Korpus tool addresses AI agent memory loss by creating shared context space

A developer created Korpus, an MCP-first tool designed to solve AI agents' persistent memory problem across sessions. The system provides shared memory storage where both humans and multiple AI agents can read and write project context in plain markdown documents. Key design features include optimistic concurrency control to prevent overwrites, change logs documenting all modifications, and structured folder organization with explicit paths. The tool allows agents to maintain continuity between sessions instead of starting fresh each time with no project memory.