---
type: "post"
id: "claude-code-v2-1-328-adds-auto-mode-8c6ac1"
slug: "claude-code-v2-1-328-adds-auto-mode-8c6ac1"
title: "Claude Code v2.1.328 adds auto mode on Bedrock, Foundry and Google Cloud and a stricter /code-review"
description: "Anthropic’s Claude Code v2.1.328 adds auto mode for Claude Opus 4.6 and Claude Sonnet 4.6 across Amazon Bedrock, Google Cloud’s Agent Platform, Microsoft Foundry, and the Claude apps gateway."
retrieval_nugget: "Anthropic’s Claude Code v2.1.328 adds auto mode for Claude Opus 4.6 and Claude Sonnet 4.6 across Amazon Bedrock, Google Cloud’s Agent Platform, Microsoft Foundry, and the Claude apps gateway. It also narrows /code-review toward bugs and CLAUDE.md violations, making permission handling and review output more explicit for teams using the coding agent."
published_at: "2026-11-21"
updated_at: "2026-11-21"
record_date: "2026-11-21"
date_kind: "published_at"
topics: ["claude","claude-code","anthropic"]
entities: ["Anthropic","Claude Code","Claude Opus 4.6","Claude Sonnet 4.6","Amazon Bedrock","Google Cloud Agent Platform","Microsoft Foundry"]
related_patterns: ["verification-bandwidth-is-the-scarce-resource"]
novelty_label: "new_fact"
what_changed: null
prior_refs: []
source_url: "https://github.com/anthropics/claude-code/releases/tag/v2.1.328"
source_urls: ["https://github.com/anthropics/claude-code/releases/tag/v2.1.328"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:claude-code-v2-1-328-adds-auto-mode-8c6ac1"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/claude-code-v2-1-328-adds-auto-mode-8c6ac1/","markdown":"https://newruntime.com/posts/claude-code-v2-1-328-adds-auto-mode-8c6ac1.md","json":"https://newruntime.com/posts/claude-code-v2-1-328-adds-auto-mode-8c6ac1.json"}
markdown_url: "https://newruntime.com/posts/claude-code-v2-1-328-adds-auto-mode-8c6ac1.md"
---

# Claude Code v2.1.328 adds auto mode on Bedrock, Foundry and Google Cloud and a stricter /code-review

## Retrieval answer

Anthropic’s Claude Code v2.1.328 adds auto mode for Claude Opus 4.6 and Claude Sonnet 4.6 across Amazon Bedrock, Google Cloud’s Agent Platform, Microsoft Foundry, and the Claude apps gateway. It also narrows /code-review toward bugs and CLAUDE.md violations, making permission handling and review output more explicit for teams using the coding agent.

## Summary
Anthropic’s Claude Code v2.1.328 adds auto mode for Claude Opus 4.6 and Claude Sonnet 4.6 across Amazon Bedrock, Google Cloud’s Agent Platform, Microsoft Foundry, and the Claude apps gateway. It also narrows `/code-review` toward bugs and `CLAUDE.md` violations, making permission handling and review output more explicit for teams using the coding agent.

## What happened
Anthropic’s release notes for Claude Code v2.1.328 say auto mode is now supported for Claude Opus 4.6 and Claude Sonnet 4.6 on Amazon Bedrock, Google Cloud’s Agent Platform, Microsoft Foundry, and the Claude apps gateway. Interactive sessions started without a permission mode now begin in auto mode. For plan mode and Bash prompts on those third-party providers, the release changes prompts to use and offer auto mode as they do on newer models. Anthropic says administrators can turn both off with `disableAutoMode`.

The release also changes `/code-review` on most models. It now reports only bugs and violations of `CLAUDE.md`, rather than combining those with broader cleanup suggestions. Findings receive one of three labels: Important, Pre-existing, or Nit. Each summary ends with a suggested fix, and a clean review now says, “Code review found no issues.” Anthropic directs users to `/simplify` for reuse, simplification, and efficiency cleanups. The change separates a check for defects and project-rule violations from a separate pass for code-quality improvements.

A new hook option gives teams a way to block an action when an MCP tool hook fails: `onFailure: "block"`. Anthropic says this behavior is now available for `mcp_tool` hooks, as it already was for command and HTTP hooks. The option makes the failure response configurable at the hook level; the release notes do not describe a broader change to MCP permissions.

