Content
82%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.
A highly actionable, well-organized single-file skill: executable commands and complete code examples for every usage path, with valuable non-obvious operational details throughout. The main improvement opportunities are trimming repeated URL/flag duplication and splitting the larger API sections into reference files.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with non-obvious specifics ("--screenshot-delay <ms>... defaults to 100", "PNG pixel size is width * dpr by height * dpr", "Treat waitForFrame() as 'a frame was submitted'") and explains nothing Claude already knows. It is not a 5 because the sample gallery URL is repeated five times across examples and the CLI-rules bullets partially restate flag behavior already shown in the commands, so a modest trim is possible. | 4 / 5 |
Actionability | Every path is copy-paste ready: complete CLI invocations with all flags, and full TypeScript examples with imports, try/finally cleanup, and error callbacks covering remote URLs, local file URLs, CDP inspection, data updates, windowed input, and OpenCard managers. Specific examples cover the common cases; nothing is pseudocode. | 5 / 5 |
Workflow Clarity | Each usage path is an unambiguous, correctly sequenced flow (construct view → load template → wait/screenshot → destroy in finally), with explicit decision guidance ("Prefer the CLI for one-off screenshots... Prefer the API when the task needs programmatic setup...") and error-recovery hooks (onCardError, try/finally). It is not a 5 because there are no verification checkpoints for the produced output (e.g., confirming the PNG was written or retrying on capture failure), leaving minor validation gaps; it is above 3 because sequences and cleanup/error checkpoints are consistently present. | 4 / 5 |
Progressive Disclosure | No bundle files exist (references/, scripts/, assets/ are absent) and the body has no dangling references, so all content is appropriately contained in a single well-sectioned file (CLI, TypeScript API, Windowed API, DebugRouter OpenCard, Operational Notes) that is easy to navigate. It is not a 5 because at roughly 250 lines the skill exceeds the under-50-line simple-skill case, and clearly separable material (the full OpenCard manager example or the CLI flag reference) could be split into one-level-deep reference files to keep SKILL.md a leaner overview. | 4 / 5 |
Total | 17 / 20 Passed |