llm-provider

To add a new LLM provider, you implement the LLMProvider interface in a new file under src/llm, ensuring both call() and stream() methods are wired through the config, factory, and tests for seamless integration.

The standard structure for integrating a new model API backend involves creating a provider file at src/llm/{provider-name}.ts that implements the LLMProvider interface, then routing it through a unified config and factory system.

You validate LLM provider behavior across environments by wiring the new provider through config, factory, and tests that enforce standard API, model validation, and streaming semantics following the described interface.

Yes, you can extend model support by creating a new provider file and wiring it through the factory routing system, which maintains a unified interface and enables seamless integration without breaking existing providers.

The best way to implement streaming responses for a new LLM API is to implement the stream() method within the LLMProvider interface, ensuring the provider config and factory routing support streaming semantics.