Skip to content

MemoryScope

Add long-term memory to an LLM chatbot

Visit official site

What MemoryScope is

The framework is described as a long-term memory system for LLM chatbots, built on a memory database and a work library. It stores, retrieves and consolidates memory fragments so a chatbot can recall user habits and preferences.

What you can do with it

  • Store recorded memory fragments in a built-in vector database
  • Split long-term memory into more than 20 work units
  • Retrieve semantic and time-related fragments when a query arrives
  • Consolidate observations and refresh insights through reflection

Application scenarios

  • Give a personal assistant chatbot memory for schedules and preferences
  • Help a companion app follow emotional changes and offer support
  • Add long-term memory to a support chatbot built with AutoGen

Getting started

  • Open the project GitHub page for details and the installation guide
  • Follow the installation document to install the framework and its dependencies
  • Read the quick start document to learn the basic workflow
  • Work through the CLI or API to run retrieval and consolidation

Pros & cons

✓ What we like

  • Memory retrieval returns both semantic matches and time-related fragments
  • Customisation is open through model APIs and existing agent frameworks

! What to watch out for

  • The default memory store relies on ElasticSearch
  • Users need to install the framework and its dependencies from the documentation

FAQ

What does memory retrieval return?

The material states it returns semantically related fragments and/or time-related fragments for a query.

Which frameworks can it work with?

AutoGen and AgentScope are named, along with OpenAI and DashScope model APIs.

Last reviewed: 2026-11-15

More AI coding tools tools

View all →

How we review