[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-11107":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":17,"stars30d":18,"stars90d":15,"forks30d":15,"starsTrendScore":19,"compositeScore":20,"rankGlobal":10,"rankLanguage":10,"license":21,"archived":22,"fork":22,"defaultBranch":23,"hasWiki":22,"hasPages":22,"topics":24,"createdAt":10,"pushedAt":10,"updatedAt":45,"readmeContent":46,"aiSummary":47,"trendingCount":15,"starSnapshotCount":15,"syncStatus":14,"lastSyncTime":48,"discoverSource":49},11107,"Agent_Memory_Techniques","NirDiamant\u002FAgent_Memory_Techniques","NirDiamant","Agent memory for LLMs: 30 runnable Jupyter notebooks covering conversation buffers, vector stores, knowledge graphs, episodic and semantic memory, MemGPT, Mem0, Letta, Zep, Graphiti, LoCoMo benchmarks, and production patterns.","https:\u002F\u002Fdiamantai.substack.com\u002F",null,"Jupyter Notebook",502,68,2,0,17,27,220,51,93.02,"Apache License 2.0",false,"main",[25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44],"agent-memory","ai-agents","anthropic","cognitive-architecture","cookbook","episodic-memory","graphiti","jupyter-notebook","knowledge-graph","langchain","letta","llm-agents","llm-memory","mem0","memgpt","openai","rag","tutorial","vector-database","zep","2026-06-12 04:00:53","# 🧠 Agent Memory Techniques\n\n\u003Cp align=\"center\">\n  \u003Cpicture>\n    \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"images\u002Fhero_dark.png\">\n    \u003Cimg src=\"images\u002Fbanner.png\" alt=\"Agent Memory Techniques for LLMs: 30 runnable Jupyter notebooks covering every major memory pattern\" width=\"100%\"\u002F>\n  \u003C\u002Fpicture>\n\u003C\u002Fp>\n\n**Learn every agent memory technique for LLM agents.**\n\n> ⭐  **If you find this useful, please star the repo** so more learners can discover it.\n\n> 🧭  **New here?** Start with [01 Conversation Buffer Memory](all_techniques\u002F01_conversation_buffer_memory\u002F) or pick a [Learning Path](#-learning-paths). Prefer a visual? See the [Decision Tree](#-which-technique-do-i-need) below. 30 runnable Jupyter notebooks covering conversation buffers, vector stores, knowledge graphs, episodic and semantic memory, working memory, MemGPT, Mem0, Letta, Zep, Graphiti, LoCoMo benchmarks, and production memory patterns.\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fwww.apache.org\u002Flicenses\u002FLICENSE-2.0\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-Apache_2.0-blue.svg\" alt=\"License: Apache 2.0\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fwww.python.org\u002Fdownloads\u002F\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpython-3.10+-blue.svg\" alt=\"Python 3.10+\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fjupyter.org\u002F\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FJupyter-Notebook-orange.svg\" alt=\"Jupyter\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fstargazers\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FNirDiamant\u002FAgent_Memory_Techniques?style=social\" alt=\"GitHub Stars\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fissues\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fissues\u002FNirDiamant\u002FAgent_Memory_Techniques\" alt=\"Issues\">\u003C\u002Fa>\n  \u003Ca href=\".github\u002FCONTRIBUTING.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fcontributions-welcome-brightgreen.svg\" alt=\"Contributions Welcome\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n---\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Feurope-west1-amt-views-tracker.cloudfunctions.net\u002Famt-tracker?notebook=main-readme&click=social-linkedin&target=https%3A%2F%2Fwww.linkedin.com%2Fin%2Fnir-diamant-759323134%2F&text=LinkedIn\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLinkedIn-Connect-blue\" alt=\"LinkedIn\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Feurope-west1-amt-views-tracker.cloudfunctions.net\u002Famt-tracker?notebook=main-readme&click=social-twitter&target=https%3A%2F%2Ftwitter.com%2FNirDiamantAI&text=Twitter\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002FNirDiamantAI?label=Follow%20%40NirDiamantAI&style=social\" alt=\"Twitter\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Feurope-west1-amt-views-tracker.cloudfunctions.net\u002Famt-tracker?notebook=main-readme&click=social-reddit&target=https%3A%2