[DOCS] Auto mode docs omit classifier handling for risky shell-command checks

Documentation Type

Missing documentation (feature not documented)

Documentation Location

https://code.claude.com/docs/en/permission-modes

Section/Topic

The auto-mode safety rules and the explanation of when a command is evaluated by the classifier rather than shown as a normal permission dialog.

Current Documentation

The page describes auto mode in the mode table as:

Everything, with background safety checks

It also lists recursive forced deletes such as rm -rf "$VAR" as blocked by default, but does not describe the changed handling of dangerous-rm checks, background-& commands, or suspicious Windows paths.

What's Wrong or Missing?

Claude Code 2.1.218 changed the dangerous-rm, background-&, and suspicious-Windows-path checks so they no longer open permission dialogs. The auto-mode classifier now adjudicates those cases.

The current page does not tell users which layer evaluates these commands or what they should expect to see in the transcript. This makes it difficult to distinguish a classifier decision from a normal permission request and leaves users without an explanation for why a command that previously opened a dialog is now allowed or blocked through auto mode.

Suggested Improvement

Add a versioned subsection to the auto-mode safety rules explaining that, as of v2.1.218, the dangerous-rm, background-&, and suspicious-Windows-path checks are handled by the auto-mode classifier rather than by a separate permission dialog.

For each category, describe whether the classifier can allow or block the action, and show where the decision reason appears. Make clear that explicit permissions.deny and permissions.ask rules remain a separate layer when applicable.

Impact

Medium - Makes feature difficult to understand

Additional Context

Affected Pages:

| Page | Context |
|------|---------|
| https://code.claude.com/docs/en/permission-modes | Auto-mode safety rules and prompt behavior |
| https://code.claude.com/docs/en/auto-mode-config | Classifier ordering and configuration boundaries |
| https://code.claude.com/docs/en/permissions | Relationship between explicit permission rules and classifier decisions |

Total scope: 3 pages should describe the same safety-check routing change.

The behavior was changed in Claude Code v2.1.218.