Repolode
Submit
NirDiamant

NirDiamant/Agent_Memory_Techniques

A learning repository of 30 runnable Jupyter notebooks covering agent memory techniques for LLM agents, from conversation buffers to Mem0, Letta, Zep and Graphiti.

View on GitHubWebsite
Stars
1,111
Forks
142
Open issues
2
Last commit
3 days ago

What is it?

Agent Memory Techniques is a hands-on learning resource for giving LLM-based agents memory across turns, sessions and tasks. Without memory, an agent has to rebuild context on every interaction, which makes personalization and long-running coherence difficult. The repository organizes 30 techniques into six families: short-term context management, long-term storage, cognitive architectures, retrieval and multi-agent patterns, frameworks, and evaluation with production patterns.

Each technique lives in its own notebook that can be run locally or opened in Colab. Early notebooks cover conversation buffers, sliding windows, summary and token-budget memory, while later ones cover vector stores, entity and knowledge graph memory, episodic and semantic memory, and forgetting or decay. Framework notebooks walk through Graphiti, Mem0, Letta (MemGPT) and Zep, and the final section covers memory evaluation, LoCoMo and LongMemEval benchmarks, and production concerns such as caching, TTLs, backups and GDPR.

The notebooks use Python and rely on OpenAI or Anthropic API keys. The repository states that its code is for education and is not production-ready.

Key features

  • 0130 separate Jupyter notebooks, each covering one memory technique.
  • 02Covers short-term buffers, vector stores, knowledge graphs, and episodic and semantic memory.
  • 03Hands-on notebooks for Mem0, Letta (MemGPT), Zep and Graphiti.
  • 04Includes LoCoMo and LongMemEval benchmark walkthroughs.
  • 05Each notebook can be opened in Colab without local setup.

Use cases

  • Learning how conversation memory works before building an agent.
  • Comparing memory approaches such as summary buffers, vector stores and knowledge graphs.
  • Evaluating frameworks like Mem0 or Zep for a specific agent project.
  • Preparing a team to design memory systems for production agents.

Quick start

  1. 1

    Open the first notebook, 01 Conversation Buffer Memory, in its folder

  2. 2

    Or open the same notebook directly in Google Colab using the Colab badge

Topics

#agent-memory#llm#ai-agents#rag#knowledge-graph#vector-database#jupyter#learning-resources#anthropic#episodic-memory

FAQ

What is Agent_Memory_Techniques?
A learning repository of 30 runnable Jupyter notebooks covering agent memory techniques for LLM agents, from conversation buffers to Mem0, Letta, Zep and Graphiti. Agent Memory Techniques is a hands-on learning resource for giving LLM-based agents memory across turns, sessions and tasks. Without memory, an agent has to rebuild context on every interaction, which makes personalization and long-running coherence difficult. The repository organizes 30 techniques into six families: short-term context management, long-term storage, cognitive architectures, retrieval and multi-agent patterns, frameworks, and evaluation with production patterns.
Is Agent_Memory_Techniques free to use? What license does it have?
Agent_Memory_Techniques is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is Agent_Memory_Techniques still maintained?
Actively maintained. The last commit to Agent_Memory_Techniques was on Oct 17, 2026.
What can I use Agent_Memory_Techniques for?
Learning how conversation memory works before building an agent. Comparing memory approaches such as summary buffers, vector stores and knowledge graphs. Evaluating frameworks like Mem0 or Zep for a specific agent project. Preparing a team to design memory systems for production agents.
How do I install Agent_Memory_Techniques?
Open the first notebook, 01 Conversation Buffer Memory, in its folder Or open the same notebook directly in Google Colab using the Colab badge
Where can I get Agent_Memory_Techniques?
The source code and installation instructions are on GitHub: https://github.com/NirDiamant/Agent_Memory_Techniques

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects