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 orchestration skill body: an 8-step workflow with real validation checkpoints and fallback routing, concrete paths and format specs, and clean one-level-deep reference signaling. The main improvement levers are trimming the repeated never-posts/scope statements and inlining a compact example of a package block and variant tag so the output format is executable without chasing external files.
Suggestions
State the never-posts rule once (e.g., in the Skill Contract) and drop the repetitions in the Scope guard, Save Results, and step 8 to reclaim tokens.
Add one compact ready-to-paste example package (copy + variant tag {formula | hook family | CTA type | signal optimized}) so the delivered shape is concrete without opening external files.
Inline the required handoff-summary field list or a minimal example of it, since the body currently delegates the shape entirely to skill-contract.md while also naming four required fields.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Dense and operational with no explanation of concepts Claude already knows, but the never-posts rule is stated ~4 times ('this skill never posts', 'No posting, scheduling, or engagement automation', 'Saving does not authorize posting', 'never marks a package published') and the Scope guard paragraph repeats exclusions already in the description. Fits 'efficient; minor instances of over-explanation that could be trimmed' — not 3, since there is no genuinely unnecessary explanation, only repetition. | 4 / 5 |
Actionability | Concrete guidance throughout: exact paths ('memory/projections/narrative.json', 'memory/social/social-creative-builder/YYYY-MM-DD-<topic>.md'), a named hook taxonomy, 'char budget at ~70% of the platform max', and '小红书 3:4, IG 4:5, LinkedIn document PDF' artboards. Not 5: there is no copy-paste example of a package block, variant tag, or handoff summary — the output shape is delegated ('Emit the standard shape from skill-contract.md'), leaving minor gaps for the common cases. | 4 / 5 |
Workflow Clarity | Eight clearly sequenced numbered steps with explicit validation checkpoints (truth-state confirmation at named offsets, claims/rights blocking in step 6, the negative checklist in step 7) and error-recovery routing ('A missing or stale card → route to platform-norm-profiler', 'Any edit requires a new version and re-gate'). Not 4: checkpoints and fallback loops are explicit, and the destructive/batch cap does not apply to this content-drafting skill. | 5 / 5 |
Progressive Disclosure | The body is a clear orchestrating overview; all detail is delegated via well-signaled, one-level-deep references, each given a purpose note in Reference Materials ('echo-benchmark.md — ECHO framework…', 'humanizer-slop.md — pre-handoff AI-tell strip'), with links also appearing inline at point of use. Content is appropriately split to norm cards, the contract, and sibling skills; navigation is easy. No bundle files exist, and no nested (2+ level) references appear. | 5 / 5 |
Total | 18 / 20 Passed |