What probably matters to youSection of the release
What
Three new hidden claude subcommands were added: edit-skill-overrides, edit-sandbox-settings, and edit-chrome-settings, each reading one JSON edit from stdin.
They dispatch to new validated settings-write functions: one for per-skill invocation state (on, name-only, user-invocable-only, off, plus handles like unqualified name, aliases, and a bare-name-reserved flag), one for sandbox mode (auto-allow, regular, disabled, plus allowUnsandboxedCommands/excludeCommand), and one for the Claude-in-Chrome default-enabled toggle (claudeInChromeDefaultEnabled).
Each write function validates its input with zod, checks for policy/plugin/author locks, and writes to local settings.
Why
This gives the VS Code extension (and other callers) a supported, validated way to change skill, sandbox, and Claude-in-Chrome settings from the command line instead of writing to settings files directly.
Read from
Names in the bundleedit-permission-rulesedit-memory-settings
What the documentation says
--helpCLI referencemodified, high confidence| `--permission-mode` | Begin in a specified [permission mode](/docs/en/permission-modes). Accepts `default`, `acceptEdits`, `plan`, `auto`, `dontAsk`, `bypassPermissions`, or `manual` as an alias for `default`. The `manual` alias selects …see the edit
How sure we are
One source agreesOne thing we can check says the same as this entry.
Anthropic's documentation agrees--help on CLI reference
