Skip to content

API Keys

wtf needs an AI provider to work. You pick: Anthropic (Claude), OpenAI (GPT), Google (Gemini), or run locally with Ollama (no API key needed!).

Supported Providers

Provider Models Best For
Anthropic Claude Sonnet 4, Claude Opus 4, Claude 3.5 Sonnet/Haiku Best overall, great at following instructions
OpenAI GPT-4o, GPT-4o Mini, o1, o3, GPT-5 Fast, widely available, reasoning models
Google Gemini 2.0, Gemini 1.5 Pro/Flash Good balance of speed and quality
Local (Ollama) Llama 3.2, Mistral, Qwen, DeepSeek, CodeLlama Free, private, offline - no API key needed

Getting an API Key

  1. Sign up at https://console.anthropic.com
  2. Go to Settings → API Keys
  3. Create a new key
  4. Free tier: $5 credit, then pay-as-you-go
$ wtf here is my anthropic api key sk-ant-...
✓ API key saved
✓ Using claude-3.5-sonnet as default model

OpenAI

  1. Sign up at https://platform.openai.com
  2. Go to API Keys
  3. Create a new secret key
  4. Pricing: Pay-as-you-go
$ wtf here is my openai api key sk-...
✓ API key saved
✓ Using gpt-4o as default model

Google

  1. Go to https://makersuite.google.com/app/apikey
  2. Create an API key
  3. Free tier: 60 requests/minute
$ wtf here is my google api key AIza...
✓ API key saved
✓ Using gemini-pro as default model

Local Models (Ollama) - No API Key Needed!

Run AI completely locally on your machine. Free, private, and works offline.

  1. Install Ollama from https://ollama.ai
  2. Pull a model:
# Pick one (or several!)
ollama pull llama3.2        # Best all-around
ollama pull mistral         # Fast and capable
ollama pull codellama       # Great for coding
ollama pull deepseek-r1     # Advanced reasoning
ollama pull qwen2.5         # Strong multilingual
  1. Run setup - wtf will auto-detect your models:
$ wtf --setup
✓ Local models detected (Ollama)

  1. Ollama Llama 3.2 (local, free)
  2. Ollama Mistral (local, free)
  ...

Why local? - 🔒 100% private - nothing leaves your machine - 💰 Free forever - no API costs - 🌐 Works offline - no internet needed - ⚡ Fast - no network latency (if you have good hardware)

Setup During First Run

The first time you run wtf, it asks you to pick a provider:

$ wtf show my git status

No API key configured. Let's set one up.

Which AI provider do you want to use?
  1. Anthropic (Claude) - Recommended
  2. OpenAI (GPT-4)
  3. Google (Gemini)

Choose [1-3]: 1

Great! Get your API key from:
https://console.anthropic.com/settings/keys

Paste your API key: sk-ant-...
✓ API key saved

Switching Providers

Change providers anytime:

$ wtf switch to OpenAI
What's your OpenAI API key?
Paste it here: sk-...
✓ Switched to OpenAI (gpt-4o)

Or re-run setup:

$ wtf --setup

Changing Models

Use a different model from the same provider:

$ wtf use claude-opus as my model
✓ Switched to claude-opus

$ wtf use gpt-3.5-turbo
✓ Switched to gpt-3.5-turbo (faster, cheaper)

Where Keys Are Stored

By default, wtf uses Simon Willison's llm library for key management:

# Keys stored in llm's keyring
~/.config/io.datasette.llm/keys.json

This is secure and shared with other llm-based tools.

Via Environment Variables

You can also use environment variables:

# Add to ~/.zshrc or ~/.bashrc
export ANTHROPIC_API_KEY="sk-ant-..."
export OPENAI_API_KEY="sk-..."
export GOOGLE_API_KEY="AIza..."

Then:

$ source ~/.zshrc
$ wtf --setup

wtf will detect the environment variable automatically.

Keys can be stored in ~/.config/wtf/config.json, but this is less secure:

{
  "api": {
    "model": "claude-3.5-sonnet",
    "key_source": "config",
    "key": "sk-ant-..."
  }
}

We don't recommend this. Use llm's keyring or environment variables instead.

Checking Your Setup

$ wtf what API key am I using?
You're using Anthropic (claude-3.5-sonnet)
API key: sk-ant-...xyz (last 3 chars)

Or check the config:

$ wtf --config
Config directory: ~/.config/wtf/
Config file: ~/.config/wtf/config.json

Current settings:
  Model: claude-3.5-sonnet
  Key source: llm

Model Costs

Approximate costs per 1M tokens (as of 2024):

Model Input Output Speed
GPT-4o $5 $15 Fast
Claude Sonnet 3.5 $3 $15 Medium
Claude Haiku $0.25 $1.25 Very fast
Gemini Pro $0.50 $1.50 Fast

For typical wtf usage: ~$0.01-0.05 per query

Troubleshooting

"Invalid API key" error

$ wtf test my setup
✗ Error: Invalid API key

Fix:
1. Double-check you copied the full key
2. Make sure the key is active (check provider dashboard)
3. Re-run setup: wtf --setup

"Rate limit exceeded"

✗ Error: Rate limit exceeded

Fix:
1. Wait a few minutes
2. Check your provider's rate limits
3. Upgrade your API plan if needed

Key not being detected

$ echo $ANTHROPIC_API_KEY
# (should show your key)

# If empty:
$ export ANTHROPIC_API_KEY="sk-ant-..."
$ source ~/.zshrc

Security Best Practices

✓ Do: - Use environment variables or llm's keyring - Keep keys in .env files (add to .gitignore) - Rotate keys periodically - Use separate keys for different projects

✗ Don't: - Commit keys to git - Share keys in Slack/email - Store keys in plain text files - Use the same key everywhere

Next Steps