Claude Code’s Code Mode has several workflow changes. Anthropic says it removes a 0.75-second pause after most scripts in auto and bypass-permissions modes. It now shows the command a script is running and its output, tells Claude which tool calls were still running when a script failed, and displays development servers as background tasks in `/tasks`. The release notes also say Code Mode uses shorter tool descriptions to reduce token use and that Claude is told when MCP servers are still connecting or their tool list changes.

The version includes self-hosted runner changes, including a `--repository-folder` option for naming a repository folder at capacity one, session-label and metadata options for spawn-runner hooks, and queue-wait metrics that distinguish reopened sessions from new ones. The release entry also lists fixes for session resumption, hooks, Remote Control, Code Mode, terminal behavior, and provider compatibility. It gives no price change or rollout schedule.

## What the archive adds
New Runtime’s earlier coverage tracked Claude Code v2.1.310’s Monitor tool parity across Bedrock, Vertex, and Foundry, alongside gateway retry and Enterprise role controls ([Claude Code v2.1.310](https://newruntime.com/posts/claude-code-v2-1-310-brings-the-monitor-24f7bd/), 2026-10-24). A separate note covered v2.1.318’s release changes ([Claude Code v2.1.318](https://newruntime.com/posts/release-v2-1-318-anthropics-claude-code-77b52d/), 2026-11-04). This is a new release in an ongoing product story: v2.1.328 adds provider auto mode and changes review behavior.

## Trend signal
This release supports [verification bandwidth is the scarce engineering resource](https://newruntime.com/patterns/verification-bandwidth-is-the-scarce-resource/). Anthropic’s decision to make `/code-review` report bugs and project-rule violations, label findings by importance or prior existence, and suggest fixes is a product-level attempt to make review output more actionable. The release does not establish whether the changes improve review accuracy or reduce the time engineers spend validating findings.

## Why it matters
For teams running Claude Code through a cloud provider, auto mode’s appearance across Bedrock, Google Cloud’s Agent Platform, and Foundry reduces a mismatch between the coding agent and provider-specific environments. But availability of auto mode does not establish that a team’s permissions and approval practices are appropriate for it. The setting deserves review alongside existing policy controls, particularly because interactive sessions without an explicit permission mode now start in auto mode.

The `/code-review` change creates a useful workflow distinction: use review for defects and documented project rules, then invoke `/simplify` when a cleanup pass is wanted. That can help teams avoid treating every stylistic suggestion as a defect, while the labels and suggested fixes give reviewers more structured material to check. The MCP hook option is similarly relevant to reliability: a team can block an action when a tool hook fails rather than proceed as if the hook succeeded. These are configurable controls, not proof that agent output is correct or safe by default.

## Limits and uncertainty
Anthropic’s release notes do not quantify the accuracy of the revised `/code-review`, how often its labels match human reviewers, or whether suggested fixes reduce review time. They do not specify the exact behavior or scope of auto mode permissions across the named providers, nor recommend configurations for different risk levels. The release describes Code Mode speed and token-use improvements without benchmark methodology or measurements beyond removal of the stated 0.75-second pause. The listed fixes have not been independently assessed here in customer environments.

## Skill to build now
Take one low-risk repository and compare the two review passes. Run `/code-review` against a small change that touches a documented `CLAUDE.md` rule, then run `/simplify` separately. Check whether the first pass surfaces only defects and rule violations, and whether the second produces useful cleanup suggestions. Record which findings a human accepts or rejects before making either pass part of a wider workflow.

## Reader takeaway
Claude Code v2.1.328 makes auto mode available across more cloud-provider setups and gives `/code-review` a narrower, more structured job. Treat both as workflow controls to evaluate, not evidence that permissions, findings, or suggested fixes are correct. Separate bug-and-policy review from optional cleanup, then validate each on a small change before wider adoption.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Claude](https://newruntime.com/topics/claude.md): Explore the claude topic hub.
- topic hub: [Claude Code](https://newruntime.com/topics/claude-code.md): Explore the claude-code topic hub.
- topic hub: [Anthropic](https://newruntime.com/topics/anthropic.md): Explore the anthropic topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)
