OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Memory & RAG

Nocturne Memory

Nocturne Memory is an MIT-licensed Python 3.10+ MCP server that stores graph-like long-term memory for AI agents across sessions, models and tools. It is designed as a self-hosted infrastructure layer, with SQLite and PostgreSQL identified as storage options and a dashboard for exploring and reviewing memory changes.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Nocturne Memory — OpenViST project profile in Agent Memory & RAG.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Nocturne Memory when an MCP-based agent needs durable context that can remain available while the surrounding model or client changes. The public demo is useful for read-only exploration, while full read-write operation is described through a self-hosted deployment.

Who it is for

  • Developers building MCP-compatible AI agents
  • Teams or individuals who need agent context to persist across sessions and model changes
  • Builders seeking a self-hosted memory service with visual review and rollback workflows

Documented uses

  • Store long-term agent memories outside an individual model or client
  • Retrieve selected memories across new conversations
  • Share one memory store across different MCP-compatible clients and models
  • Inspect, edit, review and roll back memory changes
  • Provide separate namespaces for multiple agent identities
Before you choose

Requirements & limits

The project is public and MIT-licensed. A public demo endpoint is described as read-only; full read-write capability requires deploying an instance. The repository lists Linux, macOS and Windows alongside CLI and server use.

  • MCP support is central: the memory layer is intended to connect through MCP-compatible clients.
  • Python 3.10+ and Node.js are listed prerequisites for self-hosted setup.
  • SQLite and PostgreSQL are the identified storage choices.
  • The public demo exposes read_memory and search_memory only, so writable workflows require self-hosting.
  • The dashboard supports browsing, metadata inspection, visual diffs and rollback-oriented review.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up a self-hosted Nocturne Memory server

Prepare the documented MCP memory service and its dashboard prerequisites.

Before you start

  • Python 3.10 or newer
  • Node.js
  • An MCP-compatible client for connecting to the server

Steps

  1. Clone the Nocturne Memory repository.

  2. Enter the project directory and follow the repository’s installation instructions.

  3. Start the service so the dashboard frontend can build on first startup.

  4. Connect an MCP-compatible client using the documented stdio or SSE configuration for the deployed instance.

Expected outcome

A self-hosted Nocturne Memory MCP server prepared for use as a persistent memory layer.

Guide limitations

  • The available instructions document setup prerequisites and connection modes, but do not establish a successful deployment or client compatibility guarantee.
  • Writable use requires a self-hosted instance; the public demo is documented as read-only.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
3 October 2026
Original website title
Dataojitori/nocturne_memory · Official homepage title · retained in its source language
How we review sourcesSuggest a correction