Claude Code Desktop Bypass Permissions: How to Enable It

Q: What is the Claude Code CLI command for bypass permissions?

Use claude --permission-mode bypassPermissions. The older equivalent command is claude --dangerously-skip-permissions.

Name: Soren Atelier

Updated on 4/20/2026

Bypass permissions is one of those Claude Code settings you only care about after Claude has interrupted your workflow for the tenth "Ask for permission" prompt. In the CLI, developers often start Claude Code with a bypass flag so it can edit files and run commands without asking for every small approval. In Claude Code Desktop, the same idea exists, but the setting is easier to miss.

This guide shows the current Claude Code Desktop path for enabling Bypass permissions, then adds the matching CLI commands for users who also search for --bypass-permission, --permission-mode bypassPermissions, or --dangerously-skip-permissions.

As of April 20, 2026, the Desktop path is:

Open Claude Code Desktop.
Click your profile / personal menu in the lower-left corner.
Click Settings.
In Settings, choose Claude Code.
Turn on Allow bypass permissions mode.

Officially, Anthropic describes Bypass permissions as the bypassPermissions permission mode. It runs without normal permission prompts, except for protected paths. See the official Claude Code permission modes docs (opens in a new tab) and Claude Code Desktop docs (opens in a new tab) for the underlying behavior.

If you want broader AI coding workflow context, see the AI Coding topic hub, How to Use Codex, and Parallel Code Agents Explained.

Quick Answer

To enable Claude Code Desktop bypass permissions:

Step	Action
1	Open Claude Code Desktop
2	Click the lower-left profile menu
3	Click Settings
4	Select Claude Code in the settings sidebar
5	Enable Allow bypass permissions mode

After that, use the permission mode selector in your Claude Code session if the current session still starts in a stricter mode.

Why This Setting Matters

Claude Code's default permission behavior is intentionally conservative. It asks before many file edits, shell commands, and tool actions. That is the safer default, but it becomes noisy during long coding tasks.

Typical examples:

Claude fixes one lint error, then asks before running the next command.
Claude edits a file, then asks before touching another related file.
Claude runs tests, finds a failure, and asks again before applying the obvious follow-up fix.
Claude needs to repeat the same kind of command across a repo and keeps pausing.

Bypass permissions removes most of that approval friction. It is the Desktop equivalent of running Claude Code in the CLI with bypassPermissions mode. The tradeoff is simple: fewer interruptions, much more responsibility.

Use it for contained work where you can review the final diff. Do not use it as a casual default in a sensitive workspace.

Step 1: Open the Lower-Left Profile Menu

Open Claude Code Desktop and look at the lower-left corner of the app. Click your profile or personal menu. This is where the route to Settings begins.

Open the lower-left personal menu, then choose Settings.

Step 2: Click Settings

In the menu, click Settings. This opens the full settings page for Claude Desktop and Claude Code.

The bypass permissions control is not a general account setting. It belongs to Claude Code because it changes how Claude Code handles tool approvals during coding sessions.

Step 3: Select Claude Code

In the settings sidebar, find and select Claude Code.

This page contains Claude Code-specific settings such as permission modes, preview behavior, worktree location, and related desktop coding controls.

Step 4: Enable Allow Bypass Permissions Mode

Inside the Claude Code settings page, find Claude code desktop settings and turn on Allow bypass permissions mode.

Claude Code settings screen with Allow bypass permissions mode enabled

The Bypass permissions switch lives under Settings -> Claude Code. Turn on Allow bypass permissions mode.

Once enabled, open or restart the Claude Code session where you want to use it. If you still see a permission mode selector next to the prompt box, choose Bypass permissions there.

A Note for Jupyter Workflows

Bypass permissions helps Claude Code move faster in a coding workspace, but notebook-heavy work has a different bottleneck. In Jupyter, the hard part is often not just approving edits. It is whether the agent understands the live notebook state: current variables, DataFrames, cell outputs, execution order, and errors from the kernel.

If your workflow is data analysis, EDA, feature engineering, or model debugging inside notebooks, RunCell (opens in a new tab) is worth a look. RunCell is a Jupyter-native AI agent with a stronger notebook harness: it works inside Jupyter, reads notebook context, understands DataFrames and outputs, and can help generate, run, and debug cells without forcing you to translate everything into a terminal-first coding loop.

