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 compact skill body: a concrete verification loop with explicit validation and feedback steps, lean non-obvious guidance throughout, and appropriate deferral of version-sensitive syntax to the CLI's own docs. The only minor gap is that a few common command invocations (snapshot, wait) are named but not shown.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~45-line body is lean with every token earning its place; it teaches only non-obvious operational knowledge ('References can become stale after navigation or a substantial DOM update', 'a completed screenshot command does not prove that the intended state was visible') with no filler or explanation of concepts Claude already knows. | 5 / 5 |
Actionability | Provides mostly executable guidance: copy-paste commands ('agent-browser skills get core --full', the dev-bypass URL) and named subcommands ('agent-browser console', 'agent-browser errors', 'agent-browser network requests'). It falls short of fully copy-paste-ready coverage because snapshot and wait command syntax is deferred to the CLI's own docs rather than shown for common cases. | 4 / 5 |
Workflow Clarity | The 7-step verification loop is clearly sequenced with explicit validation checkpoints (wait for state, re-snapshot, verify result/console/network), a feedback loop ('Check agent-browser console and agent-browser errors when the UI behaves unexpectedly'), and a pre-success gate ('Inspect the saved image before reporting success'). | 5 / 5 |
Progressive Disclosure | A single-purpose skill under 50 lines with no bundle files and no need for external references; it is well-organized into clear sections and appropriately defers version-sensitive command syntax one level deep to the CLI's version-matched instructions, satisfying the simple-skill exception. | 5 / 5 |
Total | 19 / 20 Passed |