agents-features-longterm-memory/ai.koog.agents.longtermmemory.feature/LongTermMemory

LongTermMemory

class LongTermMemory(retrievalSettings: RetrievalSettings? = null, ingestionSettings: IngestionSettings? = null)(source)

Memory feature that incorporates persistent storage of memory records (documents) in vector databases.

This feature provides two main capabilities that can be configured independently:

Retrieval (RAG): Retrieves relevant context from the memory record repository using a configured search request builder and inserts it into the prompt before sending to the LLM. Configured via retrievalSettings. When null, no retrieval/augmentation is performed.
Ingestion: Saves messages to the memory record repository for future retrieval. Configured via ingestionSettings. When null, no messages are persisted.

See also

RetrievalSettings

IngestionSettings

Constructors

constructor(retrievalSettings: RetrievalSettings? = null, ingestionSettings: IngestionSettings? = null)

Types

class Config : FeatureConfig

Configuration for the LongTermMemory feature.

object Feature : AIAgentGraphFeature<LongTermMemory.Config, LongTermMemory> , AIAgentFunctionalFeature<LongTermMemory.Config, LongTermMemory> , AIAgentPlannerFeature<LongTermMemory.Config, LongTermMemory>

Companion object implementing agent feature, handling LongTermMemory creation and installation.

IngestionSettingsBuilder

class IngestionSettingsBuilder

Builder for IngestionSettings used in the Config.ingestion DSL block.

RetrievalSettingsBuilder

class RetrievalSettingsBuilder

Builder for RetrievalSettings used in the Config.retrieval DSL block.

Properties

ingestionStorage

val ingestionStorage: IngestionStorage?

Property getter for IngestionStorage for usage inside strategy nodes

retrievalStorage

val retrievalStorage: RetrievalStorage?

Property getter for RetrievalStorage for usage inside strategy nodes