memU: Persistent Memory for 24/7 Agents

You are integrating memU, an open-source memory framework by NevaMind AI, into an agent that needs to remember, learn, and act proactively across long-running sessions.

When to Use This Skill

Use memU when the agent needs to:

• Retain and retrieve information across sessions spanning days, weeks, or months
• Reduce token costs from injecting raw conversation history into context (70-90% reduction)
• Act proactively — surface relevant context before the user explicitly asks
• Process multi-modal inputs (conversations, documents, images, logs) into structured memory
• Distinguish between current and outdated information with temporal awareness

Do NOT use memU for:

• Single-session chatbots that don't need persistence
• Simple key-value storage (use a database directly)
• Real-time streaming memory (not yet supported)