🧩 Compartment: Encrypted Offline Memory for AI Agents

Compartment provides a localized solution for adding long-term memory to AI agents. Operating as an MCP (Model Context Protocol) server, it focuses on being fully offline and encrypted, ensuring privacy when agents read and write contextual data. The interesting part is that it does not just handle backend storage and retrieval; it also includes a GUI with a memory map. This allows users to visually inspect and manage exactly what Claude Desktop or other MCP-compatible agents have remembered about them or their projects. For users who want their local AI assistants to maintain persistent context but are hesitant to send personal data to cloud-based vector databases, this out-of-the-box, encryption-focused local memory component is a highly practical addition.