memU
Open-source agentic memory framework for 24/7 proactive AI agents with file-system memory, intention prediction, and lower token costs.
🛡️ AgentReady threat assessment
MAESTRO 7-layer threat model + OWASP AIVSS risk score for memU, derived from its capabilities.
These scores are auto-generated from public information (the agent's own listing, docs, and repository) using the canonical OWASP AIVSS formula and the MAESTRO framework — an estimate for guidance, not a penetration test, audit, or certification. See the scoring methodology. Are you the vendor? Factual corrections are free.
Overview
memU is an open-source memory infrastructure for LLM applications and AI agents, designed for long-running, always-on assistants that need persistent, evolving memory. It organizes memories like a file system (categories as folders, memory items as files, cross-links as symlinks) and supports proactive behavior such as capturing user intent, predicting next steps, and injecting relevant memory into the agent’s context. memU aims to reduce token spend for continuous agents by caching insights and avoiding redundant LLM calls, and it includes companion components like a backend service (memU-server) and a web UI (memU-ui). In addition to the open-source framework, memU offers hosted APIs (Memory API / Response API) with usage-based pricing and a “start free” path.
Key features
- agentic memory
- long-term memory
- proactive memory
- user intention prediction
- file-system memory
- memory graph
- context injection
- rag retrieval
- multimodal inputs
- always-on agents
Use cases
- Adding long-term, structured memory to AI companions and assistants that run continuously.
- Reducing LLM context/token costs for always-on agents by caching and recalling distilled insights.
- Capturing user goals and preferences automatically and using them to act proactively.
- Building agent memory stores with multiple retrieval strategies (file-based, RAG-style, and direct reading).