Content
85%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-structured strategic reference with strong actionability, clear sequenced usage workflow, and exemplary progressive disclosure into two real companion references. The only weakness is conciseness: some quoted external prose and explanatory passages run longer than a lean skill body requires.
Suggestions
Trim or paraphrase the blockquoted external-doc prose (e.g., the Microsoft UIA 'designed for experienced C/C++ developers' quote and the Apple XCTest quotes) to only the clause that justifies a claim, since the citations already carry provenance.
Compress the 'Localisation rule of thumb' paragraph — its point is already captured by the locator-strategy table's brittle-locator column and could be a one-line callout.
Consider collapsing the two 'Operating the tests - deep references' and 'References' sections, which currently re-list the same two reference links, to remove repetition.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient and dense with tables/citations rather than padding, but the ~300-line body includes several blockquoted prose passages from external docs (e.g., 'UI Automation is designed for experienced C/C++ developers...') that restate background and could be tightened; the localisation rule-of-thumb and some quoted material are longer than strictly needed. | 2 / 3 |
Actionability | Provides concrete, executable guidance throughout: a per-OS/per-toolkit decision matrix, a locator-strategy table with specific identifiers (AutomationId, accessibilityIdentifier, setObjectName), named classes (XCUIApplication, XCUIElementQuery), and the references carry copy-paste commands (tccutil reset..., gsettings set org.gnome.desktop.interface toolkit-accessibility true). | 3 / 3 |
Workflow Clarity | 'How to use this reference' gives a clearly sequenced 4-step workflow (pick OS+toolkit, choose locators, apply async-wait primitive, pre-handle elevation hazard) with explicit ordering and a preventative checkpoint ('before it blocks the run'); the anti-pattern table pairs each failure with a concrete fix, providing recovery guidance. | 3 / 3 |
Progressive Disclosure | SKILL.md is a well-organized overview that pushes deep operational detail into two clearly signaled, one-level-deep references — references/async-waits-and-concurrency.md and references/platform-hazards-and-dpi.md — both of which exist as real bundle files and are linked with descriptive context. | 3 / 3 |
Total | 11 / 12 Passed |