Skip to main content

OllamaClient

oxide_agent::client

Trait OllamaClient 

Source 
pub trait OllamaClient: Send + Sync {
    // Required methods
    fn generate<'life0, 'async_trait>(
        &'life0 self,
        req: GenerateRequest,
    ) -> Pin<Box<dyn Future<Output = Result<GenerateResponse, OxideError>> + Send + 'async_trait>>
       where Self: 'async_trait,
             'life0: 'async_trait;
    fn chat<'life0, 'async_trait>(
        &'life0 self,
        req: ChatRequest,
    ) -> Pin<Box<dyn Future<Output = Result<ChatResponse, OxideError>> + Send + 'async_trait>>
       where Self: 'async_trait,
             'life0: 'async_trait;
    fn embed<'life0, 'async_trait>(
        &'life0 self,
        req: EmbedRequest,
    ) -> Pin<Box<dyn Future<Output = Result<EmbedResponse, OxideError>> + Send + 'async_trait>>
       where Self: 'async_trait,
             'life0: 'async_trait;
    fn list_models<'life0, 'async_trait>(
        &'life0 self,
    ) -> Pin<Box<dyn Future<Output = Result<ListModelsResponse, OxideError>> + Send + 'async_trait>>
       where Self: 'async_trait,
             'life0: 'async_trait;
    fn stream_generate(
        &self,
        req: GenerateRequest,
    ) -> BoxStream<GenerateResponse>;
    fn stream_chat(&self, req: ChatRequest) -> BoxStream<ChatResponse>;
}
Expand description

The type-safe bridge to any Ollama-compatible backend.

Decoupling agent logic from transport via a trait means you can:

  • Swap in a real HTTP client for production.
  • Swap in a MockOllamaClient for unit testing — no running server needed.
  • Wrap the client with middleware (retry, logging, rate-limiting) transparently.

Required Methods§

Source

fn generate<'life0, 'async_trait>( &'life0 self, req: GenerateRequest, ) -> Pin<Box<dyn Future<Output = Result<GenerateResponse, OxideError>> + Send + 'async_trait>>
where Self: 'async_trait, 'life0: 'async_trait,

Single-turn text completion (/api/generate).

Source

fn chat<'life0, 'async_trait>( &'life0 self, req: ChatRequest, ) -> Pin<Box<dyn Future<Output = Result<ChatResponse, OxideError>> + Send + 'async_trait>>
where Self: 'async_trait, 'life0: 'async_trait,

Multi-turn chat completion (/api/chat), including tool/function calling.

Source

fn embed<'life0, 'async_trait>( &'life0 self, req: EmbedRequest, ) -> Pin<Box<dyn Future<Output = Result<EmbedResponse, OxideError>> + Send + 'async_trait>>
where Self: 'async_trait, 'life0: 'async_trait,

Produce dense vector embeddings for one or more texts (/api/embed).

Source

fn list_models<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = Result<ListModelsResponse, OxideError>> + Send + 'async_trait>>
where Self: 'async_trait, 'life0: 'async_trait,

List models available on the Ollama server (/api/tags).

Source

fn stream_generate(&self, req: GenerateRequest) -> BoxStream<GenerateResponse>

Stream a generate response token-by-token.

Ollama sends newline-delimited JSON (NDJSON) when stream: true. This method decodes each line into a GenerateResponse without buffering the entire response body, keeping memory overhead minimal regardless of output length.

Source

fn stream_chat(&self, req: ChatRequest) -> BoxStream<ChatResponse>

Stream a chat response token-by-token.

Implementors§