mem-skill

mem-skill provides AI agents with persistent memory by automatically retrieving past experiences and proactively recording successful solutions to a private knowledge base. This transforms your AI into a second brain that learns preferences over time.

An AI knowledge base captures best practices by asking permission to proactively record successful solutions and workflows. It detects topic switches to intelligently reload relevant context, ensuring past patterns are automatically available for future tasks.

Yes, you can use QMD semantic search for enhanced memory retrieval. The system supports default JSON and Markdown indexing, while offering optional QMD semantic search to improve how past experiences and solutions are loaded into context.

The best way to initialize persistent memory in a workspace is to run the `/mem-skill init` command. This sets up the self-evolving knowledge accumulation system, enabling automatic retrieval and proactive recording of your solutions.

Your AI forgets coding preferences because it lacks a persistent memory system. Without a second brain to automatically retrieve past solutions and proactively record best practices, it cannot load previous context when you switch topics or start new sessions.