AI Aulis

AI NEWS IN ENGLISH · 19 November 2026

The most important AI news every morning. AI writes and checks the stories against the original sources – check important facts in the sources.

38 stories · next issue tomorrow morning · How the paper is made →

← Issue 19 November 2026Paper

Opus 4.6 and Sonnet 4.6 on Bedrock: Claude Code starts in auto mode if no permission mode is set

AI-generated contentWritten by AI

github.com, huggingface.co, api.github.com, note.com, vibehacker.com, claudekit.io · 6 sources · Sources published: 19 November 2026 · Some source publication dates are unknown

Claude Code 2.1.328 brings auto mode to Opus 4.6 and Sonnet 4.6 on Amazon Bedrock, Google’s Agent Platform, Microsoft Foundry and Claude apps gateway. Interactive sessions with no permission mode set now start in auto mode.

The change affects Claude Code, Anthropic’s coding agent, which runs in the terminal, among other places. Anthropic announced the change in a changelog published on GitHub on 19 November.

In auto mode, Claude Code does not ask the user’s permission for every action. Production Ready AI, a Japanese publication, describes it as a mode in which an automatic safety check replaces the permission prompt shown to a person. Anthropic has not said which actions the check lets through.

In practice, the default changes. If an organisation uses Opus 4.6 or Sonnet 4.6 through Bedrock, for example, and has not set a permission mode, interactive sessions start in auto mode. On third-party providers, plan mode and Bash permission prompts for these models now also use and offer auto mode, the changelog says. The disableAutoMode setting turns both off. Production Ready AI wrote that the changelog does not make clear whether starting in auto mode applies only to the listed platforms.

Administrators may also need to check their hooks, scripts that run automatically in certain situations. Managed PostToolUse hooks no longer fire when a tool call is refused before it runs. The hook therefore does not log these calls and cannot rewrite, block or stop them. The PostToolBatch hook receives most of these calls, but they are not marked as refused.

In Code Mode, Claude writes a script that calls tools. UserPromptSubmit hooks, managed ones included, no longer process the output of these scripts. To redact that output, Anthropic advises using a PostToolUse hook and its updatedToolOutput field.

On most models, the /code-review command now reports only bugs and CLAUDE.md violations. Clean-up suggestions come from the /simplify command. The release contains more than 140 changes, the ClaudeKit website says. Code Mode gets more than 20 fixes or improvements, and there are many fixes for self-hosted runners, Remote Control and terminal input.

The changes are in Claude Code version 2.1.328. Anthropic has not said how urgent the update is, Production Ready AI reported.

Report an error in this story