Remove Sensitive Data from a GitHub Repository

Oh no! Did you commit and push something sensitive to a public GitHub repository (e.g. secrets, credentials, student information, or solutions)? This article is for you.

Even if a repository is private, you still should NOT store student information on GitHub. If you have a script that interacts with student information, you should track that script via git/GitHub, but you should not track any associated CSV or other files that contain student information. It is useful to keep track of the format of the data, and you may do so by describing the data in your script or by tracking a data file that does not contain any student information. For example, you may track a CSV file that only contains the headers (example). If you need to share or otherwise preserve data files containing student information, we recommend using Google Drive which has been approved for P3 data.

Regardless of a repository’s visibility, it is not enough to delete files containing student information; You must also delete the git history for those files by following the steps below. If your repository has had any pull requests with the student information, you must additionally follow these steps after completing the below.

Be careful running the following commands. These are DESTRUCTIVE actions. 🌋

Option 1 (Recommended): Use git-filter-repo

Follow this official guide from the GitHub documentation.

Option 2: Remove all commits up to a certain point

1. Find the last “good” commit hash by running:

    git log
   

   The output should have entries that look something like

    commit a3cf8b9261a52a8c34098e5316278ce52e40a9b5
    Author: Cal-CS-61A-Staff <cs61a@berkeley.edu>
    Date:   Wed Mar 26 20:41:07 2025 +0000
   

2. Take the commit hash (above a3cf8b9261a52a8c34098e5316278ce52e40a9b5) and run the following locally:

    git reset --hard <hash>
   

   If you’d like to remove only the latest commit you can run the following instead:

    git reset --hard HEAD~1
   

3. Reflect your changes on the remote (public GitHub repo):

    git push origin HEAD --force
   

Option 3: Remove only specific commits

If you have a mix of “good” and “bad” commits, you’ll want to interactively rebase. Read more about rebasing.

1. Identify the last commit hash after all all of the commits that you want to remove. The following commands may be useful for this:

    git log
    git reflog
   

2. Start the interactive rebase using that hash:

    git rebase -i <hash>
   

3. Identify the commits that you want to remove and remove them in the edit screen.

4. Resolve any conflicts.

5. In the edit screen, save and exit.

6. End the rebase by running:

    git rebase --continue
   

   If you messed up the rebase, here’s your options:

   • Abort the rebase without resetting HEAD to the original branch (“clean up” the rebase):

       git rebase --quit
     

   • Abort the rebase and reset HEAD to the original branch (“undo” the rebase):

       git rebase --abort
     

   • Continue with the rebase, but skip any commits that introduce a merge conflict (any commits with merge conflicts will not be in the git history after finishing the rebase):

       git rebase --skip
     

7. Reflect your changes on the remote (public GitHub repo):

    git push origin HEAD --force