Content
81%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.
The body is a strong, contract-driven instruction skill: an unambiguous 8-step workflow with real validation gates, concrete formulas, and disciplined labeling/escalation rules. Its weaknesses are moderate — repeated scope/boundary restatements, no worked example of the deliverable spec, and reliance on cross-skill references that live outside the skill bundle and cannot be verified from this directory.
Suggestions
State the scope/boundary line once (opening paragraph or Skill Contract) and have steps 1, 7, and the Reference Materials section reference it by name instead of restating it, reclaiming several lines of token budget.
Add a short worked example or row-level template of the value-model spec / QA sheet (one revenue conversion with margin adjustment, one non-revenue proxy with its derivation) so the deliverable format is concrete rather than described in prose.
Consolidate the external file links into the Reference Materials section and confirm the target paths resolve within the published bundle, so progressive disclosure does not depend on files outside the skill directory.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is information-dense with no concept-teaching padding, but the scope/boundary line (values vs signal-fires vs R1/R2 scoring) is restated in the opening paragraph, step 1, step 7, Reference Materials, and Next Best Skill — redundant trimmable material. It is not a 3 because nothing explains concepts Claude already knows and every section carries actionable content. | 4 / 5 |
Actionability | Guidance is concrete for an instruction-only skill: exact formulas (net_value = revenue × margin (or revenue − COGS); proxy_value = avg_order_net_value × lead→sale close_rate), a Measured/User-provided/Estimated labeling scheme, specific output paths, and needs-input escalation rules instead of guessing. It falls short of 5 because the deliverable itself (value-model spec and QA sheet) is described only in prose — no worked example or row-level template shows what a finished spec looks like. | 4 / 5 |
Workflow Clarity | Eight clearly sequenced steps with explicit validation checkpoints — the step-7 value-vs-count reconciliation (with a 3× phantom-value flag example) and the step-8 launch-readiness gate that requires listing exactly what to fix — plus error-recovery routing (needs-input states, hop to attribution-reconciler or conversion-signal-qa) and a "Done when" checklist. This matches the anchor with explicit validation, feedback loops, and a checklist rather than merely having most checkpoints. | 5 / 5 |
Progressive Disclosure | Sections are well organized and every reference is signaled with its purpose, and the core instructions appropriately live inline. It is not a 5 because the skill ships no local bundle (references/, scripts/, assets/ are absent) while pointing to files two-to-three directory levels outside itself (../../scale/attribution-reconciler/SKILL.md, ../../../references/skill-contract.md, ../../../CONNECTORS.md), and the same links are scattered inline across nearly every section rather than consolidated once — minor organization gaps rather than a fully self-contained or cleanly split structure. | 4 / 5 |
Total | 17 / 20 Passed |