Expand description
Canonical model invocation and continuation contract. No Agent or transport implementation.
Re-exports§
pub use service::Image;
Modules§
Structs§
- Completion
- Provider-neutral completion returned to Runtime.
- Completion
Request - One logical LLM invocation.
- Continuation
- Provider-issued replay data attached to exactly one completed model message. Adapters own the format and compatibility key. Callers must not synthesize or edit payloads.
- Message
- Provider-neutral conversation message.
- Model
Constraints - Capabilities required by an invocation, without naming a provider or model.
- Model
Mode - Logical model mode selected by the product, independent of provider and model identifiers.
- Model
Profile - Provider-neutral limits for one frozen logical model route.
- Token
Usage - Normalized token accounting reported by an implementation.
- Tool
Call - Tool call returned by an LLM.
- Tool
Definition - A callable function exposed to the model. Execution policy belongs to the caller.
- UseCase
- Logical model use case resolved by the deployment’s LLM implementation.
Enums§
- Content
Part - One typed part of a model message.
- Finish
Reason - Why a model invocation stopped.
- Message
Role - Role of a message in a model conversation.
Constants§
- MESSAGE_
SCHEMA - Offline normative schema for persisted model messages.