Claude Code Introduces Auto Mode: Eliminates Repeated Approvals for Long Tasks—Safer Than Dangerous Mode
TechFlow Selected TechFlow Selected
Claude Code Introduces Auto Mode: Eliminates Repeated Approvals for Long Tasks—Safer Than Dangerous Mode
Currently available to Team plan users as a research preview.
Navigating Web3 tides with focused insightsAuthor: Anthropic
Compiled and translated by TechFlow
TechFlow Intro: Previously, Claude Code either required manual approval for every step or bypassed all checks entirely using --dangerously-skip-permissions.
Auto Mode offers a middle path: AI autonomously evaluates the safety of each action; risky operations are automatically blocked, enabling truly hands-off execution of long-running tasks.
Currently available as a research preview to Team plan users.
Full text below:
Auto Mode for Claude Code
Today, we’re introducing Auto Mode—a new permission mode for Claude Code. In this mode, Claude makes permission decisions on your behalf and performs safety checks before executing each operation. It is currently available as a research preview to Team plan users and will roll out to Enterprise plan and API users in the coming days.
How It Works
The default permission settings for Claude Code are intentionally conservative: every file write and bash command execution requires your manual approval. This is a safe default—but it comes at the cost of preventing you from launching large tasks and stepping away, since Claude will frequently pause for human confirmation during execution. Some developers opt to use --dangerously-skip-permissions to skip all permission checks, but doing so can lead to dangerous or destructive outcomes and should only be used in isolated environments.
Auto Mode strikes a middle ground: enabling longer tasks with fewer interruptions, while introducing less risk than skipping all permissions entirely. Before each tool call executes, a classifier reviews it to detect potentially destructive actions—such as bulk file deletion, sensitive data exfiltration, or execution of malicious code.
Operations classified as safe execute automatically; risky ones are blocked, and Claude is guided toward alternative approaches. If Claude persists in attempting repeatedly blocked actions, it will ultimately trigger a permission prompt directed to you.
Usage Expectations
Compared to --dangerously-skip-permissions, Auto Mode reduces risk—but does not eliminate it entirely. We still recommend using it only in isolated environments. The classifier may occasionally allow risky operations—for example, when user intent is ambiguous, or when Claude lacks sufficient contextual awareness of your environment to assess whether an action carries additional risk. It may also occasionally block harmless operations. We’re continuously improving this experience.
Auto Mode may have a minor impact on token consumption, cost, and latency for tool calls.
Getting Started
Auto Mode is available today as a research preview for Claude Team users and will be rolled out to Enterprise and API users in the coming days. It supports Claude Sonnet 4.6 and Opus 4.6.
Administrators: Auto Mode will soon be available to all Claude Code users across Enterprise, Team, and Claude API plans. To disable it in the CLI and VS Code extension, set "disableAutoMode": "disable" in your managed settings. Auto Mode is disabled by default in the Claude desktop app; you can enable it via Organization Settings → Claude Code.
Developers: Run claude --enable-auto-mode to enable Auto Mode, then press Shift+Tab to switch into it. In the desktop app and VS Code extension, first enable Auto Mode under Settings → Claude Code, then select it from the Permission Mode dropdown menu within your session.
Learn more in the documentation.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
@claudeai
