Conciseness

Nearly all content is actionable rules, commands, and examples with no explanations of concepts Claude already knows, but repeated heavy emphasis ("MUST READ", "CRITICAL & BLOCKING", "Do NOT generate... do not run... do not use") pads several sections. Not 5 because the emphasis and restatement in the project-ID and attribution sections could be trimmed without losing clarity.

Actionability

Fully executable guidance: a copy-paste curl command for resource descriptors, a dependency-install one-liner, a batch linter invocation, and a six-step metric-name normalization recipe with a worked example at every step, plus incorrect/correct PromQL snippets. Covers the common cases concretely.

Workflow Clarity

Clear sequence (resolve project -> inspect descriptors -> resolve resource filters -> choose aggregation -> format and validate) with a blocking gate when the project ID is missing, an explicit validation step, and a feedback loop that routes failures through the error-recovery reference before retrying. Not 4 because both checkpoints and error-recovery loops are explicit.

Progressive Disclosure

Three real, well-signaled, one-level-deep references plus a validation script, with clear section headers for navigation. Not 5 because the ~190-line body inlines sizable material (the full normalization recipe, state-label filtering rules) alongside a reference declared 'the single source of truth', a minor organization gap.

Specificity

"Generates valid PromQL queries from Cloud Monitoring metric descriptors and resource parameters" names the domain and one concrete action, but does not enumerate several distinct capabilities. Not 4 because coverage is limited to a single action rather than a list with only minor gaps.

Completeness

Explicitly answers what ("Generates valid PromQL queries from Cloud Monitoring metric descriptors and resource parameters") and when ("Use when asked to create, generate, write, or format PromQL queries...") with concrete trigger phrases. Matches the anchor-5 example; not below because neither half is missing or vague.

Trigger Term Quality

"create, generate, write, or format PromQL queries, PromQL strings, or PromQL aggregations" gives good natural keyword coverage across verb and object variations. Not 5 because a few natural phrasings users might say (e.g., charting or querying Cloud Monitoring metrics) are absent.

Distinctiveness Conflict Risk

Clear niche (PromQL generation for Cloud Monitoring) with an explicit negative boundary ("Don't use for raw metric discovery or metric selection") that separates it from the sibling metric-selection skill. Minimal conflict risk with other skills.