Conciseness

Lean and dense throughout: no background concept is explained, every sentence is operational ('Build in a worktree with physical dependencies, never an external node_modules symlink. Preserve stdout/stderr, process receipts and the validator's own exit.'), and boundaries are stated once without padding. Fully matches anchor 5's 'every token earns its place'.

Actionability

Includes copy-paste commands ('npm run check:release-green -- --quick', '--full-ci --with-build --hermetic --json'), concrete script paths, and a specific handoff field list. Falls short of anchor 5 because the report verifier's usage is described but never shown ('use scripts/ci/release-green-result.mjs when available... supply the values from the run') and flag selection is delegated to 'Inspect scripts/quality/validate-release-green.mjs' rather than stated.

Workflow Clarity

Clear sequenced workflow — read the shared base doc, resolve SHAs, isolate a worktree, select a profile, execute, interpret against the exact base, hand off — with explicit validation checkpoints throughout ('Read the report even when the outer job is green', 'Missing, contradictory, stale, cancelled or timed-out evidence cannot establish PASS', 'Re-query live refs before using the result for a later decision') and a concrete handoff checklist. The skill is diagnostic, not destructive, so no cap applies.

Progressive Disclosure

A ~52-line, well-sectioned single file whose only delegated reference ('Read ../_shared/base-green.md before interpreting remote status') is clearly signaled and one level deep. Not a 5: that reference sits outside the skill's own bundle (no references/, scripts/, or assets/ directories exist alongside SKILL.md), so the essential shared guidance cannot be verified from the skill's structure — a minor organization gap against anchor 5's 'appropriately split' content.

Specificity

Names several concrete actions — 'Run or inspect the local Release-Green observer for an exact base or combined PR candidate, separating hard failures, advisory drift and incomplete evidence' — anchored to a specific named tool. Not a 5 because the observer's actual checks (static/test/package profile) are never named, leaving minor coverage gaps; clearly above anchor 3's '1-2 concrete actions'.

Completeness

Both parts are explicit: the 'what' ('Run or inspect the local Release-Green observer... separating hard failures, advisory drift and incomplete evidence') and an explicit 'Use for maintainer preflight' clause that also draws the boundary ('full CI acceptance remains a separate requirement'). Not a 5 because the 'when' names a role/context rather than concrete user-utterance triggers like 'when the user mentions X or asks for Y'.

Trigger Term Quality

Good keyword coverage for the intended audience: 'Release-Green observer', 'combined PR candidate', 'hard failures', 'advisory drift', 'maintainer preflight', 'CI acceptance'. Falls short of anchor 5 because natural synonyms and variations a user might say ('green check', 'preflight validation', 'release readiness') are absent.

Distinctiveness Conflict Risk

A named internal tool ('Release-Green observer') with a precise niche (local preflight of a base or combined PR candidate) and an explicit scope boundary; it would not plausibly trigger for any other skill, matching anchor 5's 'clear niche with distinct triggers; minimal conflict risk'.