Set up model API keys
Testronaut uses credentials from the model providers directly. Create keys only through each provider's official console, keep them out of source control, and review that provider's billing and usage limits.
| Provider | Create or manage keys | Official documentation | Testronaut variable |
|---|---|---|---|
| OpenAI | OpenAI API keys | OpenAI API quickstart | OPENAI_API_KEY |
| Anthropic Claude | Claude Platform API keys | Claude API getting started | ANTHROPIC_API_KEY |
| Google Gemini | Google AI Studio API keys | Using Gemini API keys | GEMINI_API_KEY |
| Jev / TypeSafe AI | TypeSafe AI API console | Jev API reference | TESTRONAUT_JEV_API_KEY |
Add the keys you use to the project's .env file:
# Choose one execution-model provider.
OPENAI_API_KEY=
# ANTHROPIC_API_KEY=
# GEMINI_API_KEY=
# Optional, only when opting in to Jev guardrails.
TESTRONAUT_JEV_API_KEY=
Do not commit .env. Provider API access is separate from a ChatGPT, Claude, or Gemini consumer subscription, and may require API billing to be enabled in the corresponding developer account.
After configuration, run testronaut config --json. It reports the selected execution provider and whether a Jev credential was detected, but never prints secret values.