F%2Fwww.reddit.com%2Fr%2FEducationalAI%2F&text=Reddit\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FReddit-Join%20our%20subreddit-FF4500?style=flat-square&logo=reddit&logoColor=white\" alt=\"Reddit\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Feurope-west1-amt-views-tracker.cloudfunctions.net\u002Famt-tracker?notebook=main-readme&click=social-discord&target=https%3A%2F%2Fdiscord.gg%2FcA6Aa4uyDX&text=Discord\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join%20our%20community-7289da?style=flat-square&logo=discord&logoColor=white\" alt=\"Discord\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Feurope-west1-amt-views-tracker.cloudfunctions.net\u002Famt-tracker?notebook=main-readme&click=sponsor-github&target=https%3A%2F%2Fgithub.com%2Fsponsors%2FNirDiamant&text=Sponsor\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=Sponsor&message=%E2%9D%A4&logo=GitHub&color=ff69b4\" alt=\"Sponsor\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Feurope-west1-amt-views-tracker.cloudfunctions.net\u002Famt-tracker?notebook=main-readme--hiring-badge&click=top-badge-hiring&target=https%3A%2F%2Fdiamant-ai.com%2Fjobs&text=Hiring%20Badge\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%92%BC%20Hiring-DiamantAI%20Collective-7c3aed?style=flat-square\" alt=\"DiamantAI Collective is hiring\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cdiv align=\"center\">\n\n# 📖 [The RAG Techniques Book is HERE](https:\u002F\u002Feurope-west1-rag-techniques-views-tracker.cloudfunctions.net\u002Frag-techniques-tracker?notebook=agent-memory-techniques--readme&click=book-buy-amazon-title&target=https%3A%2F%2Fwww.amazon.com%2Fdp%2FB0D76734SZ%3Ftag%3Ddiamantai-amt-20&text=The%20RAG%20Techniques%20Book%20is%20HERE)\n\n### From the same author\n\n\u003Ca href=\"https:\u002F\u002Feurope-west1-rag-techniques-views-tracker.cloudfunctions.net\u002Frag-techniques-tracker?notebook=agent-memory-techniques--readme&click=book-buy-amazon-image&target=https%3A%2F%2Fwww.amazon.com%2Fdp%2FB0D76734SZ%3Ftag%3Ddiamantai-amt-20&text=Best%20Seller%20Image\">\u003Cimg src=\"images\u002Frag_book_best_seller.png\" alt=\"#1 Best Seller in Generative AI on Amazon\" width=\"600\">\u003C\u002Fa>\n\n**#1 Best Seller on Amazon in Generative AI**\n\nWant to go deeper on RAG (Retrieval-Augmented Generation, the technique of giving a model extra documents so it can answer better)? The book is the long version. You'll get the **intuition** behind every technique. You'll get **side-by-side comparisons** that show when each one wins and when it quietly fails. You'll get **illustrations** that make the tricky parts click.\n\n### ⏳ Launch window only: **$0.99**\n\nThe price goes up once the launch window closes. Readers who grab it now lock in the lowest price it will ever have.\n\n### 👉 [**Get the book on Amazon before the price changes**](https:\u002F\u002Feurope-west1-rag-techniques-views-tracker.cloudfunctions.net\u002Frag-techniques-tracker?notebook=agent-memory-techniques--readme&click=book-buy-amazon-cta&target=https%3A%2F%2Fwww.amazon.com%2Fdp%2FB0D76734SZ%3Ftag%3Ddiamantai-amt-20&text=Get%20the%20book%20on%20Amazon%20before%20the%20price%20changes)\n\n\u003C\u002Fdiv>\n\n\u003Cdiv align=\"center\">\n\n---\n\n\u003Cimg src=\"images\u002Fcollective-banner.png\" alt=\"DiamantAI Collective - AI engineering jobs\" width=\"600\">\n\n## 💼 Apply for open AI engineering jobs\n\n**AI-first companies are hiring through the DiamantAI Collective.**\n\n[![See open jobs and apply](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%E2%9E%A1%EF%B8%8F%20%20See%20open%20jobs%20and%20apply-7c3aed?style=for-the-badge)](https:\u002F\u002Feurope-west1-amt-views-tracker.cloudfunctions.net\u002Famt-tracker?notebook=main-readme--jobs-panel&click=jobs-panel-see-all-roles&target=https%3A%2F%2Fdiamant-ai.com%2Fjobs&text=See%20open%20jobs%20and%20apply)\n\n---\n\n\u003C\u002Fdiv>\n\n## 📫 Stay Updated\n\n\u003Cdiv align=\"center\">\n\u003Ctable>\n\u003Ctr>\n\u003Ctd align=\"center\">🚀\u003Cbr>\u003Cb>Weekly\u003Cbr>Updates\u003C\u002Fb>\u003C\u002Ftd>\n\u003Ctd align=\"center\">💡\u003Cbr>\u003Cb>Expert\u003Cbr>Insights\u003C\u002Fb>\u003C\u002Ftd>\n\u003Ctd align=\"center\">🎯\u003Cbr>\u003Cb>Top 0.1%\u003Cbr>Content\u003C\u002Fb>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n\u003Ca href=\"https:\u002F\u002Feurope-west1-amt-views-tracker.cloudfunctions.net\u002Famt-tracker?notebook=main-readme&click=newsletter-subscribe-button&target=https%3A%2F%2Fdiamantai.substack.com%2F%3Fr%3D336pe4%26utm_campaign%3Dpub-share-checklist&text=Subscribe%20to%20DiamantAI%20Newsletter\">\u003Cimg src=\"images\u002Fsubscribe-button.svg\" alt=\"Subscribe to DiamantAI Newsletter\">\u003C\u002Fa>\n\n*Join over 50,000 readers getting clear AI tutorials every week.