Module provider 

LLM provider abstraction layer.

This module defines the Provider trait and shared request/response types used by all backends (Anthropic/OpenAI/Gemini/etc).

Providers are responsible for:

• Translating crate::model::Message history into provider-specific HTTP requests.
• Emitting StreamEvent values as SSE/HTTP chunks arrive.
• Advertising tool schemas to the model (so it can call crate::tools by name).

Re-exports

pub use crate::model::StreamEvent;

Structs

Context

Inputs to a single completion request.

Model

A model definition loaded from the models registry.

ModelCost

Model pricing per million tokens.

StreamOptions

Options that control streaming completion behavior.

ThinkingBudgets

Custom thinking token budgets per level.

ToolDef

A tool definition exposed to the model.

Enums

Api

Known API types.

CacheRetention

Cache retention policy.

InputType

Input types supported by a model.

KnownProvider

Known providers.

Traits

Provider

An LLM backend capable of streaming assistant output (and tool calls).