Undoing Changes

If mistakes or unwanted changes have been committed to your Git repository, there are ways to correct them. Two common methods for reverting changes include:

• Git Reset: Resets the branch to a previous commit.
• Git Revert: Creates a new commit that reverts specified changes.

The question that decides it

Has anyone else pulled the commit you want to undo?

If not, git reset is the better tool: it moves the branch pointer and the mistake leaves no trace. If yes, git revert, because rewriting history that other people have already fetched leaves their copy and yours disagreeing, and the fix is to reset their branch and pull again.

That is the whole rule. Everything below is a variation on it.

Undo the last commit, keeping the changes

git reset --soft HEAD~1     # changes stay staged
git reset HEAD~1           # changes stay unstaged, still in your files

Use --soft when you want to recommit the same work differently, and the default mixed reset when you just want the commit to stop existing.

--hard discards the changes too, and that is unrecoverable from the working tree. It is the one to be careful with — see Undoing Changes for how to recover from that.

Undo a commit that is already shared

git revert <hash>

This records a new commit that undoes the old one. The mistake stays in the history, visibly corrected, and nobody has to do anything to their repository.

Undo changes you never committed

git restore <file>              # discard edits to a file
git restore --staged <file>     # unstage, keep the edits

Recover something that looked lost

git reflog
git switch -c recovered <hash>

The reflog records every position HEAD has held — including commits that were amended away, reset off a branch, or rebased into nothing. This is why "I lost that commit" is nearly always recoverable, and it is worth knowing the command before you need it. See Git Reflog.

Free Resources