I've found agents like Claude are absolute ass at fixing any halfway complex merge conflict. I won't trust them to do it.
You trust Claude (or any AI) to do anything right? The odds are good - but never 100%.
That said, "jj new -A @", get Claude to do it's thing, "jj squash" is going to be pretty safe.
You trust Claude (or any AI) to do anything right? The odds are good - but never 100%.
That said, "jj new -A @", get Claude to do it's thing, "jj squash" is going to be pretty safe.