Package uno.anahata.asi.agi.provider
package uno.anahata.asi.agi.provider
Provides the provider-agnostic LLM connectivity and generation interfaces.
This package defines the core adapter contracts for connecting external AI models
to the Anahata ecosystem. It orchestrates provider-level lifecycles via
AbstractAiProvider and manages active inference
runs, payload synthesis, streaming observers, and API retry handling through
AbstractModel.
- Author:
- anahata
-
ClassDescriptionThe abstract base class for all AI model providers, now with model caching.The abstract base class for a specific AI model (e.g., "gemini-1.5-pro-latest").Generic runtime exception for AiProvider implementations.Exception thrown when an AI model generation call is interrupted, either by the user or by the system.A standardized, model-agnostic enum representing the reason why a model stopped generating content.A simple, immutable record that encapsulates all the information required for a model to generate content: the configuration and the conversation history.The definitive, model-agnostic configuration object for a single API request.Response<T extends AbstractModelMessage>A standardized, model-agnostic response received from an AI model provider.Represents the modalities that an AI model can output in its generative response.A specialized exception indicating that an API call failed but is potentially retryable.Represents a server-side tool provided by an AI model (e.g., Google Search, Code Execution).StreamObserver<R extends Response<?>>A simple interface for observing a stream of values, typically used for asynchronous token streaming from an AI model.Standardized, model-agnostic enum representing the level of "thinking" or reasoning tokens that the model should generate.Defines the supported tokenizer strategies for the Anahata ASI.