LLM providers
There is no blank-imported llm_provider tool module in
digitornd. LLM access is configured per agent via brain: and
implemented by the runtime LLM stack.
Typical brain fields: provider, model, backend, config
(base_url, api_key, ...), temperature / token limits, and
native_tool_use overrides when present.
Supported paths include OpenAI-compatible HTTP APIs and Anthropic (and related backends the daemon ships). Prefer working examples in the repo over outdated provider shortcut tables.