How to use Claude without prompts?

To use Claude Code without permission prompts, run 'claude --dangerously-skip-permissions'. This command enables auto-approve mode where Claude skips all confirmation dialogs. You can also combine it with --continue to resume the last session or with -p for one-shot prompts.

How to bypass permissions in Claude CLI?

To bypass all permission checks in Claude Code CLI, use the --dangerously-skip-permissions flag. This tells Claude to skip permission prompts for file edits, command execution, and web requests. Useful for automation scripts and CI/CD pipelines in trusted environments.

How to install Claude Code CLI?

Install Claude Code on macOS with 'brew install --cask claude-code', on Linux with 'curl -fsSL https://claude.ai/install.sh | bash', or on Windows with 'winget install Anthropic.ClaudeCode'. After installing, run 'claude' in your terminal to start, or 'claude --dangerously-skip-permissions' to skip all permission prompts.

Can I use claude --dangerously-skip-permissions with --continue?

Yes, you can combine flags: 'claude --dangerously-skip-permissions --continue' resumes your last Claude Code session with all permission prompts skipped. You can also use 'claude --dangerously-skip-permissions -p "your prompt"' for one-shot auto-mode commands in CI/CD pipelines.

What permissions does claude --dangerously-skip-permissions skip?

The --dangerously-skip-permissions flag skips three types of permission prompts: file edit confirmations (creating, modifying, deleting files), shell command execution (running terminal commands), and web request approvals (fetching URLs, calling APIs). All actions are auto-approved without user interaction.

claude --dangerously-skip-permissions | Skip Permissions Claude Code CLI

Essential Commands

ESSENTIAL

claude --dangerously-skip-permissions

Run Claude Code in full auto-mode with no confirmation prompts. Skips all permission checks for file edits, command execution, and web requests. Use only when you fully trust the project repository.

claude --dangerously-skip-permissions

Install Claude Code

macOS

Install via Homebrew

Recommended for macOS — installs Claude Code as a cask.

brew install --cask claude-code

Linux / macOS

Install via curl

Universal installer script — works on Linux and macOS.

curl -fsSL https://claude.ai/install.sh | bash

Windows

Install via winget

Windows Package Manager — the native way to install on Windows.

winget install Anthropic.ClaudeCode

What does claude --dangerously-skip-permissions do?

The --dangerously-skip-permissions flag tells Claude Code CLI to skip all interactive permission prompts. Normally, Claude Code asks for confirmation before editing files, running shell commands, or making web requests. With this flag enabled, all actions are auto-approved — no interruptions, no confirmations.

This is the fastest way to use Claude Code in auto-mode. It is commonly combined with --continue to resume the last session, or with -p for one-shot prompts in CI/CD pipelines and automation scripts. Perfect for trusted repositories where you want maximum speed.

Always review changes after running in this mode. The flag name includes "dangerously" as a deliberate reminder that you are granting the AI agent full access to your project files and shell.

How to skip permissions in Claude Code

To skip permissions in Claude Code, open your terminal and run claude --dangerously-skip-permissions. This single flag bypasses all permission checks — file edits, shell commands, and web requests are all auto-approved. No more clicking "Allow" for every action.

You can also combine it with other flags for different workflows:

claude --dangerously-skip-permissions --continue — resume your last Claude session with permissions skipped
claude --dangerously-skip-permissions -p "your prompt" — run a one-shot prompt in auto-mode without any confirmation dialogs
claude --dangerously-skip-permissions --model opus — skip permissions and use a specific model

This flag works on macOS, Linux, and Windows. If you haven't installed Claude Code yet, use brew install --cask claude-code on macOS, curl -fsSL https://claude.ai/install.sh | bash on Linux, or winget install Anthropic.ClaudeCode on Windows.

When to use --dangerously-skip-permissions

Claude bypass permissions is ideal for these situations:

Personal projects — you control the codebase and trust Claude to make changes without asking
CI/CD automation — running Claude in pipelines where no human is present to approve actions
Rapid prototyping — when you want Claude to iterate quickly without pausing for confirmations
Batch operations — refactoring, testing, or deploying across multiple files at once

When NOT to use it: on untrusted or unfamiliar codebases, in production environments without review, or when working on security-sensitive code. The --dangerously-skip-permissions flag gives Claude full auto-approve access — use it wisely.

Related Claude Code commands

Besides --dangerously-skip-permissions, Claude Code CLI has other useful flags for developers:

claude --continue — continue the most recent conversation
claude --resume — resume a specific previous session
claude -p "prompt" — run a one-shot prompt and exit
claude config — view or edit Claude Code configuration
claude mcp — manage Model Context Protocol servers

Visit our About page to learn how this site was built entirely with Claude Code, or check the Danger Leaderboard to see which country lives most dangerously.