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.
A well-crafted, tool-invocation-style skill body: executable JSON examples, an explicit mode-selection table, anti-patterns, and negative-scope guidance, with no filler or background explanations. Its single weakness is redundancy — the Modes table and the dedicated mode sections repeat the same information almost verbatim.
Suggestions
Drop the standalone 'Schema Mode' and 'Brand Mode' sections (or the corresponding Modes table rows) — the table already states the provenance and null-on-hallucination behavior verbatim, so pick one location for each.
State the 'always prefer structured over tables' rule once (e.g., in the Modes section) instead of repeating it in the Modes emphasis, Anti-Patterns, and the Quick Reference comment.
In the description, add a short deflection for plain page content (e.g., 'for markdown content use wigolo-fetch') to mirror the existing agent-tool boundary and sharpen single-page conflict behavior.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and teaches only wigolo-specific behavior (modes, anti-patterns, token budget) rather than concepts Claude already knows. However, the Modes table rows for `schema` and `brand` are restated nearly verbatim in the "Schema Mode" and "Brand Mode" sections, and the "use structured, not tables" guidance appears three times (Modes emphasis, Anti-Patterns, Quick Reference comment) — minor trimming would earn the top score. | 4 / 5 |
Actionability | The Quick Reference gives copy-paste-ready tool-call payloads for every common case (structured, schema, selector, metadata, raw-HTML), plus a concrete `named_schema` example and a decision table with explicit "when to use" guidance. This is fully executable instruction covering the common cases. | 5 / 5 |
Workflow Clarity | This is a simple single-action skill (choose a mode, invoke the tool) and the mode-selection decision is unambiguous: `structured` is marked as the default in three places, alternatives have explicit "when to use" criteria, and "When NOT to use" plus Anti-Patterns handle the failure paths. No destructive or batch operations exist that would require validation checkpoints. | 5 / 5 |
Progressive Disclosure | There are no bundle files (no references/, scripts/, or assets/ directories) and none are needed at this size; the body is compact and organized into clearly headed sections with all detail appropriately inline. The only external pointers are the one-level-deep, clearly signaled "See Also" links to sibling skills — no nested or buried references. | 5 / 5 |
Total | 19 / 20 Passed |