Script-tooling tests for the SQLite reliability bench, CI workflow guards and git-owner auth fixture, Docker build helper, iOS node e2e, the OpenClaw e2e instance, Parallels smoke model, plugin SDK API diff, unchanged-plugin Docker update, Android Gradle runner, and Vitest worker-artifact verification polled files, ports, and PIDs against 500 ms to 15 s deadlines, and raced child readiness/exit against timers of up to 30 s, so a fixture child that outlasted the bound on a loaded host failed the test.
Waits now await child IPC, inherited readiness pipes, durable fixture receipts, native exit/close promises, and existing read-admission deferreds; Python fixtures join their retained subprocess handles under the caller's lifetime guard. Suspended test bodies are bound to the Vitest signal and fixture roots are retained until native children settle; the tracked Docker build awaits its inherited pipe EOF through close before its TERM record is checked. Where no reaper handle is owned, one deadline-free, signal-bound foreign-PID check remains per file. ci-git-owner-settlement's native Windows Job cases keep main's bytes until a Windows host can prove them. No product source, test timeout, product timeout, or assertion meaning changed.
Part of the polling audit from #162274. Proof on Blacksmith Testbox: delay probes fail the original bytes and pass these bytes, forced-abort probes reach cleanup, 20/20 standalone runs per touched file, 3/3 replays of each owning shard (core-tooling-12 on the rebased head), type-aware lint, base-aware timeout-race ratchet; Codex autoreview clean; ClawSweeper found no findings.
