notmemory: open-source library to track and undo AI agent memory
A new Python library called notmemory lets you see exactly what an AI agent remembers and roll back any mistakes. Most existing memory tools for agents are a black box — notmemory logs every change and supports rollbacks. It's at v0.1.0 and looking for contributors.
AI agents store information between tasks or conversations so they can refer back to it later. The problem is that most memory systems don't show you what was stored, when it changed, or let you fix a bad entry without a lot of manual work.
notmemory tackles this by keeping a full change log every time the agent's memory is updated, making it auditable — you can inspect the history at any time. If something goes wrong, you can roll back to an earlier state. It's posted in the LangChain community, suggesting it's designed to work alongside LangChain-based agents. At version 0.1.0, it's very early-stage, so expect rough edges, and the author is actively looking for people to help build it out.
Key points
Quick term guide
- memory
- A ChatGPT feature that lets it use details from past chats in future chats.
- AI agent
- An AI program that can inspect information and suggest what to do next.
- agents
- AI helpers that follow your instructions and make changes for you.
- rollback
- A plan or action to return software to a previous working version.
- AI agents
- AI agents are AI tools that can carry out steps toward a goal, not just answer once.
- auditable
- Designed so you can inspect a full record of what changed, when, and why
- LangChain
- A popular open-source framework for building AI agents and applications that chain together language model calls.
- production
- The live version of a service that real users use.