prompt-executor-ollama-client/ai.koog.prompt.executor.ollama.client/OllamaClient

OllamaClient

class OllamaClient @JvmOverloads constructor(httpClient: KoogHttpClient, clock: KoogClock = KoogClock.System, contextWindowStrategy: ContextWindowStrategy = ContextWindowStrategy.Companion.None, toolDescriptorConverter: ToolDescriptorSchemaGenerator = OllamaToolDescriptorSchemaGenerator()) : LLMClient(source)

Client for interacting with the Ollama API with comprehensive model support.

Implements:

LLMClient for executing prompts and streaming responses.
LLMEmbeddingProvider for generating embeddings from input text.

Parameters

httpClient

A preconfigured Koog HTTP client used for API calls. Must have the Ollama base URL and timeouts already embedded. To create a client with standard defaults, use the secondary constructor that accepts a base URL and a KoogHttpClient.Factory.

clock

Clock instance used for tracking response metadata timestamps.

contextWindowStrategy

The ContextWindowStrategy to use for computing context window lengths. Defaults to ContextWindowStrategy.None.

toolDescriptorConverter

Generates JSON schemas for tool descriptors sent to the model.

Constructors

OllamaClient

@JvmOverloads

constructor(httpClient: KoogHttpClient, clock: KoogClock = KoogClock.System, contextWindowStrategy: ContextWindowStrategy = ContextWindowStrategy.Companion.None, toolDescriptorConverter: ToolDescriptorSchemaGenerator = OllamaToolDescriptorSchemaGenerator())

@JvmOverloads

constructor(httpClientFactory: KoogHttpClient.Factory, baseUrl: String = DEFAULT_BASE_URL, headers: Map<String, String> = emptyMap(), queryParameters: Map<String, String> = emptyMap(), timeoutConfig: ConnectionTimeoutConfig = ConnectionTimeoutConfig(), clock: KoogClock = KoogClock.System, contextWindowStrategy: ContextWindowStrategy = ContextWindowStrategy.Companion.None, toolDescriptorConverter: ToolDescriptorSchemaGenerator = OllamaToolDescriptorSchemaGenerator())

Secondary constructor that builds the underlying KoogHttpClient from a KoogHttpClient.Factory.

Types

Companion

object Companion

Properties

clientName

open val clientName: String

Functions

open override fun close()

embed

open suspend override fun embed(text: String, model: LLModel): List<Double>

Embeds the given text using the Ollama model.

open suspend override fun embed(inputs: List<String>, model: LLModel): List<List<Double>>

Embeds the given inputs using the Ollama embeddings API.

execute

open suspend override fun execute(prompt: Prompt, model: LLModel, tools: List<ToolDescriptor>): Message.Assistant

executeMultipleChoices

open suspend fun executeMultipleChoices(prompt: Prompt, model: LLModel, tools: List<ToolDescriptor>): LLMChoice

executeStreaming

open fun executeStreaming(prompt: Prompt, model: LLModel): Flow<StreamFrame>

open override fun executeStreaming(prompt: Prompt, model: LLModel, tools: List<ToolDescriptor>): Flow<StreamFrame>

getBasicJsonSchemaGenerator

open fun getBasicJsonSchemaGenerator(): BasicJsonSchemaGenerator

getModelOrNull

suspend fun getModelOrNull(name: String, pullIfMissing: Boolean = false): OllamaModelCard?

Returns a model card by its model name, on null if no such model exists on the server.

getModels

suspend fun getModels(): List<OllamaModelCard>

Returns the model cards for all the available models on the server.

getStandardJsonSchemaGenerator

open fun getStandardJsonSchemaGenerator(): StandardJsonSchemaGenerator

llmProvider

open override fun llmProvider(): LLMProvider

Provides the type of Language Learning Model (LLM) provider used by the client.

models

open suspend fun models(): List<LLModel>

moderate

open suspend override fun moderate(prompt: Prompt, model: LLModel): ModerationResult