Conciseness

The body is dense rule-directives with no concept explanations Claude already knows ("A baseline that refuses input or performs different work is a control, not a performance comparison" is a rule, not padding). Fits anchor 4 (efficient, minor trims possible) rather than 5 due to slight redundancy between the repair paragraphs and the compressed rerun guidance.

Actionability

Directives are methodologically concrete ("Run the baseline first and prove one planted slowdown fails the budget", "Use `pstack:benchmark-checklist`", "Discover runners from... `benchmarks/editor`") but there are no executable commands, code, budget values, or concrete runner examples — the reader is told what to do, not how to run anything. Matches anchor 3 (some concrete guidance but incomplete, key details missing); not 4 because nothing is copy-paste executable.

Workflow Clarity

Paragraphs follow a real sequence (discover runners → record identity → select cohorts → freeze budgets → baseline with planted slowdown → alternate samples → correctness guards → repair measurement first → rerun on final bytes → checklist gate) with explicit checkpoints ("prove one planted slowdown fails the budget", "Use `pstack:benchmark-checklist` before accepting or reporting"). Fits anchor 4 (clear sequence, most checkpoints present); the sequence is prose-ordered rather than enumerated, keeping it below 5.

Progressive Disclosure

Under 50 lines with no bundle files present; the external reference `pstack:benchmark-checklist` is clearly signaled and one level deep. Fits anchor 4 (good structure, minor organization gaps); not 5 because the body has no section headers to organize the dense paragraphs, which the simple-skill exception requires for a 5.

Specificity

"Measure, review or repair Plate performance" names the domain and several actions, but the verbs are generic and the qualifiers ("honest cohorts, baseline identity, correctness guards and causal proof") are jargon attributes rather than concrete capability statements like "extracts text, fills forms". Matches anchor 3 (domain plus a few actions, not comprehensive); below anchor 4 because no action names a concrete operation or deliverable.

Completeness

Both parts are explicit: the *what* ("Measure, review or repair Plate performance with honest cohorts...") and the *when* ("Load before choosing, running or waiving a scale or latency gate"). Not 5 because the when-clause is jargon-tinged ("waiving a scale or latency gate") and less naturally phrased than anchor 5's concrete trigger examples; comfortably above anchor 3 since the when is explicit, not implied.

Trigger Term Quality

Natural terms present: "performance", "latency", "scale", "measure", "repair", "gate" — good coverage of the perf-gate domain. Missing common synonyms such as "benchmark", "slow", "speed", so it fits anchor 4 (good coverage, a few natural terms missing) rather than anchor 5, and is clearly above anchor 3.

Distinctiveness Conflict Risk

The "Plate" qualifier plus "scale or latency gate" scopes a clear niche with minimal conflict risk, matching anchor 4 (mostly distinct, minor overlap). Not 5 because the generic "measure, review or repair performance" core would overlap a general profiling skill absent the Plate qualifier.