B-6AHH-baseline-capture-concurrent-lock
Status: promoted/task
Two concurrent /sdlc:task-work runs that both run Step 3a ‘quality baseline capture’ against the same origin/main SHA (writing the same —baseline-dir) hung for 25+ min during T-KESH. Capture has no lock/timeout. Add a per-SHA capture lock (or an ‘already capturing for SHA’ short-circuit + bounded timeout) so parallel sessions don’t deadlock the full gate. Surfaced by T-KESH-config-surface-sdlc-yaml-zod-safeparse post-mortem.