Content
100%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 API-skill body: lean, entirely non-obvious content organized into scannable sections, with executable code, concrete parameter examples, full error semantics, and a content-safety guardrail. Nothing needs to be moved or trimmed.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every section carries non-obvious, API-specific facts — auth header and its Bearer-vs-API-key distinction, entitlement gating, parameter table, response field semantics, error codes — with zero explanation of concepts Claude already knows. Nothing is padded and every token earns its place. | 5 / 5 |
Actionability | The body is copy-paste executable: a complete curl with headers and urlencoded params, a real example key format, concrete HS/reporter code examples (842, 156, 2709, 8542), a worked JMESPath projection, a full response schema, and jq post-processing. Not 4 because there are no gaps — even rate-limit backoff and upstream-staleness semantics are specified. | 5 / 5 |
Workflow Clarity | This is a single-call API skill whose one action is unambiguous (auth header -> endpoint -> params -> interpret response), and error-recovery guidance is explicit: 401/403/429 meanings, retry with backoff, and the caveat that 'upstreamUnavailable: true ... does not prove zero trade'. The 'When NOT to use' section adds a disambiguation checkpoint. No destructive or batch operation, so no validation cap applies. | 5 / 5 |
Progressive Disclosure | The skill is self-contained with no bundle directories (references/, scripts/, assets/ are empty), and all inline content is appropriate for a single-endpoint skill. External references — OpenAPI operation 'ListComtradeFlows' and the auth-matrix doc — are one level deep, clearly signaled in a References section, and easy to navigate. | 5 / 5 |
Total | 20 / 20 Passed |