Partial Clone
In one line
A partial clone downloads the repo's structure but omits some file contents (blobs), fetching them on demand.
In simple words
A partial clone (--filter) downloads a repository's commits, trees, and tags, but omits some file contents (blobs) based on a filter. When you later checkout or diff a file whose content is missing, Git fetches just that blob on demand. This dramatically reduces initial clone time and disk usage for large repos.
The most common filter is --filter=blob:none, which omits all file contents — you get the full history structure but no file bodies until you need them. --filter=tree:0 omits even tree objects for a more aggressive partial clone. The server must support the feature (GitHub, GitLab do).
Partial clone pairs with sparse checkout: clone without blobs, then materialize only the directories you need. This is the modern way to work with huge monorepos — you pay only for what you actually read. The trade-off is a small latency on first access to each file, as its blob is fetched.
Example
# Clone without any file contents (blobs)
git clone --filter=blob:none https://github.com/org/huge-repo.git
cd huge-repo
# The history is complete, but file contents are fetched on demand
git log --oneline # works — no blobs needed
cat README.md # triggers a fetch of README.md's blob
# Combine with sparse checkout for maximum efficiency
git sparse-checkout init --cone
git sparse-checkout set packages/web--filter=blob:none skips all file contents at clone time. Reading a file triggers an on-demand fetch of just that blob. Combined with sparse checkout, you only download what you use.
Related terms
Related Resources
Related Lessons
- Git Advanced
Learn partial clone.
Decode error messages
Plain-English explanations of common errors — what they mean, why they happen, and how to fix them.