AWS Bedrock AgentCore Memory Adds Strictly Consistent Metadata
Amazon Bedrock AgentCore Memory now allows developers to directly attach metadata to long-term memory records, ensuring data integrity and enabling precise filtering. This feature improves data organization, facilitates compliance, and supports multi-tenant architectures by guaranteeing metadata values are preserved without LLM inference. Strictly consistent metadata is available in all AWS Regions where AgentCore Memory is supported.
- →Strictly consistent metadata for Bedrock AgentCore long-term memory
- →Improved data isolation and grouping with strictly consistent metadata
- →Configuration and availability of strictly consistent metadata
Features (1) ›
- Strictly consistent metadata for Bedrock AgentCore long-term memory
Developers can now attach metadata directly to long-term memory records, ensuring values are preserved unchanged and are not subject to LLM inference. This feature supports data organization, filtering, and routing for retrieval. Events with the same strictly consistent metadata values are grouped together for extraction and consolidation.
Enhancements (1) ›
- Improved data isolation and grouping with strictly consistent metadata
Strictly consistent metadata enables better data isolation for use cases like department-scoped retrieval, compliance boundaries, and multi-tenant memory. Events with different values for strictly consistent keys are never merged, even if semantically similar.
Notes (1) ›
- Configuration and availability of strictly consistent metadata
Up to three strictly consistent STRING keys can be configured per strategy, supporting semantic, user preference, and episodic strategies. Both LLM-inferred and strictly consistent keys can coexist on the same memory resource. The feature is available in all AWS Regions where AgentCore Memory is supported.
https://aws.amazon.com/about-aws/whats-new/2026/05/agentcore-memory-scmetadata