elite-longterm-memory

To maintain durable AI agent context across sessions, use a tiered memory system that orchestrates data across hot RAM for current tasks, LanceDB warm storage for semantic auto-recall, and Git-Notes cold storage for permanent decision logs.

Enable semantic search for an AI memory system by configuring LanceDB as a warm storage layer, which allows the agent to perform semantic auto-recall of relevant past interactions and context during ongoing runtime sessions.

Yes, you need a local environment with LanceDB configured for semantic warm storage, while optional OpenAI integration and Git-Notes cold storage provide extended capabilities for cross-device synchronization and audit trails.

Tiered AI agent memory works by keeping current tasks in hot RAM (SESSION-STATE.md), using LanceDB for semantic warm storage, and writing permanent decision logs to a Git-Notes cold store to create an audit-friendly knowledge graph.

Yes, you can synchronize AI agent memory across multiple devices by utilizing the optional cloud backup feature, which works with the daily logs and MEMORY.md archive to ensure persistent context is available cross-device.

Use a permanent cold storage layer, like the Git-Notes knowledge graph, when your AI agents require long-running sessions, multi-agent collaboration, or audit-friendly decision tracking that must endure beyond the current runtime session.