crab fetch
Pre-fetch Crab objects for the current repository into the local cache. This warms data needed by later hydrate, checkout, or push operations without materializing files.
Synopsis
crab fetch [OPTIONS]The command reads the canonical repository URL from crab.toml. For a
managed URL it resolves the active exact-authority profile, refreshes
authentication when needed, obtains a bounded read grant, and uses the same
verified cache/reconstruction path as other reads.
crab fetch does not negotiate Git refs. Run git fetch origin first when you
also need remote Git ref and pack updates.
Options
| Option | Default | Description |
|---|---|---|
--include <PATTERN> | None | Include matching file/object patterns; repeat as needed |
--exclude <PATTERN> | None | Exclude matching patterns; repeat as needed |
--all | Disabled | Fetch objects for all refs rather than only HEAD |
--dry-run | Disabled | Resolve and report selected files without credentials, network access, or cache writes |
--no-sync-chunk-index | Disabled | Skip post-fetch local chunk-index warming |
--json | Disabled | Emit one terminal JSON envelope |
--jsonl | Disabled | Stream JSONL progress and a terminal result |
Examples
Fetch current managed repository data
cd models
crab fetchUpdate refs and pre-warm all refs
git fetch origin
crab fetch --allPre-warm selected formats
crab fetch --include '*.safetensors' --exclude 'archive/**'Inspect without downloading
crab fetch --dry-run --jsonFetch uses Git pointer blobs as the live inventory. It batch-checks blob sizes,
deduplicates identical file hashes across refs, then reconstructs selected files
into a discard sink through the same replica, file-index, shard, and xet-core
range-cache path used by hydrate. Reconstructed bytes are hash- and size-checked;
peak memory does not scale with logical file size. --all means every local ref,
so run git fetch origin first when remote refs must be current.