Content
85%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, highly actionable instruction skill with an excellent evidence-gated workflow and clean one-level-deep disclosure into a real templates file. Its main cost is density and repetition: the scope boundary and mode deliverables are each stated multiple times, so the body could be materially shortened without losing guidance.
Suggestions
State the STAR/scope boundary once (in the Scope guard section) and have the intro and Next Best Skill sections reference it instead of restating it, cutting three redundant passages.
Enumerate the mode deliverables once — in the Skill Contract's Expected output — and reduce the intro and Instructions step lists to mode names and step titles.
Trim the connector paragraphs in Data Sources to a single purpose clause each (e.g., "social listening — community language, topics, brand sentiment") since the detail also lives in CONNECTORS.md.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body assumes Claude's competence and never explains known concepts, but states the STAR/scope boundary three times (intro "computes neither the Suitability/Trust/Appeal/Return dimension scores nor the SQS", the Scope guard paragraph, and again in Next Best Skill) and enumerates the mode deliverables in the intro, the Skill Contract, and the Instructions. Not 4: the redundancy is more than minor; not 2: there is no padded concept explanation, just tighten-able repetition. | 3 / 5 |
Actionability | Concrete, executable guidance throughout: literal Quick Start commands ("Analyze the target audience for [brand/product/category]"), steps that name their sub-fields and cite verified template anchors (§A3, §N3), exact save paths (memory/influencer/audience-mapper/YYYY-MM-DD-<topic>.md), and an explicit evidence gate. Not 4: the specific execution detail is fully specified for the common cases of both modes. | 5 / 5 |
Workflow Clarity | Clear sequenced workflow — set mode → evidence gate → branch steps → done-when — with explicit validation ("Unsupported fields stay TBD and produce a query plan"), a NEEDS_INPUT error-recovery path with exact queries, and a 3-item "Done when" checklist. Not 4: both explicit checkpoints and error-recovery feedback are present. | 5 / 5 |
Progressive Disclosure | The body is a genuine overview; fill-in detail lives one level deep in references/templates.md, which exists with headings matching every cited anchor (§1, §A2–A9, §N2–N7), and each reference is signaled with its purpose. Not 4: content is appropriately split, navigation is by explicit section anchor, and no bulk reference material is inlined. | 5 / 5 |
Total | 18 / 20 Passed |