For a deeper comparison, see Jupyter AI RunCell for Notebook Debugging and Data Work and Can Claude Code Analyze Jupyter Notebooks for Data Science?.

Claude Code Desktop vs CLI

Claude Code Desktop exposes Bypass permissions as a settings toggle plus a session permission mode. Claude Code CLI exposes the same underlying mode through command-line flags.

Interface	How to enable bypass permissions
Claude Code Desktop	Settings -> Claude Code -> Allow bypass permissions mode, then choose Bypass permissions in the session if needed
Claude Code CLI	Start with `claude --permission-mode bypassPermissions`
Older CLI habit	Start with `claude --dangerously-skip-permissions`

The key name is bypassPermissions. That is the permission mode used by Claude Code.

How to Use Bypass Permissions in Claude Code CLI

If you also use Claude Code from the terminal, the current explicit command is:

claude --permission-mode bypassPermissions

Anthropic's docs also say the older flag is equivalent:

claude --dangerously-skip-permissions

You may see people search for or describe this as --bypass-permission, --bypass-permissions, or "Claude Code bypass permission flag." The current documented CLI spelling is not --bypass-permission; use --permission-mode bypassPermissions if you want the clear, mode-based form.

There is also an enable-without-activating variant:

claude --allow-dangerously-skip-permissions

That makes Bypass permissions available in the session mode cycle without immediately starting in bypass mode. In the CLI, you can cycle permission modes with Shift+Tab after the mode is available.

Should You Use Bypass Permissions?

Use Bypass permissions when the workspace is contained and the cost of repeated prompts is higher than the risk of letting Claude continue.

Good fits:

a disposable branch
a devcontainer
a VM
a sandboxed local repo
repetitive lint, format, or test-fix work
generated boilerplate that you will review with git diff

Poor fits:

repos with production deploy scripts
workspaces containing secrets or customer data
cloud CLIs already logged into production
tasks that can delete, migrate, or publish real resources
repos with code you have not reviewed

Anthropic's docs are blunt about this: bypassPermissions is for isolated environments. If you want fewer interruptions with more safety checks, consider Auto mode instead.

Bypass Permissions vs Auto Permissions

In the same Claude Code settings area, you may also see Allow auto permissions mode. It is related, but it is not the same setting.

Mode	What happens	Best for
Default	Claude asks before many edits and commands	New projects, sensitive work, review-heavy sessions
Auto accept edits	Claude can edit files and run common filesystem commands with fewer prompts	Normal coding where you still want command approvals
Auto	Claude runs with background safety checks	Longer tasks where you want fewer prompts but not full bypass
Bypass permissions	Claude skips normal permission prompts except protected paths	Sandboxed work where speed matters more than approval gates

For most developers, Bypass permissions should be temporary. Turn it on for a contained task, review the diff, then return to a stricter mode.

Troubleshooting

I enabled the toggle, but Claude still asks for permission

Start a new Claude Code session and check the session permission mode selector. The settings toggle makes the mode available, but your active session may still be running in another mode.

I do not see Allow bypass permissions mode

Check that you are in Settings -> Claude Code. Also update Claude Desktop and restart the app. If you are on a managed Team or Enterprise environment, an admin may have disabled Bypass permissions mode.

Which CLI flag should I use?

Use:

claude --permission-mode bypassPermissions

The older equivalent is:

claude --dangerously-skip-permissions

Do not rely on --bypass-permission unless your local Claude Code version explicitly documents it. The current official docs use --permission-mode bypassPermissions.

Is Bypass permissions safe?

It is faster, not safer. It removes many prompts that normally slow Claude down. Use it only when you have a contained workspace and a review plan.

FAQ

How do I enable Bypass permissions in Claude Code Desktop?

Open the lower-left profile menu, click Settings, select Claude Code, then enable Allow bypass permissions mode.

What is the Claude Code CLI command for bypass permissions?

Use claude --permission-mode bypassPermissions. The older equivalent command is claude --dangerously-skip-permissions.

Is `--bypass-permission` the right Claude Code flag?

Not in the current official docs. The documented mode-based command is claude --permission-mode bypassPermissions.

Is Bypass permissions the same as Auto mode?

No. Auto mode uses background safety checks. Bypass permissions skips normal permission prompts except protected paths and should be used only in isolated environments.

Why is Bypass permissions useful?

It prevents Claude Code from stopping repeatedly for manual approval during long coding tasks, especially repetitive edit-test-fix loops.

Related Guides

📚