Content
88%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, operationally dense skill body: every workflow ships executable commands, exact endpoints, and explicit validation and error-recovery loops. The weaknesses are length in a few over-dense sections and reliance on inline sections plus external URLs rather than bundled reference files for progressive disclosure.
Suggestions
Trim or externalize the densest inline sections — the vanity-URL and stale-base-protection paragraphs could be shortened to a rule plus a docs link, cutting the body's token cost without losing operational value.
Move secondary topics (folders, Drive usage, client attribution, state-file format) into bundled reference files under references/ and keep one-line pointers in SKILL.md, improving progressive disclosure from 4 to 5.
Deduplicate the docs pointers: '→ https://here.now/docs' appears in 'Current docs' and 'Beyond publish.sh', and a bare 'Full docs:' line closes the file — consolidate to one clearly marked section.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Almost every line carries skill-specific operational knowledge Claude cannot know (endpoints, flags, state-file semantics, claim-token rules), and it does not explain concepts Claude already knows. Not 5 because sections like the vanity-URL paragraph, the dense stale-base-protection paragraph, and 'What to tell the user' run long and could be tightened or pushed to the docs; not 3 because padding is minor relative to the volume of operational content. | 4 / 5 |
Actionability | Fully executable throughout: copy-paste commands ('./scripts/publish.sh {file-or-dir}', curl auth requests with complete JSON bodies, the credentials-file one-liner), a complete flag table, and exact API endpoints with response-field names ('publish_result.account_url', 'code: "version_conflict"'). Common cases are covered with specific examples. | 5 / 5 |
Workflow Clarity | Multi-step flows are clearly sequenced with explicit validation: create -> upload -> finalize ('the site is not live until finalize succeeds'), a numbered 5-step API-key acquisition flow, drift checking ('if the id differs from your state file's versionId, read the live files before editing'), and a version_conflict recovery loop with two explicit remediation options. The destructive update path has stale-base validation, so no cap applies. | 5 / 5 |
Progressive Disclosure | Good structure: the body points to one-level-deep, clearly signaled external references (the '→ https://here.now/docs#…' markers) and both bundled scripts ('./scripts/publish.sh', './scripts/drive.sh') exist in the bundle, and detail is deliberately deferred to the live docs and openapi.json. Not 5 because no reference files exist in the bundle and several self-contained topics (folders, Drive, client attribution) are inlined in SKILL.md that could be split into reference files; not 3 because navigation is clear and nothing is buried. | 4 / 5 |
Total | 18 / 20 Passed |