Rememb

Gives your AI agent a persistent memory layer backed by local JSON files, no server or API keys needed. Exposes MCP tools for reading, writing, and semantically searching structured entries across sections like project context, actions taken, system architecture, and user preferences. Ships with a local web UI for browsing and editing memory, runs over stdio or SSE transport, and uses SentenceTransformers for offline semantic search with configurable model unloading. The agent automatically loads context at session start and writes back learnings, so you stop re-explaining your stack and preferences every conversation. Reach for this when you want stateful sessions without cloud dependencies or when building agents that need to remember project-specific details across runs.