Diagnostics from the hung run show the stuck podman exec is the ONLY process with the libpod lock segment mapped: the SHM container-lock semaphore was leaked by a process that died holding it, and SHM semaphores never auto-release. Switch libpod to lock_type=file (flock releases on holder death) before any podman use. Also raise the wrapper's hang guard back to 900s for cold-cache lint runs. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NaDYk2PGWfSLEXeN5CBJFN