Reference
crab clone
Clone a Git or crab:// repository. Crab remotes receive filter setup and
optional hydration after Git clone completes.
Synopsis
crab clone [OPTIONS] <URL> [DIR]Arguments
| Argument | Required | Description |
|---|---|---|
URL | Yes | Git URL/path, direct crab://bucket/prefix, canonical managed URL, or managed organization/repository shorthand |
DIR | No | Target directory; defaults to the repository name derived from the URL |
Managed shorthand is expanded only when an active service profile exists. It is a Crab CLI convenience and does not reinterpret arbitrary Git paths.
Options
| Option | Default | Description |
|---|---|---|
-b, --branch <BRANCH> | Remote default | Branch to check out |
--depth <DEPTH> | Full history | Shallow clone depth |
--lazy | Enabled | Leave managed files as pointers |
--no-lazy | Disabled | Hydrate all managed files immediately |
--eager | Disabled | Alias for --no-lazy |
--include <PATTERN> | None | Glob to hydrate immediately; repeat for multiple patterns |
--exclude <PATTERN> | None | Glob to exclude from post-clone hydration; repeat as needed |
--sync-chunk-index | Disabled | Warm the local chunk-index cache after clone |
--json | Disabled | Emit one terminal JSON envelope |
--jsonl | Disabled | Stream JSONL events followed by the terminal result |
Managed examples
Canonical hosted URL
crab login https://crab.build
crab clone crab://crab.build/acme/modelsActive-profile shorthand
crab clone acme/modelsBoth forms persist crab://crab.build/acme/models in .git/config and
crab.toml. Resolved placement and transfer credentials remain in memory.
Hydrate selected content
crab clone acme/models --include 'configs/**' --include '*.json'Hydrate everything
crab clone --no-lazy crab://crab.build/acme/modelsDirect BYOC example
crab clone crab://team-bucket/ml/modelsAn authority other than crab.build remains direct unless an exact managed
service profile binds it.