picurv storage is now supported, with its compression levels, offload policies, and the checkpoints, logs, analysis, visualization, and inputs retention components. A campaign against a Google Drive remote on 3.5 GiB turbulent-channel runs restored every file bit for bit, left exactly each policy's documented components, restarted from an offloaded-then-restored checkpoint, and measured each compression level (fast 70% of the raw size, maximum 65% at 2.3 times the time and 3.8 GB of memory). raw-output stays experimental. Page 61's compression guidance now reflects these measurements.
A workspace restored to another directory no longer rewrites paths inside its asset objects, which left them failing their own checksums; checkpoint bundles are likewise left untouched, and rewritten files no longer change hard-linked copies.
storage protect or offload with an explicit --compression no longer reuses an existing archive made at another level; without --compression, protect-then-offload still reuses it.
storage verify --workspace verifies a workspace's own archive.
storage prune reports an asset object it could not delete and exits non-zero, instead of counting it as removed.
Restoring into a run's own directory skips components it already holds.