You'll noticeTier: how much it should matter to you
2Useful: my rating, 1 to 5
3Signal: worth watching, 1 to 5
Auto ModeArea: what it touches
ImprovementsKind: in v2.1.283,
ImprovementsSection of the release
Unclear It is not clear what the classifier then does with these edits.
What
In auto mode, Claude Code uses a classifier, which is a model that judges actions, to decide how they are handled. In acceptEdits mode, file edits are approved automatically. When an edit is allowed that way, Claude Code can now also pass that approval through auto mode's classifier. This only happens if the session's model qualifies. A model qualifies when either of these is true:
CLAUDE_CODE_AUTO_MODE_CLASSIFY_EDITS is set to true
the model is on a list of models sent by the server
The environment variable already existed, but it was not used for these edits before. The server's list is empty when nothing is sent.
Why
More automatically approved edits can now go through auto mode's classifier, which can change how edits are approved in auto mode. If you do not set the variable and your model is not on the server's list, nothing changes.
The entry above is what we published on the day. These lines were added later, as Anthropic's own pages caught up, and they sit beside the original rather than replacing it.
Added sinceA small documentation edit on Hooks reference touched a line naming acceptEdits after this was published.| `setMode` | `mode`, `destination` | Changes the permission mode. Valid modes are `default`, `auto`, `acceptEdits`, `dontAsk`, `bypassPermissions`, `plan`, and `manual` as an alias for `default` |hookssee the edit
Confirmed sinceAnthropic's documentation has since written up acceptEdits, on Choose a permission mode.In addition to file edits, `acceptEdits` mode auto-approves common filesystem Bash commands: `mkdir`, `touch`, `rm`, `rmdir`, `mv`, `cp`, and `sed`. These commands are also auto-approved when prefixed with safe environment variables such a…permission-modessee the edit
How sure we are
Something disagreesSomething we can check disagrees with this entry, or the writer said they could not settle it.
The writer flagged doubtIt is not clear what the classifier then does with these edits.
Anthropic's documentation agreesAnthropic's documentation has since written up acceptEdits, on Choose a permission mode.
