Claude Code Configuration: 50+ Settings You Should Know

“I’ve been using Claude Code for a while now, but I had no idea you could configure all of this!” – I hear this regularly from developers in my courses. Time to change that.

Claude Code is far more than an AI coding assistant. It is a highly configurable development tool with over 50 environment variables, hierarchical settings levels and granular permission controls. In this article I will show you the most important configuration options.

Understanding the Settings Hierarchy

Claude Code uses a hierarchical configuration system. This means: settings at higher levels override lower ones. The order (from highest to lowest priority):

1. Enterprise Managed Policies (managed-settings.json) – Deployed by IT/DevOps, not overridable
2. Command Line Arguments – Temporary overrides for a session
3. Local Project Settings (.claude/settings.local.json) – Personal, non-committed settings
4. Shared Project Settings (.claude/settings.json) – Team settings in the repository
5. User Settings (~/.claude/settings.json) – Global personal settings

Practical tip: Use .claude/settings.local.json for experiments – this file is automatically ignored by Git.

The Most Important Environment Variables

Here are the environment variables that make the biggest difference in practice:

Performance & Costs

Bash & Shell Behaviour

Enterprise & Proxy

Alternative Model Providers

Permission Settings: Granular Control

The permissions settings in settings.json give you precise control over what Claude Code may and may not do:

{
  "permissions": {
    "allow": [
      "Bash(npm run lint)",
      "Bash(npm run test:*)",
      "Read(~/.zshrc)"
    ],
    "deny": [
      "Bash(curl:*)",
      "Read(./.env)",
      "Read(./.env.*)",
      "Read(./secrets/**)"
    ],
    "defaultMode": "acceptEdits"
  }
}

The Three Permission Types

• allow: Tools/commands that are executed without asking
• ask: Tools that require confirmation
• deny: Completely blocked tools/paths

Security tip: Use deny rules for sensitive files such as .env, credentials and secrets. Claude Code will then ignore these files entirely.

Sandbox Settings: Isolation for More Security

The sandbox isolates Bash commands from the filesystem and network. Especially important for teams working with sensitive code:

{
  "sandbox": {
    "enabled": true,
    "autoAllowBashIfSandboxed": true,
    "excludedCommands": ["git", "docker"],
    "network": {
      "allowUnixSockets": ["/var/run/docker.sock"],
      "allowLocalBinding": true
    }
  }
}

Important: The sandbox is currently only available on macOS and Linux.

Team Announcements: Communication Within the Team

An underrated feature – companyAnnouncements displays messages on startup:

{
  "companyAnnouncements": [
    "Willkommen! Unsere Code Guidelines findest du unter docs.company.com",
    "Reminder: Code Reviews sind für alle PRs erforderlich",
    "Neue Security Policy seit 01.02.2026 aktiv"
  ]
}

Perfect for onboarding, policy updates or team announcements.

Hooks: Automation Before and After Tool Execution

With hooks you can run your own commands before or after Claude Code uses tools:

{
  "hooks": {
    "PreToolUse": {
      "Bash": "echo 'Running command...'"
    }
  }
}

Use cases:

• Automatic formatting after file edits
• Logging/auditing of commands
• Blocking certain operations

My Top 5 Configurations for Everyday Use

From my daily work with Claude Code and teaching at FFHS/TSBE, here are my personal favourites:

1. Extended Thinking for Complex Tasks

export MAX_THINKING_TOKENS=10000

2. Protect Sensitive Files

{
  "permissions": {
    "deny": ["Read(./.env)", "Read(./secrets/**)", "Read(./*.pem)"]
  }
}

3. Project-Specific Model

{
  "model": "claude-sonnet-4-5-20250929"
}

4. Disable Co-Authored-By in Commits

{
  "includeCoAuthoredBy": false
}

5. Custom Status Line for Context

{
  "statusLine": {
    "type": "command",
    "command": "~/.claude/statusline.sh"
  }
}

Conclusion: Invest 30 Minutes in Your Configuration

Most developers use Claude Code “out of the box” – and thereby give away enormous potential. A well-thought-out configuration makes you more productive, protects sensitive data and adapts the tool to your workflow.

My advice: Take 30 minutes, create a ~/.claude/settings.json for your global preferences and a .claude/settings.json for each important project.

You can find the full documentation at code.claude.com/docs/en/settings.

Which Claude Code settings do you use? Write to me on LinkedIn – I’m curious about your setups!