Content
78%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 disciplined, security-conscious instruction-only skill body: lean token use, concrete env vars/scopes/commands, and unusually strong validation and consent checkpoints for a destructive publishing operation. Its main defect is that the central drafting workflow is delegated to workflows/publish.md, which is missing from the bundle, leaving the most important procedure unreachable.
Suggestions
Ship or restore workflows/publish.md in the bundle (or inline the drafting steps into SKILL.md) so the 'follow workflows/publish.md' link resolves — the core procedure is currently unreachable.
Add one complete example invocation, e.g. `scripts/youtube-publish publish --draft draft.json`, showing how the executable subcommand, draft path, and env vars combine.
Turn the post-failure guidance into an explicit recovery loop (check channel state via channels.list → report result to user → re-approve if needed) so interrupted-write handling matches the precision of the pre-write checks.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body (~30 lines) is lean and operational throughout — 'The consumer supplies YOUTUBE_CHANNEL_ID and YOUTUBE_APPROVER_EMAIL in the command environment', 'Use youtube.force-ssl for comments and replies' — with no padding and no explanation of concepts Claude already knows; every sentence carries a constraint, env var, scope, or policy pointer. | 5 / 5 |
Actionability | Concrete, near-executable guidance is present: named env vars, exact OAuth scopes ('youtube.upload', 'youtube.readonly', 'youtube.force-ssl'), the executable path 'scripts/youtube-publish' (verified to exist), and the command 'Run publish --draft PATH only after that yes'. Minor gaps keep it below 5: no full example invocation, and the drafting procedure itself is delegated to workflows/publish.md, which is absent from the bundle. | 4 / 5 |
Workflow Clarity | The draft → approval_record comment → publish sequence is clear with real validation checkpoints for a destructive operation: 'The script checks the token's channel with channels.list before a write', 'It checks the author, decision, item and digest before any YouTube request', and the error-recovery rule 'Stop and check the channel with the user before any new attempt. Do not repeat a successful request'. It is not 5 because the drafting steps themselves live in the missing workflows/publish.md and recovery is stop-and-check rather than an explicit validate-fix-retry loop. | 4 / 5 |
Progressive Disclosure | Sections are well-organized and the body correctly stays an overview, but the single key reference — 'follow [workflows/publish.md](workflows/publish.md)' — points to a file that does not exist in the bundle (only scripts/youtube-publish and scripts/youtube-publish.py are present), breaking navigation to the core workflow. This is more than the 'minor organization gaps' of a 4, so it lands at 3. | 3 / 5 |
Total | 16 / 20 Passed |