* ***Subscribers also get early access and a 33% discount on my book.***\n\u003C\u002Fdiv>\n\n\u003Ca href=\"https:\u002F\u002Feurope-west1-amt-views-tracker.cloudfunctions.net\u002Famt-tracker?notebook=main-readme&click=newsletter-subscribe-image&target=https%3A%2F%2Fdiamantai.substack.com%2F%3Fr%3D336pe4%26utm_campaign%3Dpub-share-checklist&text=DiamantAI%20newsletter\">\u003Cimg src=\"images\u002Fsubstack_image.png\" alt=\"DiamantAI newsletter\">\u003C\u002Fa>\n\n---\n\n\n---\n\n## 💡 Why Agent Memory Matters\n> ### 💡 Quick Answer (for search engines and skimmers)\n>\n> **Agent memory** is the set of techniques that let an LLM-based agent (a system built around a Large Language Model) remember information across turns, sessions, and tasks. Without memory, an agent re-derives context every time and cannot personalize, learn, or maintain coherence over long interactions. This repository documents 30 distinct memory techniques, grouped into six families: short-term context management, long-term storage, cognitive architectures, retrieval and multi-agent patterns, batteries-included frameworks, and production deployment patterns.\n\n\nThink about a friend who forgets every conversation you've ever had. Every morning you're strangers again. That's what most AI agents are like today.\n\nEvery AI agent eventually hits the same wall: **it forgets**.\n\nIn 2026, AI agents are everywhere. But most of them still forget what you told them yesterday. Without strong memory, an agent can't keep context across conversations. It can't learn from past chats. It can't build a lasting relationship with you.\n\nThe landscape is shifting fast:\n\n- **Anthropic's 7 Layers of Memory** (March 2026): from conversation context to cross-project knowledge, defining the memory hierarchy for Claude Code\n- **Mem0**: managed memory layer gaining rapid adoption for personalized AI\n- **Letta (MemGPT)**: self-editing memory with inner\u002Fouter monologue architecture\n- **Zep**: temporal knowledge graphs for long-term agent memory\n- **Graphiti**: episodic-to-semantic knowledge graph extraction\n- **MemOS & Memori**: memory-as-infrastructure platforms for production agents\n\nBut there's no single hands-on guide that teaches you **how each technique works, when to use it, and how to build it yourself**.\n\nThat's why this repository exists. **30 techniques. Runnable notebooks. Real code you can use today.**\n\n---\n\n## 🗺️ Taxonomy of Agent Memory Techniques\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"images\u002Ftaxonomy.png\" alt=\"Agent memory taxonomy: 30 techniques across 6 families (short-term, long-term, cognitive architectures, retrieval, frameworks, production)\" width=\"720\"\u002F>\n\u003C\u002Fp>\n\nThe 30 techniques fall into six families. Each family solves a different memory problem. Each technique lives in its own notebook.\n\n| Family | What it solves | Techniques |\n|---|---|---|\n| **Short-term** | Keep recent turns in memory without filling up the context window. | 01 - 05 |\n| **Long-term** | Save knowledge across sessions, users, and time. | 06 - 11 |\n| **Cognitive architectures** | Working, hierarchical, and reflective memory systems. | 12 - 19 |\n| **Retrieval & routing** | Choose what to recall and when. | 20 - 23 |\n| **Frameworks** | Production-ready memory libraries (Mem0, Letta, Zep, Graphiti). | 24 - 27 |\n| **Evaluation & production** | Measure, benchmark, and deploy memory. | 28 - 30 |\n\n\n---\n\n## 🧭 Which Technique Do I Need?\n\n30 techniques grouped by what you are building. Pick the group that matches your goal, then open the technique inside it.\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"images\u002Fdecision_tree.svg\" alt=\"Decision tree: which agent memory technique do I need?\" width=\"100%\"\u002F>\n\u003C\u002Fp>\n\n\n\n\u003C!-- decision-tree-text-fallback -->\n**Quick text version:**\n\n- Need to manage the current chat? Start with **01-05** (short-term memory).