---
title: 'In the News: November 18, 2026 (Extra 3)'
description: 'Claude Code 2.1.328 starts some cloud-provider sessions in auto mode and stops managed PostToolUse hooks firing on refused calls.'
canonical_url: 'https://darkfactory.dev/news/2026-11-18-extra-3'
markdown_url: 'https://darkfactory.dev/news/2026-11-18-extra-3.md'
collection: news
date_published: '2026-11-18T23:05:00-05:00'
date_modified: '2026-11-18T23:05:00-05:00'
---

# In the News: November 18, 2026 (Extra 3)


Claude Code 2.1.328, released on the evening of November 18, changes two defaults for teams that run it through a cloud provider or under managed settings. Sessions on Opus 4.6 or Sonnet 4.6 through Amazon Bedrock, Google Cloud's Agent Platform, Microsoft Foundry or the Claude apps gateway now start in auto mode when no permission mode is set. Managed PostToolUse hooks no longer fire for tool calls that were refused before they ran.

## 1. Claude Code 2.1.328 brings auto mode to older models on cloud providers and changes what managed hooks see

**[Claude Code v2.1.328](https://github.com/anthropics/claude-code/releases/tag/v2.1.328)** · Anthropic · GitHub release notes, November 18, 2026

The first line of the notes adds auto mode for Opus 4.6 and Sonnet 4.6 on Bedrock, Agent Platform, Foundry and the Claude apps gateway, and says "interactive sessions with no permission mode start in it". A later line changes plan mode and Bash prompts for those models on third-party providers "to use and offer auto mode, as on newer models". The notes name `disableAutoMode` as the setting that turns both of those off.

Two hook changes affect anyone who logs or filters agent actions through managed settings. Managed PostToolUse hooks now "no longer fire for calls refused before they ran: no log entry, rewrite, block or stop there". The notes say PostToolBatch receives most of those calls, "not marked as refused". Code Mode also stops running UserPromptSubmit hooks, managed ones included, on scripts' output; the notes point to a PostToolUse hook with `updatedToolOutput` for redacting that output instead. In the other direction, `mcp_tool` hooks gain `"onFailure": "block"`, which command and HTTP hooks already had, so a hook whose call fails now blocks the action.

The built-in review command is narrower. On most models, /code-review now reports "only bugs and CLAUDE.md violations", marks each finding Important, Pre-existing or Nit, and ends each summary with a suggested fix. Reuse, simplification and efficiency cleanups move to /simplify.

We read the full notes. We did not install the release or test any of these behaviors, and the notes give no figures on how many sessions are affected.

**Why it matters:** Teams that run Opus 4.6 or Sonnet 4.6 through a cloud provider and leave the permission mode unset now get auto mode, which this release adds for those models there. If you want a person approving each action in those sessions, set the permission mode explicitly rather than relying on the default. If your audit trail comes from managed PostToolUse hooks, calls refused before they ran no longer reach them, so check that your logging also reads PostToolBatch.
