Claude Code Permission Modes Explained (2026)

TL;DR: Claude Code has three permission modes — default (prompts for everything), allowEdits (auto-approves file edits, prompts for commands), and bypassPermissions (auto-approves everything). Configure via CLI flag, settings.json, or PreToolUse hooks.

The Problem

Claude Code keeps asking for permission on every tool call, slowing down your workflow:

Permission rule Bash requires confirmation for this command:
 cat package.json
 Allow? [y/n]

Or conversely, you want to restrict Claude Code from running certain commands but the permission system does not seem configurable enough.

Why This Happens

Claude Code defaults to the safest permission mode — prompting for confirmation on every impactful tool call (file writes, shell commands, etc.). This is intentional for security, but becomes friction for experienced users who want more autonomy.

The Fix

Step 1 — Choose Your Permission Mode

Mode File Edits Shell Commands Use Case
default Prompt Prompt New users, shared machines
allowEdits Auto-allow Prompt Trusted projects, edit-heavy workflows
bypassPermissions Auto-allow Auto-allow CI/CD, sandboxed environments, power users

Step 2 — Configure via CLI Flag

# Allow file edits without prompts
claude --permission-mode allowEdits
# Bypass all permissions (use with caution)
claude --dangerously-skip-permissions

Step 3 — Configure via Settings File

For persistent configuration, edit ~/.claude/settings.json:

{
 "defaultMode": "bypassPermissions",
 "skipDangerousModePermissionPrompt": true,
 "permissions": {
 "allow": [
 "Bash(npm test)",
 "Bash(npm run *)",
 "Bash(git *)",
 "Read(*)",
 "Write(src/*)"
 ],
 "deny": [
 "Bash(rm -rf *)",
 "Bash(sudo *)"
 ]
 }
}

Step 4 — Fine-Grained Control with PreToolUse Hooks

For the most control, use a PreToolUse hook that programmatically decides permissions:

{
 "hooks": {
 "PreToolUse": [
 {
 "matcher": "Bash",
 "hooks": [
 {
 "type": "command",
 "command": "/path/to/permission-hook.sh",
 "timeout": 30
 }
 ]
 }
 ]
 }
}

permission-hook.sh:

#!/bin/bash
# Read the tool input from stdin
INPUT=$(cat)
# Extract the command being run
COMMAND=$(echo "$INPUT" | python3 -c "import json,sys; print(json.load(sys.stdin).get('input',{}).get('command',''))" 2>/dev/null)
# Allow safe commands, deny dangerous ones
case "$COMMAND" in
 rm\ -rf*|sudo*|chmod\ 777*)
 echo '{"hookSpecificOutput":{"hookEventName":"PreToolUse","permissionDecision":"deny","reason":"Blocked by safety hook"}}'
 ;;
 *)
 echo '{"hookSpecificOutput":{"hookEventName":"PreToolUse","permissionDecision":"allow"}}'
 ;;
esac

chmod +x /path/to/permission-hook.sh

Step 5 — Verify Your Configuration

# Start Claude Code and check the active mode
claude --permission-mode allowEdits
# In the session, run a test command
> Read the file package.json
# Should auto-approve without prompting
> Run: echo "hello world"
# In allowEdits mode, this should still prompt

Common Variations

Scenario Cause Quick Fix
Permissions reset after background task Known bug (issue #47810) Use hook-based permissions as workaround
bypassPermissions still prompts Settings not loaded Check file path, restart Claude Code
Hook not being called Wrong matcher pattern Use "matcher": ".*" to match all tools
Want to allow git but deny rm Need fine-grained rules Use permissions.allow and permissions.deny arrays
Sandbox mode overrides settings Sandbox has its own rules Set sandbox.autoAllowBashIfSandboxed: true

Prevention

  • Start with allowEdits: It is a good balance between safety and productivity for most developers.
  • Use project-level settings: Put permissive settings in project .claude/settings.json rather than global settings to limit scope.
  • Review hook logs: If using PreToolUse hooks, log every decision for auditability.
**Written by Michael** — solo dev, Da Nang, Vietnam. 50K+ Chrome extension users. $500K+ on Upwork (100% Job Success). Runs 5 Claude Max subs in parallel. Built this site with autonomous agent fleets. [See what I'm building →](https://zovo.one)
This took me 3 hours to figure out. I put it in a CLAUDE.md so I'd never figure it out again. Now Claude gets it right on the first try, every project. 16 framework templates. Next.js, FastAPI, Laravel, Rails, Go, Rust, Terraform, and 9 more. Each one 300+ lines of "here's exactly how this stack works." Copy into your project. Done. **[See the templates →](https://zovo.one/lifetime?utm_source=ccg&utm_medium=cta-config&utm_campaign=claude-code-permission-modes)** $99 once. Yours forever. I keep adding templates monthly.

Set it up → Build your permission config with our Permission Configurator.

Find commands → Search all commands in our Command Reference.

Try it: Paste your error into our Error Diagnostic for an instant fix.

Last verified: 2026-04-14. Found an issue? Open a GitHub issue.

Common Questions

How do I get started with claude code permission modes?

Begin with the setup instructions in this guide. Install the required dependencies, configure your environment, and test with a small project before scaling to your full codebase.

What are the prerequisites?

You need a working development environment with Node.js or Python installed. Familiarity with the command line and basic Git operations is helpful. No advanced AI knowledge is required.

Can I use this with my existing development workflow?

Yes. These techniques integrate with standard development tools and CI/CD pipelines. Start by adding them to a single project and expand once you have verified the benefits.

Where can I find more advanced techniques?

Explore the related resources below for deeper coverage. The Claude Code documentation and community forums also provide advanced patterns and real-world case studies.