agents-features-longterm-memory/ai.koog.agents.longtermmemory.retrieval

Package-level declarations

Types

data class KeywordSearchRequest(val query: String, val limit: Int = 10, val similarityThreshold: Double = 0.0, val filterExpression: String? = null) : SearchRequest

Search request for keyword/full-text search. Uses traditional text matching instead of vector similarity.

class KeywordSearchStrategy(val topK: Int = 10, val similarityThreshold: Double = 0.0, val filterExpression: String? = null) : SearchStrategy

Keyword search mode using full-text/lexical matching.

data class RetrievalSettings(val storage: RetrievalStorage, val searchStrategy: SearchStrategy? = null, val promptAugmenter: PromptAugmenter = SystemPromptAugmenter(), val namespace: String? = null)

Settings controlling how memory records are retrieved and injected into prompts (RAG).

An interface for retrieving (searching) memory records from a storage. An implementation of this interface is responsible for embedding.

Base interface for search requests.

data class SearchResult(val record: MemoryRecord, val similarity: Double)

Represents a result of a SearchRequest.

Search strategy for creating search requests during prompt augmentation.

Intermediate builder that lets callers select a SearchStrategy implementation.

data class SimilaritySearchRequest(val query: String, val limit: Int = 10, val similarityThreshold: Double = 0.0, val filterExpression: String? = null) : SearchRequest

Search request for pure vector similarity search using text query. The text will be embedded and used for vector similarity search.

class SimilaritySearchStrategy(val topK: Int = 10, val similarityThreshold: Double = 0.0, val filterExpression: String? = null) : SearchStrategy

Similarity search mode using vector embeddings for semantic search.