Memory v0.4 Release

What's New

Memory v0.4 introduces a fully documented API for managing persistent AI memory — enabling developers to create, update, analyze, and track memory objects with synced database and vector-store operations.

Memory CRUD Operations

• Create: Store structured memory objects with metadata, timestamps, and source attribution.

• Read: Query memories by ID, semantic similarity, time range, or custom filters.

• Update: Modify existing memories with version tracking and conflict resolution.

• Delete: Remove memories with cascading cleanup across vector stores and databases.

Memory Analysis

• Track how memories evolve over time with full version history.

• Analyze memory density, coverage gaps, and retrieval patterns.

• Identify stale or conflicting memories automatically.

• Export memory analytics for compliance and auditing.

Synced Operations

Every memory write is now atomic across both the relational database and vector store:

• No more inconsistencies between stored text and embeddings.

• Automatic rollback on partial failures.

• Real-time sync status visible in the dashboard.

Developer Experience

• Full API documentation with interactive playground.

• Python and TypeScript SDK examples for every endpoint.

• Webhook support for memory lifecycle events (created, updated, deleted).

• Rate limits: 1,000 memory operations/minute on Pro tier.