\n- Need to persist across sessions? Start with **06 Vector Store** or **21 Cross-Session Memory**.\n- Building a cognitive architecture with multiple stores? See **12-19**.\n- Using a framework? Go straight to **24 Graphiti**, **25 Mem0**, **26 Letta**, or **27 Zep**.\n- Evaluating or shipping to production? See **28-30**.\n\n**Still not sure?** Start with [01 Conversation Buffer](all_techniques\u002F01_conversation_buffer_memory\u002F). Almost every other technique builds on it.\n\n---\n\n## 📐 Compare Techniques at a Glance\n\nLooking to filter by constraint (persistence, retrieval style, token cost, best-for use case)? See the [side-by-side comparison matrix](docs\u002Fcomparison.md) covering all 30 techniques in one table.\n\n---\n\n## 📚 All 30 Techniques\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"images\u002Fsections\u002Fshort_term.png\" alt=\"Short-term memory techniques for LLM agents: conversation buffers, sliding window, summary, token budget\" width=\"100%\"\u002F>\n\u003C\u002Fp>\n\n### 🔄 Short-Term Memory (Techniques 1-5)\n\nManage the conversation inside a single chat.\n\n| # | Technique | Description | Notebook |\n|---|-----------|-------------|----------|\n| 01 | [Conversation Buffer Memory](all_techniques\u002F01_conversation_buffer_memory\u002F) | Save the full conversation, word for word. The simplest pattern, and the base for everything else. | ✅ [Notebook](all_techniques\u002F01_conversation_buffer_memory\u002Fconversation_buffer_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F01_conversation_buffer_memory\u002Fconversation_buffer_memory.ipynb) |\n| 02 | [Sliding Window Memory](all_techniques\u002F02_sliding_window_memory\u002F) | Keep only the last few messages. You limit the size, but you keep the recent parts. | ✅ [Notebook](all_techniques\u002F02_sliding_window_memory\u002Fsliding_window_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F02_sliding_window_memory\u002Fsliding_window_memory.ipynb) |\n| 03 | [Summary Memory](all_techniques\u002F03_summary_memory\u002F) | Replace old turns with a short summary written by the model. You lose length but keep the meaning. | ✅ [Notebook](all_techniques\u002F03_summary_memory\u002Fsummary_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F03_summary_memory\u002Fsummary_memory.ipynb) |\n| 04 | [Summary Buffer Memory](all_techniques\u002F04_summary_buffer_memory\u002F) | Summarize older turns, but keep recent messages word for word. You get both. | ✅ [Notebook](all_techniques\u002F04_summary_buffer_memory\u002Fsummary_buffer_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F04_summary_buffer_memory\u002Fsummary_buffer_memory.ipynb) |\n| 05 | [Token Buffer Memory](all_techniques\u002F05_token_buffer_memory\u002F) | Trim the history to fit a strict token budget. Drop the oldest messages first. | ✅ [Notebook](all_techniques\u002F05_token_buffer_memory\u002Ftoken_buffer_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F05_token_buffer_memory\u002Ftoken_buffer_memory.ipynb) |\n\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"images\u002Fsections\u002Flong_term.png\" alt=\"Long-term memory techniques for LLM agents: vector store, entity, knowledge graph, episodic, semantic, procedural\" width=\"100%\"\u002F>\n\u003C\u002Fp>\n\n### 💾 Long-Term Memory (Techniques 6-11)\n\nStorage that survives across sessions and users.\n\n| # | Technique | Description | Notebook |\n|---|-----------|-------------|----------|\n| 06 | [Vector Store Memory](all_techniques\u002F06_vector_store_memory\u002F) | Turn past messages into vectors (number lists that capture meaning). Search them later by similarity. | ✅ [Notebook](all_techniques\u002F06_vector_store_memory\u002Fvector_store_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F06_vector_store_memory\u002Fvector_store_memory.ipynb) |\n| 07 | [Entity Memory](all_techniques\u002F07_entity_memory\u002F) | Pull out and track facts about people, projects, and preferences. Update them as the conversation grows. | ✅ [Notebook](all_techniques\u002F07_entity_memory\u002Fentity_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F07_entity_memory\u002Fentity_memory.ipynb) |\n| 08 | [Knowledge Graph Memory](all_techniques\u002F08_knowledge_graph_memory\u002F) | Build a graph of how