Setup¶

So you've installed wtf. Now what? Don't worry, it's not complicated.

First Run¶

The first time you run wtf, it'll walk you through a brief setup. No forms to fill out. No accounts to create. Just three questions.

$ wtf "help me"

The Setup Wizard¶

wtf will ask you:

Choose your AI provider (Anthropic, OpenAI, or Google)
Enter your API key
Select your default model

That's it. Takes about 30 seconds.

API Keys

Don't have an API key? See API Keys for where to get one.

What Gets Configured¶

The wizard creates a config file at ~/.config/wtf/config.yaml with:

Your chosen AI provider
Your default model
Basic settings (all sane defaults)

Your API key is stored securely in your system keychain when possible, or in the config file with appropriate permissions.

Testing It Out¶

Once setup is complete, try a simple command:

$ wtf "what's in my git status?"

If you see wtf respond with something helpful (or sarcastic), you're good to go.

Configuration Files Created¶

After first run, you'll have:

~/.config/wtf/
├── config.yaml          # Main configuration
├── conversation.jsonl   # Conversation history
└── allowlist.json       # Auto-execute permissions (created as needed)

Next Steps¶

Quick Tour - See what wtf can do
Memories - Teach wtf your preferences
Permissions - Control what runs automatically

Reconfiguring¶

Want to change providers or models later? Just run:

$ wtf "use a different model"

Or edit ~/.config/wtf/config.yaml directly. We're not picky.

Troubleshooting Setup¶

"API key invalid"¶

Double-check you copied the key correctly
Make sure the key has the right permissions
Some providers have separate keys for different models

"Can't find config directory"¶

wtf creates ~/.config/wtf/ automatically. If it fails:

Check you have write permissions to ~/.config/
Try creating it manually: mkdir -p ~/.config/wtf

"Model not available"¶

Your API key might not have access to that model. Try a different one or check your provider's console.

Need more help? Check Troubleshooting.