Content
86%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 well-structured CLI skill body: executable examples, a gotchas table, command/subcommand tables, and clean one-level-deep reference routing. It is concise for its breadth, with only minor room to tighten and to make validation feedback loops more explicit.
Suggestions
Add an explicit validate→fix→retry loop for destructive/batch operations (e.g., after broadcasts send or contacts imports, show 'verify with emails get / logs get; on failure, fix payload and re-run') to lift workflow_clarity above 4.
Trim a few mildly explanatory lines (e.g., 'Prefer a package manager when available') to push conciseness toward the leanest anchor.
Consider a short 'verify after send' checklist item alongside the Common Patterns send example so the validation checkpoint is visible inline rather than only in the Common Mistakes table.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Dense, fluff-free reference material with no padding about what email/CLI concepts are, and it assumes Claude's competence; the sheer breadth (~160 lines across many command groups) carries inherent volume, with a few mildly explanatory lines that could be trimmed, placing it just above the midpoint rather than at the leanest anchor. | 4 / 5 |
Actionability | Copy-paste-ready bash commands throughout — install checks, 'Common Patterns' (send, inline image, React Email, domain flow, broadcast, CI/CD, doctor), a subcommand table, and exact flag contracts — fully executable and covering the common cases. | 5 / 5 |
Workflow Clarity | Multi-step flows are sequenced (domain create → verify → get; broadcast create with --send) and validation gates exist (--dry-run for sends, 'resend doctor', 'emails get <id>' to check last_event), but explicit validate→fix→retry feedback loops are implied rather than spelled out for destructive/batch operations. | 4 / 5 |
Progressive Disclosure | SKILL.md is a clear overview pointing to 17 real one-level-deep reference files, each mapped to a task in the 'When to Load References' section; all referenced paths exist on disk and navigation is easy. | 5 / 5 |
Total | 18 / 20 Passed |