Content
63%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 well-structured and lean with strong progressive disclosure and a clear validated workflow, but it is notably weak on actionability for an automation skill: it directs Claude to run AppleScript without supplying any concrete commands or pointing to a present script bundle.
Suggestions
Add at least one concrete, executable AppleScript snippet (or a reference to a real helper script in the bundle) showing how to inspect or change Atlas tabs/bookmarks, since the skill repeatedly instructs running AppleScript but provides none.
Make the Validation step concrete by naming the actual command or check to run (e.g., a specific osascript invocation or audit script) instead of 'the smallest command or test that exercises the changed behavior'.
Either move the AppleScript helper scripts into the references/ bundle and link them from the Progressive Disclosure section, or explicitly state where the executable helpers live so the guidance is actionable rather than aspirational.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly lean bullet-point sections that assume Claude's competence and avoid explaining known concepts, though the Philosophy and Anti-Patterns sections carry some abstract boilerplate that could be trimmed. | 4 / 5 |
Actionability | Despite repeatedly invoking AppleScript ('Run AppleScript or helper commands with minimal scope'), the body provides no concrete commands or examples, and the referenced files are metadata contracts rather than scripts — leaving only high-level hints without the specific steps to execute. | 2 / 5 |
Workflow Clarity | The Workflow section gives a clear sequenced process and the Validation section adds a fail-fast checkpoint with a retry loop, but the validation test itself is abstract ('the smallest command or test that exercises the changed behavior') with no concrete command. | 4 / 5 |
Progressive Disclosure | A dedicated Progressive Disclosure section points to verified one-level-deep references (contract.yaml, evals.yaml, task-profile.json) plus a clearly labeled legacy path, keeping the body an overview with easy navigation. | 5 / 5 |
Total | 15 / 20 Passed |