Content
70%Weight 40%Scale 1-3Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is a well-structured, lean overview with clear sequenced workflow, validation checkpoints, and well-signaled one-level references. Its weaker points are mild reinforcement redundancy across sections and drafting steps that lean on external cards rather than self-contained content examples.
Suggestions
Consolidate the repeated 'never verbatim cross-posting' and claims-handling statements into one canonical location and reference it, to remove redundancy and tighten conciseness.
Add one short worked example (e.g., a sample X thread or 小红书 note with variant tags applied) so the drafting step is self-contained rather than deferring entirely to external norm/voice cards.
Inline the hook taxonomy and tag schema as a compact reference block within step 3/step 8 so the actionability does not depend on recall of external cards.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body assumes Claude's competence (no concept tutorials) and every section is purposeful, but key constraints are restated across sections — "never verbatim cross-posting" appears in both the intro and step 4, and claims handling is revisited in the intro, step 6, and references — so it could be tightened. It is not level 3 because of this reinforcement redundancy, and not level 1 because it never pads with concepts Claude already knows. | 2 / 3 |
Actionability | Operational guidance is concrete — named read/write paths (memory/projections/narrative.json, registry-events.py), a hook taxonomy, the {formula | hook family | CTA type | signal optimized} tag schema, and ~70%-of-max char budgets — but the core drafting steps ("Draft each package natively") defer the actual spec to external norm/voice cards with no worked content example. It is not level 3 because nothing is copy-paste ready or exemplified, and not level 1 because concrete paths, budgets, and schemas are specified. | 2 / 3 |
Workflow Clarity | Eight numbered steps are clearly sequenced with explicit validation checkpoints — the Done-when bar, the step-7 negative checklist, claims "block publish-ready status", and stale-card routing to platform-norm-profiler. It is not level 2 because checkpoints are explicit with error-recovery routing rather than implicit. | 3 / 3 |
Progressive Disclosure | The body is a well-organized overview with one-level-deep, clearly signaled references (echo-benchmark.md, skill-contract.md, humanizer-slop.md, SECURITY.md, sibling skills) split into Reference Materials and Next Best Skill sections. It is not level 2 because references are signaled and shallow, not buried or chained two-plus levels deep; no bundle files are present so structure is evaluated on the reference organization. | 3 / 3 |
Total | 10 / 12 Passed |