entities connect. Walk the graph to reason over what the agent has learned. | ✅ [Notebook](all_techniques\u002F08_knowledge_graph_memory\u002Fknowledge_graph_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F08_knowledge_graph_memory\u002Fknowledge_graph_memory.ipynb) |\n| 09 | [Episodic Memory](all_techniques\u002F09_episodic_memory\u002F) | Store complete interactions with when-and-where context. Good for \"what happened when\" questions. | ✅ [Notebook](all_techniques\u002F09_episodic_memory\u002Fepisodic_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F09_episodic_memory\u002Fepisodic_memory.ipynb) |\n| 10 | [Semantic Memory](all_techniques\u002F10_semantic_memory\u002F) | Pull general facts out of interactions. Store them on their own, away from the raw episodes. | ✅ [Notebook](all_techniques\u002F10_semantic_memory\u002Fsemantic_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F10_semantic_memory\u002Fsemantic_memory.ipynb) |\n| 11 | [Procedural Memory](all_techniques\u002F11_procedural_memory\u002F) | Capture \"how-to\" knowledge: the procedures and workflows the agent picks up over time. | ✅ [Notebook](all_techniques\u002F11_procedural_memory\u002Fprocedural_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F11_procedural_memory\u002Fprocedural_memory.ipynb) |\n\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"images\u002Fsections\u002Fcognitive.png\" alt=\"Cognitive architecture memory patterns: working memory, hierarchical layers, consolidation, compaction, self-reflection, routing, temporal, forgetting\" width=\"100%\"\u002F>\n\u003C\u002Fp>\n\n### 🧩 Cognitive Architectures (Techniques 12-19)\n\nPatterns borrowed from how humans remember.\n\n| # | Technique | Description | Notebook |\n|---|-----------|-------------|----------|\n| 12 | [Working Memory & Context Window](all_techniques\u002F12_working_memory_context_window\u002F) | Manage the agent's limited attention. Prioritize, pin, and evict context on the fly. | ✅ [Notebook](all_techniques\u002F12_working_memory_context_window\u002Fworking_memory_context_window.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F12_working_memory_context_window\u002Fworking_memory_context_window.ipynb) |\n| 13 | [Hierarchical Memory Layers](all_techniques\u002F13_hierarchical_memory_layers\u002F) | Tiered storage with hot, warm, and cold layers. Promote and demote items as they age. | ✅ [Notebook](all_techniques\u002F13_hierarchical_memory_layers\u002Fhierarchical_memory_layers.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F13_hierarchical_memory_layers\u002Fhierarchical_memory_layers.ipynb) |\n| 14 | [Memory Consolidation](all_techniques\u002F14_memory_consolidation\u002F) | Merge, deduplicate, and strengthen memories. Inspired by how the brain consolidates during sleep. | ✅ [Notebook](all_techniques\u002F14_memory_consolidation\u002Fmemory_consolidation.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F14_memory_consolidation\u002Fmemory_consolidation.ipynb) |\n| 15 | [Memory Compaction](all_techniques\u002F15_memory_compaction\u002F) | Compress stored memories with summaries, entity extraction, or distillation. Save storage and tokens. | ✅ [Notebook](all_techniques\u002F15_memory_compaction\u002Fmemory_compaction.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F15_memory_compaction\u002Fmemory_compaction.ipynb) |\n| 16 | [Self-Reflection Memory](all_techniques\u002F16_self_reflection_memory\u002F) | The agent looks back at its own actions. It writes notes on what worked, and uses them next time. | ✅ [Notebook](all_techniques\u002F16_self_reflection_memory\u002Fself_reflection_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F16_self_reflection_memory\u002Fself_reflection_memory.ipynb) |\n| 17 | [Memory Routing](all_techniques\u002F17_memory_routing\u002F) | Pick the right memory store to read from or write to. Route by content type and intent. | ✅ [Notebook](all_techniques\u002F17_memory_routing\u002Fmemory_routing.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F17_memory_routing\u002Fmemory_routing.ipynb) |\n| 18 | [Temporal Memory](all_techniques\u002F18_temporal_memory\u002F) | Attach timestamps to memories. Retrieve with time awareness and weight recent items higher. | ✅ [Notebook](all_techniques\u002F18_temporal_memory\u002Ftemporal_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F18_temporal_memory\u002Ftemporal_memory.ipynb) |\n| 19 | [Forgetting & Decay](all_techniques\u002F19_forgetting_and_decay\u002F) | Forget on purpose. Use decay, access counts, or relevance to prune. | ✅ [Notebook](all_techniques\u002F19_forgetting_and_decay\u002Fforgetting_and_decay.