A quick reference for how to clone only a specific directory from a Git repository.
Introduction
I needed to clone only a specific directory from a large repository to my local machine.
It wasn't immediately obvious how to do this, so I'm leaving a note here for future reference.
Note: This article was translated from my original post.
How to Clone Only a Specific Directory with Git
Commands
You can clone only a specific directory using the following commands:
git clone --filter=blob:none --sparse https://github.com/user/repo.git cd repo git sparse-checkout set path/to/directory
- Line 1:
git clone --filter=blob:none --sparse https://...
This first line uses --filter=blob:none to start without downloading any actual file content, and --sparse to enable sparse-checkout.
Sparse-checkout is a Git feature that limits the working tree (the part that actually gets checked out) to only a subset of the repository ("sparse" means "thinly scattered"). At this point, only the files directly under the root directory are checked out.
- Line 2:
cd repo
This simply moves into the cloned repository directory (which at this stage only contains files from the root level).
- Line 3:
git sparse-checkout set path/to/directory
The set command specifies the directory path to check out, and sparse-checkout ensures only that directory gets checked out.
With these Git commands, you can clone only a specific directory from a repository.
Let's walk through an actual example.
Example
We'll clone only the modules/cloud-run-v2 directory from Google Cloud's Terraform modules collection.
First, run the clone command pointing to the repository root, with the --filter=blob:none --sparse flags:
git clone --filter=blob:none --sparse https://github.com/GoogleCloudPlatform/cloud-foundation-fabric.git
Next, move into the repository directory:
cd cloud-foundation-fabric
If we check the files at this point, we can see that only the root-level files have been downloaded:
$ ls -la total 1632 drwxr-xr-x 23 bioerrorlog staff 736 2 23 22:07 . drwxr-xr-x 93 bioerrorlog staff 2976 2 23 22:07 .. -rw-r--r-- 1 bioerrorlog staff 53 2 23 22:07 .codespellrc drwxr-xr-x 13 bioerrorlog staff 416 2 23 22:07 .git -rw-r--r-- 1 bioerrorlog staff 85 2 23 22:07 .gitattributes -rw-r--r-- 1 bioerrorlog staff 556 2 23 22:07 .gitignore -rw-r--r-- 1 bioerrorlog staff 4090 2 23 22:07 .pre-commit-config.yaml -rw-r--r-- 1 bioerrorlog staff 78 2 23 22:07 .style.yapf -rw-r--r-- 1 bioerrorlog staff 182 2 23 22:07 .tflint.hcl -rw-r--r-- 1 bioerrorlog staff 560 2 23 22:07 .yamllint lrwxr-xr-x 1 bioerrorlog staff 9 2 23 22:07 AGENTS.md -> GEMINI.md -rw-r--r-- 1 bioerrorlog staff 311444 2 23 22:07 CHANGELOG.0.md -rw-r--r-- 1 bioerrorlog staff 340612 2 23 22:07 CHANGELOG.md -rw-r--r-- 1 bioerrorlog staff 63422 2 23 22:07 CONTRIBUTING.md -rw-r--r-- 1 bioerrorlog staff 10981 2 23 22:07 CURSED_KNOWLEDGE.md -rw-r--r-- 1 bioerrorlog staff 7815 2 23 22:07 FABRIC-AND-CFT.md -rw-r--r-- 1 bioerrorlog staff 3533 2 23 22:07 GEMINI.md -rw-r--r-- 1 bioerrorlog staff 11357 2 23 22:07 LICENSE -rw-r--r-- 1 bioerrorlog staff 6487 2 23 22:07 README.md -rw-r--r-- 1 bioerrorlog staff 3568 2 23 22:07 REFERENCES.md -rw-r--r-- 1 bioerrorlog staff 1110 2 23 22:07 default-versions.tf -rw-r--r-- 1 bioerrorlog staff 1114 2 23 22:07 default-versions.tofu -rw-r--r-- 1 bioerrorlog staff 24436 2 23 22:07 diagram.svg
Finally, specify the target directory path modules/cloud-run-v2 and run the sparse-checkout:
git sparse-checkout set modules/cloud-run-v2
This downloads the specified directory.
Let's verify:
$ ll total 1632 drwxr-xr-x 24 bioerrorlog staff 768 2 23 22:13 ./ drwxr-xr-x 93 bioerrorlog staff 2976 2 23 22:07 ../ -rw-r--r-- 1 bioerrorlog staff 53 2 23 22:07 .codespellrc drwxr-xr-x 13 bioerrorlog staff 416 2 23 22:13 .git/ -rw-r--r-- 1 bioerrorlog staff 85 2 23 22:07 .gitattributes -rw-r--r-- 1 bioerrorlog staff 556 2 23 22:07 .gitignore -rw-r--r-- 1 bioerrorlog staff 4090 2 23 22:07 .pre-commit-config.yaml -rw-r--r-- 1 bioerrorlog staff 78 2 23 22:07 .style.yapf -rw-r--r-- 1 bioerrorlog staff 182 2 23 22:07 .tflint.hcl -rw-r--r-- 1 bioerrorlog staff 560 2 23 22:07 .yamllint lrwxr-xr-x 1 bioerrorlog staff 9 2 23 22:07 AGENTS.md@ -> GEMINI.md -rw-r--r-- 1 bioerrorlog staff 311444 2 23 22:07 CHANGELOG.0.md -rw-r--r-- 1 bioerrorlog staff 340612 2 23 22:07 CHANGELOG.md -rw-r--r-- 1 bioerrorlog staff 63422 2 23 22:07 CONTRIBUTING.md -rw-r--r-- 1 bioerrorlog staff 10981 2 23 22:07 CURSED_KNOWLEDGE.md -rw-r--r-- 1 bioerrorlog staff 7815 2 23 22:07 FABRIC-AND-CFT.md -rw-r--r-- 1 bioerrorlog staff 3533 2 23 22:07 GEMINI.md -rw-r--r-- 1 bioerrorlog staff 11357 2 23 22:07 LICENSE -rw-r--r-- 1 bioerrorlog staff 6487 2 23 22:07 README.md -rw-r--r-- 1 bioerrorlog staff 3568 2 23 22:07 REFERENCES.md -rw-r--r-- 1 bioerrorlog staff 1110 2 23 22:07 default-versions.tf -rw-r--r-- 1 bioerrorlog staff 1114 2 23 22:07 default-versions.tofu -rw-r--r-- 1 bioerrorlog staff 24436 2 23 22:07 diagram.svg drwxr-xr-x 4 bioerrorlog staff 128 2 23 22:13 modules/ # The modules directory has been downloaded $ ls modules/ README.md cloud-run-v2 # Inside the modules directory, only the specified cloud-run-v2 (plus files directly under the directory) has been downloaded
Conclusion
That was a quick reference on how to clone only a specific directory with Git.
That's all!
[Related Articles]
