Auto Mode by Claude Code

Let Claude make permission decisions on your behalf

Auto Mode by Claude Code 介绍

Auto Mode by Claude Code is a feature that allows Claude to autonomously make permission decisions for file writes and bash commands, using an AI classifier to assess the safety of each action. Safe actions are executed automatically, while risky ones are blocked or trigger user prompts, providing a balanced approach for longer tasks with fewer interruptions and reduced risk compared to skipping permissions entirely. It is designed for use in isolated environments and is available as a research preview for Claude Team and Enterprise users.

核心功能

AI-powered classifier reviews tool calls for risks like mass file deletion or malicious code
Automatic execution of actions deemed safe without user intervention
Blocking and redirection of risky actions to maintain security
Fallback to user permission prompts for persistently blocked actions
Support for Claude Sonnet 4.6 and Opus 4.6 models in the research preview

典型使用场景

Developers automating routine coding tasks like file writes and command execution with minimal manual oversight
Teams using Claude Code for long-running projects to reduce interruptions and improve productivity
Enterprises seeking to enhance developer efficiency while maintaining controlled risk in AI-assisted workflows
Startups accelerating development cycles by delegating safe, repetitive operations to AI

为什么适合初创团队

Startups need to maximize efficiency and accelerate development to compete effectively. Auto Mode by Claude Code helps by automating safe coding tasks, reducing manual interruptions, and allowing developers to focus on innovation. This leads to faster product iterations and improved resource utilization with managed security risks.

可替代选择

GitHub Copilot's autonomous coding features, OpenAI's AI models with execution capabilities, Microsoft's AI-powered developer tools

常见问题

How does auto mode work?

Auto mode uses a classifier to review each tool call for potential risks such as mass file deletion, sensitive data exfiltration, or malicious code execution; safe actions proceed automatically, while risky ones are blocked or redirected, with user prompts triggered for persistent blocked actions.

Who can use auto mode?

Currently, auto mode is available as a research preview for Claude Team users, with plans to roll out to Enterprise and API users; it supports Claude Sonnet 4.6 and Opus 4.6 models, and can be enabled through settings in the desktop app, CLI, or VS Code extension.

What are the risks of using auto mode?

While safer than skipping all permissions, auto mode does not eliminate risks entirely; the classifier may occasionally block harmless actions or allow risky ones if context is unclear, so it is recommended for use in isolated environments.

How is auto mode different from skipping permissions?

Auto mode provides a middle path with safety checks via a classifier, unlike '--dangerously-skip-permissions' which has no checks, reducing risk while allowing more autonomy for longer tasks with fewer interruptions.