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F19_forgetting_and_decay\u002Fforgetting_and_decay.ipynb) |\n\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"images\u002Fsections\u002Fretrieval.png\" alt=\"Memory retrieval and multi-agent patterns: retrieval patterns, cross-session memory, multi-agent shared memory, memory as tools\" width=\"100%\"\u002F>\n\u003C\u002Fp>\n\n### 🔍 Retrieval & Multi-Agent (Techniques 20-23)\n\nHow agents find and share memories.\n\n| # | Technique | Description | Notebook |\n|---|-----------|-------------|----------|\n| 20 | [Memory Retrieval Patterns](all_techniques\u002F20_memory_retrieval_patterns\u002F) | Compare retrieval strategies: semantic search, recency, hybrid scoring, diversity, and re-ranking. | ✅ [Notebook](all_techniques\u002F20_memory_retrieval_patterns\u002Fmemory_retrieval_patterns.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F20_memory_retrieval_patterns\u002Fmemory_retrieval_patterns.ipynb) |\n| 21 | [Cross-Session Memory](all_techniques\u002F21_cross_session_memory\u002F) | Save and reload agent state across sessions. The user picks up where they left off. | ✅ [Notebook](all_techniques\u002F21_cross_session_memory\u002Fcross_session_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F21_cross_session_memory\u002Fcross_session_memory.ipynb) |\n| 22 | [Multi-Agent Shared Memory](all_techniques\u002F22_multi_agent_shared_memory\u002F) | Shared stores, message passing, and agreement protocols for multi-agent teams. | ✅ [Notebook](all_techniques\u002F22_multi_agent_shared_memory\u002Fmulti_agent_shared_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F22_multi_agent_shared_memory\u002Fmulti_agent_shared_memory.ipynb) |\n| 23 | [Memory with Tools](all_techniques\u002F23_memory_with_tools\u002F) | Give the agent memory tools it can call: save, search, forget. Treated like any other tool. | ✅ [Notebook](all_techniques\u002F23_memory_with_tools\u002Fmemory_with_tools.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F23_memory_with_tools\u002Fmemory_with_tools.ipynb) |\n\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"images\u002Fsections\u002Fframeworks.png\" alt=\"Agent memory frameworks and libraries: Graphiti, Mem0, Letta (MemGPT), Zep\" width=\"100%\"\u002F>\n\u003C\u002Fp>\n\n### 🔧 Frameworks & Platforms (Techniques 24-27)\n\nWork with the leading memory frameworks, hands-on.\n\n| # | Technique | Description | Notebook |\n|---|-----------|-------------|----------|\n| 24 | [Graph Memory with Graphiti](all_techniques\u002F24_graph_memory_graphiti\u002F) | Use Zep's Graphiti to build time-aware knowledge graphs from chat. Extract episodes and general facts. | ✅ [Notebook](all_techniques\u002F24_graph_memory_graphiti\u002Fgraph_memory_graphiti.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F24_graph_memory_graphiti\u002Fgraph_memory_graphiti.ipynb) |\n| 25 | [Mem0 Patterns](all_techniques\u002F25_mem0_patterns\u002F) | Use Mem0's managed memory layer. It handles extracting, storing, and fetching user-specific memories. | ✅ [Notebook](all_techniques\u002F25_mem0_patterns\u002Fmem0_patterns.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F25_mem0_patterns\u002Fmem0_patterns.ipynb) |\n| 26 | [Letta (MemGPT) Patterns](all_techniques\u002F26_letta_memgpt_patterns\u002F) | Build MemGPT's self-editing memory. Covers inner monologue, heartbeat events, and memory pressure. | ✅ [Notebook](all_techniques\u002F26_letta_memgpt_patterns\u002Fletta_memgpt_patterns.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F26_letta_memgpt_patterns\u002Fletta_memgpt_patterns.ipynb) |\n| 27 | [Zep Memory](all_techniques\u002F27_zep_memory\u002F) | Use Zep for dialog classification, entity extraction, and time-aware graphs. Built for production. | ✅ [Notebook](all_techniques\u002F27_zep_memory\u002Fzep_memory.