Content
90%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 instruction-only skill body: concrete tools, a complete publish example, explicit negative guidance, and crisp routing against sibling skills. The only gaps are a missing post-publish verification step and the absence of any external references despite a body length that is starting to justify them.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and efficient across ~90 lines: trigger examples, a 5-step workflow, one compact JSON example, one shell command, and a Don't list — every token earns its place with zero explanation of concepts Claude already knows. | 5 / 5 |
Actionability | Fully executable guidance: named tools to call (get-plan-blocks, list-plan-components), a per-question-type block-selection list, a complete copy-paste-ready visual-answer-source.json example, and the full publish command with flags (--source-url, --prev-plan-id, --visibility). | 5 / 5 |
Workflow Clarity | The 5-step sequence is clear and guardrailed ('Do not invent endpoints, schema fields, UI states'), but there is no explicit validation checkpoint (e.g., confirm blocks rendered or check visual-answer-url.txt after publish). Publishing is not destructive or batch, so the anchor-3 cap does not apply; it still lacks the validate/verify feedback loop of anchor 5. | 4 / 5 |
Progressive Disclosure | Well-organized sections with no nested references and all content appropriately inline for a skill of this size. The body exceeds the ~50-line simple-skill threshold, so the automatic 5 does not apply, and growing detail (block-type matrix, terminal handoff) could eventually warrant a reference file. | 4 / 5 |
Total | 18 / 20 Passed |