Fork It, Keep It Private, Stay in Sync with Upstream

Ever found an open source project that does most of what you need, but not quite all of it? I run into this all the time. You want to add your own features, keep them private, and still pull in upstream bug fixes without losing your mind.

The standard fork workflow on GitHub works fine for public contributions. But what if you're building something proprietary on top of someone else's open source work? You need a different approach.

Here's the setup I've landed on after burning through a few different strategies. It's not fancy. But it works.

The Setup

First, clone the original project. Then swap the remotes so your private repo becomes origin and the original becomes upstream.

git clone https://github.com/original-author/some-project
cd some-project
git remote set-url origin https://github.com/you/your-private-repo
git remote add upstream https://github.com/original-author/some-project
git push origin main

Why do it this way? Two reasons. First, origin is git's default remote. When you type git push without thinking, it goes to your private repo. That's safe. Second, keeping the original as upstream makes it obvious where updates come from.

I've seen people skip this and just push directly to their fork's remote. That works too. But I like being explicit about which remote is which. It saves me from pushing something to the wrong place at 11pm.

The Sync Loop

Here's where the real workflow lives. The original project keeps moving. You need to pull those changes in without breaking your custom work.

git fetch upstream
git checkout main
git rebase upstream/main
git push origin main --force-with-lease

Four commands. That's it. Let me walk through why each one matters.

Fetch, Don't Pull

git fetch upstream downloads the latest commits from the original project. It doesn't merge them into your branch. That's deliberate. You want to see what's coming before you apply it.

I used to run git pull upstream main out of habit. Bad idea. Pull does a fetch and an implicit merge. You lose control over how your changes get combined with theirs.

The Rebase

git rebase upstream/main rewinds your commits, applies the upstream changes, then replays your commits on top. Your custom work stays on top. The upstream changes stay below.

Force Push With a Safety Net

Rebasing rewrites history. The commit hashes change. Git will yell at you if you try a normal push. That's where --force-with-lease comes in.

git push origin main --force-with-lease

This is safer than a plain --force. It checks that nobody else has pushed to the remote branch since your last fetch. If they have, git rejects the push. You don't overwrite anyone's work by accident.

The Merge Alternative

Some people argue for merge over rebase. They say rebasing rewrites history and that's bad. I get the argument. In a shared branch with multiple collaborators, I'd agree. Merge commits create a record of when things happened.

But for a private fork? You're the only one working on it. A clean linear history is more valuable than a timestamped record of every sync. You want to look back in six months and see your custom commits stacked neatly on top of upstream releases. Not a tangled web of merge bubbles.

If you are working with a team on the fork, use merge. Communicate with your team. Pick one strategy and stick with it. Nothing worse than half the team rebasing and the other half merging.

Handling Conflicts

Rebasing will occasionally produce conflicts. It's not fun, but it's manageable.

git rebase upstream/main
# git says: conflict in some-file.ts
# Fix the file manually
git add some-file.ts
git rebase --continue

The key insight: resolve conflicts in your code, not in theirs. The upstream changes are facts. Your custom modifications need to adapt to whatever the upstream maintainers changed.

If the conflict is too large, you have options:

• Abort and merge instead: git rebase --abort undoes everything. Try git merge upstream/main if you're pressed for time.
• Split your commit: A big commit that touches many files conflicts more often. Keep your commits small and focused.
• Reach out: If upstream changed an API you depend on, the maintainers might have migration advice.

When Not to Use This Workflow

This approach isn't right for every situation. Here's when I'd pick something else:

• You plan to contribute back: Just fork publicly on GitHub and submit PRs. No need for the private repo dance.
• The upstream changes constantly: Daily releases mean constant rebasing. Consider vendoring the dependency instead.
• Your changes are huge: If you're rewriting half the project, maintaining a fork adds friction. You might be better off building your own thing.
• The license won't let you: This is the big one.

A Quick Reference

Here's the full workflow in one block. Bookmark it.

# Initial setup
git clone https://github.com/original-author/some-project
cd some-project
git remote set-url origin https://github.com/you/your-private-repo
git remote add upstream https://github.com/original-author/some-project
git push origin main

# Each sync cycle
git fetch upstream
git checkout main
git rebase upstream/main
git push origin main --force-with-lease

Key Takeaways & Best Practices

I've been using this workflow for about two years across several projects. It's not perfect. Conflicts still happen. Large rebases can be tedious. But it beats the alternatives.

The alternatives are worse. Either you never sync upstream and your fork rots. Or you merge blindly and your history turns into a knot that nobody can read. Or you give up on privacy and open source everything.

You Might Also Like

• Creating a Modern Terminal Setup
• Advanced GraphQL API Design
• Claude API Function Calling: JSON Schema Optimization Guide
• Getting the Most Out of Claude's Free Plan (Including MCP & Claude Desktop)

Test Your Knowledge