Anamnesion is a persistent memory server for LM Studio and other Model Context Protocol (MCP) compatible AI clients.
It provides long-term memory, knowledge graph management, sequential reasoning workflows, and full-text retrieval that survive session restarts.
The goal is to give language models structured, persistent context so they can operate consistently across sessions.
Clone the repository:
git clone https://gitlab.com/Rexodus/anamnesion-memory-server.gitInstall dependencies:
pip install -r requirements.txtStart the server:
python memory.pyLanguage models normally lose context when a session ends. Anamnesion adds an external memory layer that enables persistent context, structured reasoning, and long-term information storage.
It allows AI systems to store observations, maintain knowledge graphs, and continue reasoning workflows across multiple sessions.
Track multi-step reasoning sessions that persist across restarts, allowing structured problem-solving over time.
Store entities, relationships, and observations in a persistent graph structure backed by SQLite.
Full-text search, metadata filtering, importance scoring, and structured retrieval across stored memory objects.
This project is experimental and intended for local, human-supervised environments. It is not designed for production or public internet deployment.
Anamnesion is a local persistent memory system for AI assistants, designed to extend context, structure reasoning, and enable long-term collaboration.