Content
86%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-sequenced instruction-only skill: it names the exact tools to call, gives a crisp categorization taxonomy, and includes approval and post-change verification. The only soft spots are the unnamed 'plan tools' in the application step and the absence of a recovery loop when verification shows waste continuing.
Suggestions
Name the specific plan tools used to apply negatives in step 4, so the execution step is as concrete as the read steps.
Add one recovery instruction after the re-pull verification, e.g. what to do when a negated term's waste persists or a converting term was accidentally excluded.
Optionally specify where the search-term export should be requested from and in what format, since step 1 falls back to it when query-level detail is missing.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | About 30 lean lines with no explanations of concepts Claude already knows; every line carries operational weight ('Reads do not require approval; adding negatives is a change, so confirm before shipping'). | 5 / 5 |
Actionability | Names concrete tools (list_connected_accounts, research_campaign_opportunity, pull_google_ads_performance) and gives specific categorization and level-selection rules, but 'Apply negatives through the plan tools' leaves the actual application tooling unnamed. Not 3: guidance is concrete and executable apart from that gap. | 4 / 5 |
Workflow Clarity | Clear 4-step sequence (read report, categorize, choose level, add on approval) with an approval checkpoint and a verification step ('Re-pull the search-term report after a few days to verify the waste stopped'). Not 5: no error-recovery loop telling Claude what to do if waste persists after negatives are applied. | 4 / 5 |
Progressive Disclosure | Under 50 lines with well-organized numbered sections and no need for external references (no references/, scripts/, or assets/ bundle exists; the keyword-research mention is a cross-skill pointer, not a nested file reference), satisfying the simple-skill exception. | 5 / 5 |
Total | 18 / 20 Passed |