The body is dense decision matrices and tables with no concept explanations Claude already knows (no 'what a CI is' or 'what a test is' padding); nearly every token earns its place. Not a 2 because the length is decision-content density rather than unnecessary explanation that could be tightened.

Provides concrete, copy-paste-ready patterns such as 'concurrency: group: ${{ github.workflow }}-${{ github.head_ref }}' + 'cancel-in-progress: true', 'interruptible: true', 'buildDiscarder(logRotator(...))', and '::add-mask::$VALUE', plus specific shard counts and timeout bands. Not a 2 because the guidance is executable rather than pseudocode or abstract direction.

A sequenced 6-step 'How to use this reference' workflow with a closing 'Then re-check the workflow against the conventions below' checkpoint, reinforced by a fully walked 25-minute Playwright worked example and a flake tag->quarantine->review feedback loop. Not a 2 because explicit checkpoints and a feedback loop are present, and this is a reference skill rather than a destructive batch operation.

SKILL.md is a decision surface that defers per-CI reporting and per-language reporter/cache-key lookups to a single, clearly signaled, one-level-deep companion at references/junit-and-cache-lookups.md (a real bundle file). Not a 2 because content is appropriately split and navigation is explicit rather than inline-bloated or nested.

Lists many specific concrete conventions - 'when to shard (and how many shards), retry policy (which failures are safe to retry), flake-quarantine integration, artifact retention, per-trigger cadence ... per-job timeouts, secret management, and cross-CI portability' - matching the 'lists multiple specific concrete actions' anchor. Not a 2 because it is comprehensive rather than naming only some actions; third-person voice, so no voice penalty.

Clearly answers 'what' (the enumerated conventions) and provides explicit trigger guidance via 'Use as the team's reference for CI test-workflow design across GitHub Actions / GitLab CI / Jenkins / CircleCI'. Not a 2 because explicit use guidance is stated rather than merely implied, so the missing-'Use when' cap does not apply.

Covers natural terms a CI user would say - 'GitHub Actions / GitLab CI / Jenkins / CircleCI', 'shard', 'retry policy', 'timeouts', 'secret management' - alongside more specialized phrasing, giving good coverage of natural trigger terms. Not a 2 because keyword coverage well exceeds the 'Works with PDF files' bar.

Occupies a clear cross-CI conventions niche explicitly distinguished from per-CI implementation skills ('per-CI reporting ... live in references/'), making wrong-skill conflicts unlikely. Not a 2 because the scope is sharply bounded rather than overlapping with sibling skills.