createLLMProviderFromConfig
steamharness.createLLMProviderFromConfig(): LLMProvider
createLLMProviderFromConfig(config?: ConfigurationValues): LLMProvider
The runtime factory reads the current execution snapshot. The named SDK export also accepts an explicit configuration object. Neither reads environment variables.
| Key | Aliases | Meaning |
|---|---|---|
LLM_TYPE | TYPE | openai-completion, openai-responses, anthropic |
LLM_API_KEY | API_KEY, APIKEY | Required API key |
LLM_BASE_URL | BASE_URL | Optional endpoint |
LLM_MODEL | MODEL | Optional model |
Declare these string fields in configuration.json and enter values in Project configurations. Mark API key fields sensitive. Missing required keys or unknown provider types produce input_validation. Defaults are gpt-4.1-mini for OpenAI and claude-3-haiku-20240307 for Anthropic.
const provider = steamharness.createLLMProviderRef("default", () =>
steamharness.createLLMProviderFromConfig());