Content
78%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 tight, well-organized instruction-only skill: concrete formulas, named tools, explicit anti-fabrication rules, and clear decision routing. Its main gaps are the absence of a worked end-to-end numeric example and an implicit rather than explicit step ordering across sections.
Suggestions
Add a short worked example that takes real spend and revenue inputs through ROAS, break-even, and profit to a scale/kill verdict, covering the most common case.
Make the workflow sequence explicit — e.g., a numbered flow (confirm account → pull figures → compute blended and per-platform → reconcile attribution → report with decision) — so checkpoint ordering is unambiguous.
Trim minor over-explanations of concepts Claude already knows (e.g., 'What one conversion costs', 'A ratio, before costs') to tighten token efficiency.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean — terse formulas, short sections, no padded explanations — but includes minor over-explanation Claude already knows ('What one conversion costs', 'A ratio, before costs') and a slightly flowery opener ('the numbers that decide whether a campaign lives'). Not 5 because a few phrases could be trimmed; not 3 because there are no padded sections or unnecessary explanations of bulk material. | 4 / 5 |
Actionability | Guidance is concrete: named tools ('pull_google_ads_performance', 'list_connected_accounts'), exact formulas, worked margin anchors ('At 50% margin you break even at 2.0x; at 25% margin, 4.0x'), and explicit decision routing to other skills. Not 5 because there is no end-to-end worked example taking real spend/revenue through to a profit verdict; not 3 because the guidance is directly executable, not high-level hints. | 4 / 5 |
Workflow Clarity | A clear sequence is present — confirm the account, pull real figures, compute blended and per-platform, reconcile attribution, report, decide — with checkpoints ('ask rather than assume', 'Reconcile with the attribution skill before trusting a single platform's ROAS'). Not 5 because the sequence is distributed across sections rather than an explicit ordered flow, leaving the checkpoint ordering implicit; not 3 because validation checkpoints are explicitly stated. | 4 / 5 |
Progressive Disclosure | The skill is under 50 lines with no need for external references, and the sections (core numbers, how to use it, report, rules) are well-organized; cross-skill pointers (attribution, optimize, kill-scale-rules, mmm-budget-planner) are clearly signaled and one level deep. Per the simple-skill guideline, this earns the top anchor. | 5 / 5 |
Total | 17 / 20 Passed |