Show HN: Demarkus – De-centralized Markup for Us:memory for AI agents and humans (github.com)

Demarkus has launched a new protocol that redefines how humans and AI agents interact with information, centered around a decentralized markdown format. This innovation addresses key issues such as privacy, security, and simplicity by eliminating complex rendering processes and tracking mechanisms. Content is transmitted via QUIC to ensure high performance, while the protocol maintains user anonymity and incorporates mandatory encryption. As it stands, the protocol supports five core commands—FETCH, LIST, VERSIONS, PUBLISH, APPEND, and ARCHIVE—enabling users and agents to easily manage and retrieve documents. For the AI/ML community, Demarkus represents a significant shift towards collaborative frameworks where both humans and AI can share, reflect, and store knowledge efficiently. Notably, it facilitates persistent memory for agents, allowing them to track changes and learn from previous interactions. The protocol is designed for seamless integration with tools like Claude Desktop, enhancing the development of AI applications that require maintaining context over time. By encouraging decentralized and open-source principles, Demarkus aims to create a more user-centric internet, providing a blueprint for future AI-driven ecosystems.