Claude Code 2.1.328 adds auto mode for Opus 4.6 and Sonnet 4.6 on third-party providers and reworks `/code-review` output
Claude Code v2.1.328, per its release notes, is mostly fixes, with several additions and changes that users may notice.
Additions:
- Auto mode is available for Claude Opus 4.6 and Sonnet 4.6 on Amazon Bedrock, Google Cloud's Agent Platform, Microsoft Foundry and the Claude apps gateway. Interactive sessions with no permission mode start in it, and plan mode and Bash prompts on these models now offer auto mode as on newer models;
disableAutoModeturns both off. "onFailure": "block"formcp_toolhooks, as for command and HTTP hooks: a hook whose call fails blocks the action.CLAUDE_CODE_DISABLE_DEFERRED_TOOL_RESUMElets a host stopclaude -p --resumeand--continuefrom running a tool call that a hook deferred.- A warning when a message from a session that skips permission prompts may be held and dropped by a cloud session, and a notice in
/statusandclaude doctorwhen an OpenTelemetry exporter has no endpoint. - Self-hosted runner:
--repository-folder,--hook-labeland--metadata key=valueoptions, and new orchestrator metrics for queue wait and for sessions Anthropic gave up starting.
Changes that may affect existing setups:
/code-reviewon most models now reports only bugs and CLAUDE.md violations (run/simplifyfor reuse and efficiency cleanups), marks each finding Important, Pre-existing or Nit, and ends with a suggested fix.- Code Mode no longer runs
UserPromptSubmithooks on scripts' output; to redact it, use aPostToolUsehook withupdatedToolOutput. - Managed
PostToolUsehooks no longer fire for calls refused before they ran. - A shell command rewritten by a
PermissionRequesthook to a sandbox-excluded command runs outside the sandbox. /bugasks Yes or No before sending.- A sandbox that can't start now refuses the next sandboxed Bash command and says the sandbox is off for the session, unless
sandbox.failIfUnavailableis set.
Fixes cover many Code Mode problems, zsh commands failing at words that start with =, automatic compaction reusing the earlier system prompt, scheduled tasks being erased or deleted on save, MCP tools going missing on third-party providers, cloud and Remote Control session problems, terminal input and display bugs, and self-hosted runner issues. Resuming large sessions is faster, and memory use is lower when resuming sessions with many images.
Sources
- v2.1.328claude
Get news like this as a push.
AI Tracker alerts you when something changes for the tools you use.