It is a series of auto-commits that took place at a certain interval. Is there any command or algorithm to simply delete commit A without needing to adjust the following commits ?
In fact, B already contains all of A’s changes, and subsequent commits to other files are not affected by its loss.

• Each commit contains changes to only one file.
• Commits marked with a character contain changes in one file,
• commits marked with numbers indicate changes in another file.

For comparison, I have first given an example with a normal commit –amend.

             master
             ↓
 1 --- 2 --- A 

$ git commit --amend → B

                   master
                   ↓
 1 --- 2 --- A     B
                 /
         _______/

             master
             ↓
 1 --- A --- 2 

$ ???

                   master
                   ↓
 1 --- A     2 --- B
           /
   _______/

Using git rebase creates an unnecessary merge conflict. Interesting non-interactive options.