Skip to main content

YOLO Mode

YOLO mode does exactly what the name suggests - Caret auto-approves everything. Check the box in feature settings, and it executes file changes, terminal commands, even transitions from Agent to Act mode without asking.

Think of it as auto-approve on steroids - instead of granular permissions, YOLO mode gives Caret complete autonomy.

ℹ️Note

Warning: This is dangerous. YOLO mode disables all safety rails. Caret will execute anything it decides to do, without permission.

What Gets Auto-Approved​

When YOLO mode is active, Caret automatically approves:

  • All File Operations - Reading, writing, modifying files anywhere in your system
  • All Terminal Commands - Including potentially destructive operations
  • Browser Actions - Web scraping, form submissions, navigation
  • MCP Server Tools - External integrations and API calls
  • Mode Transitions - Automatic switching from Agent to Act mode

Essentially, all safety guardrails are removed. Caret operates with full autonomy.

How to Enable YOLO Mode​

Go to Caret Settings → Features and check the "YOLO Mode" checkbox. That's it - no confirmation dialog, no additional warnings. Once enabled, Caret immediately starts auto-approving all operations.

To disable, just uncheck the box. Pending operations still require approval once YOLO mode is off.

When to Use It​

YOLO mode was primarily built for scriptable CLI where full autonomous execution is needed. In GUI, you might use it for:

Rapid Prototyping: When there's zero friction and you don't care about mistakes. Perfect for throwaway experiments or quickly exploring new ideas.

Validated Repetitive Tasks: When you've already validated Caret's approach and want to eliminate approval overhead. Think routine refactoring or established patterns.

Demo Purposes: When you want to showcase Caret's capabilities without constant interruptions.

What Could Go Wrong?​

Since YOLO mode removes all safety nets, Caret can:

  • Delete critical files without warning
  • Execute commands that modify system settings
  • Make network requests to external services
  • Overwrite configuration files
  • Install or remove software packages
  • Commit and push changes to version control

The risk level entirely depends on what you ask Caret to do. Simple tasks are relatively safe, but complex requests can lead to unpredictable results.

Best Practices​

If you decide to use YOLO mode:

Start in Isolated Environments. Use it first in throwaway projects or sandbox environments. Never enable in production codebases until you understand the risks.

Be Specific with Requests. Vague instructions combined with unlimited permissions lead to unexpected results. The clearer your requirements, the more predictable Caret's behavior.

Monitor the Output. Even though Caret doesn't ask permission, it still shows you what it's doing. Watch terminal output and file changes in real-time.

Have Version Control Ready. Make sure you can easily revert changes when something goes wrong. Git becomes your safety net when YOLO mode is your workflow.

The Inspiration: What Becomes Possible?​

With YOLO mode enabled:

Build Entire Applications in Single Prompts: Describe what you want and Caret handles everything - creating files, installing dependencies, configuring settings, even deployment scripts.

Automate Complex Workflows: Automate tasks that normally require dozens of approval clicks. Data processing pipelines, build system setups, or multi-step refactoring jobs become seamless.

Rapid Iteration Cycles: Test ideas quickly without approval friction. Perfect for exploring different approaches or experimenting with new technologies.

Live Demos: Showcase Caret's full capabilities without stopping to approve every action. Great for presentations or teaching scenarios.

The key is understanding that YOLO mode transforms Caret from interactive assistant to autonomous agent. Use that power wisely.

Questions or feedback? Reach out in GitHub Discussions.