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 body: copy-paste-ready commands covering every workflow, explicit success/failure validation and a retry loop, and a lean overview that delegates detail to six well-signaled, verified reference files. The only nit is a handful of guardrail sentences that could be tightened.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and operational — CLI invocations, decision rules, and error handling with no explanations of concepts Claude already knows — but a few guardrail sentences run slightly long ("Loading or reading it does not require a second approval when the user already asked for its use", "it need not meet an additional complexity threshold"). This fits the efficient-with-minor-trimmable-instances anchor rather than the every-token-earns-its-place top anchor. | 4 / 5 |
Actionability | Commands are copy-paste ready with realistic arguments: `skillnet search "pdf" --limit 5 --json`, the vector-mode variant, `skillnet download "<skill_url>" --target-dir "./downloaded_skills" --json`, `skillnet create --prompt "A skill for checking CSV headers..." --evaluate --json`, `skillnet evaluate "./generated_skills/example" --json`, plus `doctor` and `python -m skillnet_ai` fallbacks. These cover all common cases; the one delegated workflow (routing) defers its command to a reference file, which is appropriate disclosure rather than a gap. | 5 / 5 |
Workflow Clarity | Sequences are clearly ordered per workflow (search → choose → download → read → inspect → apply; create → structure check → evaluate), and validation is explicit: "A successful command has ok: true. A failed command exits nonzero", `skillnet doctor --json` for config validation, structure checked before evaluation, and a concrete recovery loop ("If creation succeeded but evaluation failed, preserve data.paths and retry evaluation, not creation"). Explicit checkpoints with feedback loops match the top anchor. | 5 / 5 |
Progressive Disclosure | The 128-line body is an overview with executable commands, while all detail lives in six references (routing, setup, platforms, workflow-patterns, api-reference, security-privacy) — all verified to exist — each linked from the body with a clear purpose signal (e.g. "Read [api-reference.md](references/api-reference.md) for flags, output fields, provider compatibility and error recovery"). References are effectively one level deep: the few cross-links between reference files (api-reference → routing, platforms → setup) target files also directly linked from SKILL.md, so navigation stays flat. Scripts in scripts/ are reachable via workflow-patterns.md. | 5 / 5 |
Total | 19 / 20 Passed |