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F27_zep_memory\u002Fzep_memory.ipynb) |\n\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"images\u002Fsections\u002Fevaluation.png\" alt=\"Agent memory evaluation and production: memory evaluation, LoCoMo and LongMemEval benchmarks, production deployment patterns\" width=\"100%\"\u002F>\n\u003C\u002Fp>\n\n### 📊 Evaluation & Production (Techniques 28-30)\n\nMeasure your memory. Then ship it.\n\n| # | Technique | Description | Notebook |\n|---|-----------|-------------|----------|\n| 28 | [Memory Evaluation](all_techniques\u002F28_memory_evaluation\u002F) | Measure memory quality. Check retrieval precision and recall, staleness, contradictions, and user satisfaction. | ✅ [Notebook](all_techniques\u002F28_memory_evaluation\u002Fmemory_evaluation.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F28_memory_evaluation\u002Fmemory_evaluation.ipynb) |\n| 29 | [Memory Benchmarks (LoCoMo)](all_techniques\u002F29_memory_benchmarks_LoCoMo\u002F) | Run your memory against LoCoMo and LongMemEval benchmarks. See how it does over long conversations. | ✅ [Notebook](all_techniques\u002F29_memory_benchmarks_LoCoMo\u002Fmemory_benchmarks_locomo.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F29_memory_benchmarks_LoCoMo\u002Fmemory_benchmarks_locomo.ipynb) |\n| 30 | [Production Memory Patterns](all_techniques\u002F30_production_memory_patterns\u002F) | Run memory at scale. Caching, TTLs (time-to-live), sharding, backups, GDPR, and observability. | ✅ [Notebook](all_techniques\u002F30_production_memory_patterns\u002Fproduction_memory_patterns.ipynb) · [![Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fblob\u002Fmain\u002Fall_techniques\u002F30_production_memory_patterns\u002Fproduction_memory_patterns.ipynb) |\n\n\n---\n\n## 🎯 Learning Paths\n\n### Beginner: Foundations\n\nNew to agent memory? Start here. These are the building blocks.\n\n```\n01 Conversation Buffer → 02 Sliding Window → 03 Summary Memory →\n05 Token Buffer → 06 Vector Store Memory → 21 Cross-Session Memory\n```\n\n### Intermediate: Structured Memory\n\nReady for more? Add entities, graphs, and smarter retrieval.\n\n```\n07 Entity Memory → 08 Knowledge Graph → 09 Episodic Memory →\n10 Semantic Memory → 20 Retrieval Patterns → 22 Multi-Agent Shared Memory\n```\n\n### Advanced: Cognitive Architectures\n\nBuild human-inspired memory patterns for advanced agents.\n\n```\n12 Working Memory → 13 Hierarchical Layers → 14 Consolidation →\n16 Self-Reflection → 17 Memory Routing → 19 Forgetting & Decay\n```\n\n### Practitioner: Frameworks & Production\n\nConnect to production tools and measure what you've built.\n\n```\n25 Mem0 → 26 Letta\u002FMemGPT → 24 Graphiti → 27 Zep →\n28 Evaluation → 29 Benchmarks → 30 Production Patterns\n```\n\n---\n\n## 🚀 Quick Start\n\n> 💡  **Prefer not to install anything?** Every notebook renders on GitHub directly. Click a technique in the table above to read it in your browser. Or use the Colab badges to run it in the cloud.\n\n\n```bash\n# Clone the repository\ngit clone https:\u002F\u002Fgithub.com\u002FNirDiamant\u002FAgent_Memory_Techniques.git\ncd Agent_Memory_Techniques\n\n# Create a virtual environment\npython -m venv venv\nsource venv\u002Fbin\u002Factivate  # On Windows: venv\\Scripts\\activate\n\n# Install dependencies\npip install -r requirements.txt\n\n# Set up your API keys\ncp .env.example .env\n# Edit .env with your OPENAI_API_KEY and\u002For ANTHROPIC_API_KEY\n\n# Launch Jupyter and start with the first technique\njupyter notebook all_techniques\u002F01_conversation_buffer_memory\u002F\n```\n\n---\n\n## 📁 Project Structure\n\n```\nAgent_Memory_Techniques\u002F\n├── README.md                           # You are here\n├── ROADMAP.md                          # Current state and what's next\n├── LICENSE                             # Apache 2.0\n├── CITATION.cff                        # How to cite this work\n├── requirements.txt                    # Python dependencies\n├── .env.example                        # API key template\n├── llms.txt                            # LLM-discoverability index\n│\n├── all_techniques\u002F                     # 30 technique folders, each with notebook + README\n│   ├── 01_conversation_buffer_memory\u002F\n│   ├── 02_sliding_window_memory\u002F\n│   ├── ...