precisely-specified overrides onto a delegated parent procedure with concrete labels, a decision table, and an ambiguity checkpoint. The only flaw is a minor duplicated framing between the intro and Procedure sections.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and delegates the base flow to the parent skill, but the opening paragraph and the Procedure line both state that the skill extends create-issue with two overrides — a minor duplication that could be merged into one framing. | 4 / 5 |
Actionability | Fully concrete for an instruction-only skill: exact label names, explicit override anchor points tied to specific parent-skill steps, a decision table for the rt/* classification, explicit inference rules, and a precise rule for the ambiguous case (ask the user, never guess or apply more than one). | 5 / 5 |
Workflow Clarity | Sequence is unambiguous via explicit anchoring to create-issue's steps (Override 1 replaces Step 3, Override 2 runs alongside Step 2), and the ask-the-user checkpoint on ambiguous classification serves as a validation gate. Issue creation is not a destructive or batch operation, so no cap applies. | 5 / 5 |
Progressive Disclosure | Under 50 lines with no external references needed; sections are well-organized with the label taxonomy appropriately kept inline and the parent-skill details properly delegated to the referenced create-issue skill. | 5 / 5 |
Total | 19 / 20 Passed |