# Memory OS — Hermes Agent Memory Operating System

A seven-layer local memory system for Hermes Agent that stores structured facts, session history, and a self-curating wiki, and injects the right context back into conversations automatically.

Memory OS turns Hermes Agent into a long-term collaborator by adding seven memory layers, automatic context injection, structured facts with trust scoring, a self-curating wiki pipeline, and semantic search across past conversations. It aims to let the agent recall a user's projects, decisions, and reasoning between sessions instead of starting over each time.

The memory infrastructure runs entirely on the user's own machine using Qdrant, SQLite, and an Icarus plugin, and works with any LLM provider, including OpenRouter, OpenAI, Anthropic, Ollama, or local models, with no separate memory subscription.

Installation is a single setup script that configures the Docker services, databases, and plugin; the project also ships smoke tests and ingestion tests and is under active community development.

Website: https://github.com/ClaudioDrews/memory-os