Content
68%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 highly actionable, with concrete, mostly copy-paste-ready AppleScript and peekaboo commands plus honest limitations and rules. Its main weakness is the absence of validation checkpoints in the capture-and-analyze loops, leaving workflows without error recovery. Organization is clean for a single-file skill.
Suggestions
Add verification steps between capture and analysis, e.g. "Check the screenshot is non-empty and shows the map before running vision_analyze; if the window was missed, re-activate FindMy and recapture."
Fill the placeholder in the AirTag tracking workflow (step 2) with a real command, such as the peekaboo click invocation from Method 2, and note that `screencapture -w` interactively waits for a window selection, which breaks the unattended loop.
De-duplicate the foreground/AirTag-update guidance, which appears in both the Workflow comments and Limitations and again in Rules, to reclaim tokens.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes Claude's competence (no concept filler; "Since Apple doesn't provide a CLI" is brief, justified context), with only minor redundancy such as the foreground-update rule repeated in both Workflow and Rules. | 4 / 5 |
Actionability | Mostly executable guidance with concrete osascript, screencapture, and peekaboo commands, but the AirTag workflow's step 2 ("Click on the AirTag item") has no actual command, and the vision_analyze blocks are tool invocations rather than shell-executable steps. | 4 / 5 |
Workflow Clarity | Methods 1 and 2 have clear sequences, but no verification checkpoints exist (e.g., confirming the screenshot captured the map before analyzing it), and the AirTag patrol loop is a batch operation without validation, which caps workflow clarity at 3. | 3 / 5 |
Progressive Disclosure | No bundle files exist (references/, scripts/, assets/ are all absent) and the ~120-line body is well organized into clearly headed sections (Prerequisites, When to Use, Methods, Workflow, Limitations, Rules); it misses a 5 only because the body exceeds the under-50-line simple-skill exception. | 4 / 5 |
Total | 15 / 20 Passed |