The body is lean and command-driven; its few asides ("keep the multi-thousand-line log out of context", the gradle-runner mutation warning) convey non-obvious gotchas rather than concepts Claude already knows, matching the level-3 lean anchor.

Every step ships copy-paste-ready executable commands (bootstrap exports, git diff, the gradle line, git status, git checkout), meeting the level-3 fully-executable anchor rather than the level-2 pseudocode case.

A clear numbered 1–5 sequence with an explicit validation checkpoint in step 4 ("confirm the only changes are yours: git status --short") and a feedback loop (treat PASS as suspect and re-run inline), matching the level-3 anchor for risky push-gate operations.

A single well-organized file under 50 lines with no external references needed; per the rubric's simple-skill note, well-organized sections alone justify the level-3 anchor here.

Lists multiple concrete actions — "bootstrap, then spotlessApply/spotlessCheck/detekt/assembleDebug/test/allTests", "restore the host-render screenshot diff", and "report a single pass/fail" — matching the level-3 anchor for several specific concrete actions rather than the level-2 partial coverage.

It states both what (the full verification task list and report behavior) and an explicit when ("Use before every push"), satisfying the level-3 anchor for explicit triggers instead of the level-2 implied-when case.

Natural trigger phrasings a developer would actually say are present — "pre-push baseline", "before every push", "CI fails" — giving good coverage rather than the level-2 partial-keyword case.

Scoped to "Meshtastic-Android" with repo-specific tasks and the gradle-runner subagent, giving it a clear niche unlikely to fire for unrelated skills, matching the level-3 anchor.