agents-features-memory/ai.koog.agents.memory.feature.nodes

Package-level declarations

Functions

Link copied to clipboard
inline fun <T> AIAgentSubgraphBuilderBase<*, *>.nodeLoadAllFactsFromMemory(name: String? = null, subjects: List<MemorySubject> = MemorySubject.registeredSubjects, scopes: List<MemoryScopeType> = MemoryScopeType.entries): AIAgentNodeDelegate<T, T>

Node that loads all facts about the subject from memory for a given concept

Link copied to clipboard
inline fun <T> AIAgentSubgraphBuilderBase<*, *>.nodeLoadFromMemory(name: String? = null, concept: Concept, subject: MemorySubject, scope: MemoryScopeType = MemoryScopeType.AGENT): AIAgentNodeDelegate<T, T>
inline fun <T> AIAgentSubgraphBuilderBase<*, *>.nodeLoadFromMemory(name: String? = null, concepts: List<Concept>, subject: MemorySubject, scope: MemoryScopeType = MemoryScopeType.AGENT): AIAgentNodeDelegate<T, T>
inline fun <T> AIAgentSubgraphBuilderBase<*, *>.nodeLoadFromMemory(name: String? = null, concepts: List<Concept>, subjects: List<MemorySubject> = MemorySubject.registeredSubjects, scopes: List<MemoryScopeType> = MemoryScopeType.entries): AIAgentNodeDelegate<T, T>

Node that loads facts from memory for a given concept

Link copied to clipboard
inline fun <T> AIAgentSubgraphBuilderBase<*, *>.nodeSaveToMemory(name: String? = null, concept: Concept, subject: MemorySubject, scope: MemoryScopeType, retrievalModel: LLModel? = null): AIAgentNodeDelegate<T, T>
inline fun <T> AIAgentSubgraphBuilderBase<*, *>.nodeSaveToMemory(name: String? = null, subject: MemorySubject, scope: MemoryScopeType, concepts: List<Concept>, retrievalModel: LLModel? = null): AIAgentNodeDelegate<T, T>

Node that saves a fact to memory

Link copied to clipboard
inline fun <T> AIAgentSubgraphBuilderBase<*, *>.nodeSaveToMemoryAutoDetectFacts(name: String? = null, scopes: List<MemoryScopeType> = listOf(MemoryScopeType.AGENT), subjects: List<MemorySubject> = MemorySubject.registeredSubjects, retrievalModel: LLModel? = null): AIAgentNodeDelegate<T, T>

Node that automatically detects and extracts facts from the chat history and saves them to memory. It uses LLM to identify concepts about user, organization, project, etc.

Link copied to clipboard
fun parseFactsFromResponse(content: String): List<Pair<MemorySubject, Fact>>

Parsing facts from response.