VertexAiMemoryBankService
A MemoryService backed by Vertex AI Agent Engine Memory Bank.
This is a Kotlin port of the Python ADK VertexAiMemoryBankService. Memories are scoped by {app_name, user_id}; addSessionToMemory and addEventsToMemory send event content for server-side memory generation, addMemory writes explicit memories (directly, or via generation when consolidation is requested), and searchMemory retrieves memories by semantic similarity.
It follows the existing Vertex service ports: a thin VertexAiMemoryBankClient over a shared GoogleApiClient handles the REST calls, and @Serializable wire DTOs live in the dto sub-package.
Types
Fluent builder for VertexAiMemoryBankService, provided primarily for Java callers. Any property left unset falls back to the same default as the constructor.
Functions
Adds events for memory generation. Uses memories:ingestEvents by default; if customMetadata carries a key only memories:generate accepts (e.g. ttl, metadata, disable_consolidation), the generate path is used. See GENERATE_ONLY_KEYS for the keys.
Writes memories as explicit facts, or consolidates them server-side via generate when enable_consolidation is set in customMetadata (the generate config keys in customMetadata then apply). Only each entry's text is persisted; per-entry author, timestamp, and customMetadata are not written (Python 2.0 maps them to revision labels and merged metadata).
Adds a session to the memory service.
Searches for sessions that match the query asynchronously.