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 SKILL.md body: concrete commands and parameters throughout, a well-sequenced workflow with rejection criteria and a real validation loop, and a clean split between overview and the bundled script. The only improvement space is trimming the slight overlap between Overview and Notes and the duplicated CLI defaults.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and efficient with no explanation of concepts Claude already knows, but minor instances could be trimmed: the Notes section partially restates the Overview's scroll-video caveat, and the options block duplicates the script's own CLI defaults ('--viewport 1440x1100 # default', '--step 950', '--wait 2000'). Fits 'efficient; minor instances of over-explanation' rather than 'every token earns its place'. | 4 / 5 |
Actionability | Fully executable guidance: a copy-paste-ready node invocation with the manifest argument, concrete option flags with defaults, parameterized workflow steps (viewport-sized overlapping steps, ~2s settle wait), and runnable validation commands (sips, git diff --check, node checks, ffprobe). | 5 / 5 |
Workflow Clarity | A clear 9-step sequence (warm-up scroll, return to top, stepped scrolling, settle, capture, stitch, crop, validate) plus explicit rejection criteria for bad candidates and a dedicated Validation section with executable checks — an explicit feedback loop for what is a batch repair operation. | 5 / 5 |
Progressive Disclosure | The body is a well-organized ~75-line overview with clear section headers; the single bundle script (scripts/stitch_full_page_capture.mjs, verified to exist) is referenced with an exact path and summarized rather than inlined, and there are no nested or buried references. | 5 / 5 |
Total | 19 / 20 Passed |