git clone Like a Pro: From -b develop to Partial & Sparse Clones (Basic Expert)

This content originally appeared on DEV Community and was authored by Cristian Sifuentes

git clone Like a Pro: From -b develop to Partial & Sparse Clones (Basic → Expert)

You don’t just clone a repo—you curate what you bring home: branches, history depth, files, and even large assets. Here’s your tiered playbook.

BASIC — “Just grab what I need”

Clone default branch

git clone https://github.com/owner/repo.git

By default Git creates remote-tracking branches for all remote branches and checks out an initial branch matching the remote’s active branch.

Clone and check out a specific branch (e.g., develop)

git clone -b develop https://github.com/owner/repo.git
# Tip: Add --single-branch to fetch only that branch’s history
git clone -b develop --single-branch https://github.com/owner/repo.git

--single-branch limits the clone to the chosen branch’s history (or the remote’s HEAD if -b is omitted).

Fast start (shallow)

git clone --depth 1 -b develop --single-branch https://github.com/owner/repo.git

--depth N fetches only the most recent N commits; shallow clones also only fetch history for the branch you ask for.

MEDIUM — “Make it tidy & predictable”

Choose HTTPS vs SSH

# HTTPS (easy to start)
git clone https://github.com/owner/repo.git
# SSH (great for contributors)
git clone [email protected]:owner/repo.git

Name the remote differently or set configs at clone time

git clone -o upstream https://github.com/owner/repo.git
git clone -c core.autocrlf=input https://github.com/owner/repo.git

Use -o to rename origin, and -c key=value to seed config on the new repo.

Skip tags (lighter fetch)

git clone --no-tags --single-branch -b develop https://github.com/owner/repo.git

--no-tags avoids fetching tag refs on clone. You can fetch tags later with git fetch --tags.

GitHub CLI pass-through

gh repo clone owner/repo -- -b develop --depth 1 --single-branch

Everything after -- is forwarded to git clone.

ADVANCED — “Big repos, small footprint”

Partial clone (blobless)

git clone --filter=blob:none https://github.com/owner/repo.git
# Combine for speed:
git clone --filter=blob:none --depth 1 -b develop --single-branch https://github.com/owner/repo.git

Partial clone avoids downloading file contents (blobs) up front; blobs are fetched on demand during checkout. Great for huge codebases.

Sparse checkout (mono-repo surgery)

git clone --filter=blob:none https://github.com/owner/repo.git
cd repo
git sparse-checkout init --cone
git sparse-checkout set apps/web libs/ui

Sparse checkout materializes only selected directories in your working tree—perfect for mono-repos and CI.

Submodules, the right way

git clone --recurse-submodules https://github.com/owner/repo.git
# (Optional) keep submodules shallow too:
git clone --recurse-submodules --shallow-submodules https://github.com/owner/repo.git

--recurse-submodules initializes & updates submodules during clone; --shallow-submodules keeps them light.

Git LFS: skip heavy assets at clone time

# one-off
GIT_LFS_SKIP_SMUDGE=1 git clone https://github.com/owner/huge-lfs-repo.git
# or set globally
git lfs install --skip-smudge

Skips automatic LFS download during clone; fetch assets later when needed with git lfs pull.

EXPERT — “Admin, mirrors, provenance & power moves”

Mirror vs bare (for backups & CI mirrors)

# Exact mirror of all refs (includes notes, refs, etc.)
git clone --mirror https://github.com/owner/repo.git /backups/repo.git
# A bare working repo (no working tree)
git clone --bare https://github.com/owner/repo.git

--mirror implies --bare and tracks all refs for 1:1 replication. Use git remote update to refresh.

Separate .git directory (ephemeral working trees)

git clone --separate-git-dir=/var/git/repo.git https://github.com/owner/repo.git worktree

Keeps the object database outside your working directory—handy for build agents and scratch trees.

Reproducible, minimal checkouts (combine techniques)

git clone --filter=blob:none --single-branch -b release/2025.09   --depth 20 --no-tags https://github.com/owner/big-mono.git
cd big-mono
git sparse-checkout init --cone
git sparse-checkout set services/api

You now have a tiny, fast, branch-specific, directory-focused working copy.

Cheat-Sheet: Pick your recipe

Common pitfalls (and quick fixes)

• Cloned with -b develop but fetch pulled all branches? Add --single-branch to limit history to that branch on the initial clone.
• Shallow clone breaks tools like git describe or deep merges? Deepen when needed: git fetch --deepen 200 or unshallow with a full fetch: git fetch --unshallow.
• Partial clone feels “missing files”? That’s expected; blobs download on demand at checkout. Consider --filter=blob:limit=<bytes> when some large blobs are needed.
• Sparse checkout not reducing clone time? Sparse limits the working tree; combine with --filter=blob:none and/or shallow depth to shrink transfer.
• Submodules out of sync? Use --recurse-submodules at clone, or run git submodule update --init --recursive.

Mental model: Why so many knobs?

Think of git clone as a pipeline:
1) Discover refs (branches/tags) →

2) Decide history (--single-branch, --depth) →

3) Decide objects (--filter partial clone) →

4) Decide files on disk (sparse checkout) →

5) Extras (submodules, LFS, mirror/bare, configs).

Dial each stage only as much as your use‑case needs.

Final Thoughts

• For everyday work on one branch: -b <name> --single-branch --depth 1 is a great default.
• For massive repos: add --filter=blob:none and sparse-checkout your sub-tree.
• For CI mirrors/backups: --mirror is your friend.
• For GitHub CLI users: remember the -- pass‑through.

Clone less. Build more.

This content originally appeared on DEV Community and was authored by Cristian Sifuentes