What problem does it solve? Building AI agents that remember users across conversations requires managing two memory tiers — raw session history and semantically searchable long-term facts — plus the async promotion pipeline between them. This Skill provides the correct patterns for provisioning, authenticating, writing, and searching Redis Agent Memory on Redis Cloud without common pitfalls like duplicate records, lost turn order, or flaky tests. ## Core Features & Use Cases - Session Memory Management: Append conversation events with correct session IDs, roles, and tz-aware UTC timestamps, and retrieve sessions or individual events efficiently. - Long-Term Memory Operations: Bulk-create up to 100 memories with idempotent IDs, organize them with ownerId, namespace, topics, and memoryType, and search semantically with structured filters and pagination. - Background Promotion Handling: Understand the 5-minute deduplication window and eventual-consistency model so tests poll instead of asserting synchronously. - Use Case: You are wiring a support chatbot to remember each customer's preferences. Use this Skill to append every turn as a session event, let promotion extract durable facts, and recall them later with an ownerId-scoped semantic search. ## Quick Start Ask the AI to set up the Redis Agent Memory SDK and write code that appends a session event and searches long-term memory for a user.
