What problem does it solve? AI applications and agents lose context between sessions, forcing users to repeat themselves and preventing personalization. This Skill provides the architectural rules, API patterns, and integration playbooks needed to add persistent, tenant-isolated long-term memory to any LLM application using Supermemory (supermemory.ai), covering ingestion, retrieval, SDK wiring, self-hosting, and memory lifecycle management. ## Core Features & Use Cases - Ingestion & Data Modeling: Ingest documents, files, and conversations with deterministic containerTag scoping, customId deduplication with diff billing, and async dreaming controls for read-your-writes timing. - Retrieval & Memory Lifecycle: Query extracted facts, raw chunks, or both via /v4/search modes, use per-container profiles for low-latency stable facts, and manage versioned updates plus soft forgetting with forget-matching dry runs. - SDK & Framework Integration: Wire the TypeScript/Python SDKs and @supermemory/tools v2 wrappers into Vercel AI SDK, OpenAI, Mastra, Claude's memory tool, LangChain, CrewAI, and Microsoft Agent Framework, plus the hosted Memory MCP and self-hosted binary on localhost:6767. - Use Case: You are building a support chatbot that must remember each customer's preferences across sessions. Use this Skill to design a per-user containerTag scheme, wrap your model with withSupermemory in full mode, and verify the deployment with the read-only API and search probe scripts. ## Quick Start Ask the AI to add Supermemory persistent memory to your agent with a per-user containerTag, using the withSupermemory wrapper and scoped API keys.
