ClaudioDrews/memory-os — reverse-engineered prompt

Build me a local memory system for Hermes Agent so it stops forgetting between chats.

I want it to remember my projects, preferences, decisions, useful facts, and past conversations, then bring back only the relevant context when I’m talking to it later. It should run on my own machine, work with whatever LLM provider Hermes supports, and not depend on a paid cloud memory service.

Please include persistent conversation search, structured facts with some kind of confidence or trust score, semantic recall across old sessions, and a simple wiki or knowledge vault that can organize important concepts over time. It should avoid dumping too much context into the model, dedupe repeated memories, skip useless small talk, and keep things token efficient.

Use Python and local infrastructure like Docker where needed. Make it easy to install, configure, and run with Hermes. Add clear setup docs and sensible defaults so a lightly technical user can clone it and get working memory without having to understand every layer first.