\n│   └── 30_production_memory_patterns\u002F\n│\n├── docs\u002F                               # Project documentation\n│   ├── architecture.md                 # Memory system design patterns\n│   ├── comparison.md                   # Side-by-side comparison of all 30 techniques\n│   ├── glossary.md                     # Key terms and definitions\n│   ├── learning_path.md                # Detailed learning path guide\n│   ├── topics.md                       # Keyword index\n│   ├── roadmap.md                      # Original planning archive\n│   ├── FAQ.md                          # Frequently asked questions\n│   └── CONTENT_STANDARDS.md            # Writing-style rules\n│\n├── .github\u002F                            # GitHub community files\n│   ├── CONTRIBUTING.md                 # How to contribute\n│   ├── CODE_OF_CONDUCT.md              # Community guidelines\n│   ├── SECURITY.md                     # Security policy\n│   ├── FUNDING.yml                     # Sponsorship config\n│   ├── ISSUE_TEMPLATE\u002F                 # Issue templates\n│   ├── pull_request_template.md        # PR template\n│   └── workflows\u002F                      # CI workflows\n│\n├── utils\u002F                              # Shared helpers and validators\n│   ├── helpers.py                      # Env loading, LLM clients, cosine, tokens\n│   ├── validate_cells.py               # Notebook cell-structure validator\n│   └── validate_style.py               # Prose-style validator\n│\n├── tests\u002F                              # pytest smoke tests\n├── data\u002F                               # Small sample datasets\n└── images\u002F                             # Diagrams and visuals\n```\n\n---\n\n## 🤝 Contributing\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002FNirDiamant\u002FAgent_Memory_Techniques\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Fcontrib.rocks\u002Fimage?repo=NirDiamant\u002FAgent_Memory_Techniques\" alt=\"Contributors\" \u002F>\n\u003C\u002Fa>\n\n\nWe welcome contributions. You can fill in a notebook, fix a bug, improve the docs, or propose a new technique. Every contribution helps the next reader.\n\nSee [CONTRIBUTING.md](.github\u002FCONTRIBUTING.md) for the details.\n\n**Where we need help the most:**\n- More techniques we haven't covered yet (propose one via an issue)\n- Architecture diagrams (Mermaid or ASCII)\n- More memory benchmarks and evaluation metrics\n- Integration examples for new frameworks\n\n---\n\n---\n\n## 💖 Sponsors\n\nSupporting this project helps keep educational AI content free and open. If your company uses agent memory in production, consider sponsoring to get your logo below.\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsponsors\u002FNirDiamant\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=Sponsor&message=%E2%9D%A4&logo=GitHub&color=ff69b4\" alt=\"Become a Sponsor\"\u002F>\u003C\u002Fa>\n\n---\n\n## 🔗 Related Work\n\nThis repo is part of a bigger collection of AI technique tutorials.\n\n| Repository | Stars | Focus |\n|-----------|-------|-------|\n| [RAG Techniques](https:\u002F\u002Fgithub.com\u002FNirDiamant\u002FRAG_Techniques) | 26k+ | Retrieval-Augmented Generation techniques |\n| [GenAI Agents](https:\u002F\u002Fgithub.com\u002FNirDiamant\u002FGenAI_Agents) | 21k+ | Generative AI agent architectures |\n| [Agents Towards Production](https:\u002F\u002Fgithub.com\u002FNirDiamant\u002Fagents-towards-production) | 18k+ | Production-grade agent deployment |\n| [Prompt Engineering](https:\u002F\u002Fgithub.com\u002FNirDiamant\u002FPrompt_Engineering) | 7k+ | Prompt engineering techniques |\n\n---\n\n## 🏷️ Topics Covered\n\nThis repository is a practical reference for agent memory in Large Language Model (LLM) applications. For the full keyword index covering short-term, long-term, cognitive architectures, retrieval, frameworks, evaluation, and production patterns, see [docs\u002Ftopics.md](docs\u002Ftopics.md).\n\n---\n\n## ⚠️ Disclaimer\n\nThis repository is for educational purposes. The code here shows how agent memory techniques work. It is not production-ready software. Do not use it as-is for handling regulated data, medical decisions, legal advice, or any high-stakes application without a careful review. The authors accept no responsibility for how you use this material.\n\n---\n\n## 📄 License\n\nThis project is licensed under the Apache License 2.0. See the [LICENSE](LICENSE) file for details.\n\n---\n\n## 📖 Citation\n\nIf you use this repository in your research or teaching, please cite:\n\n```bibtex\n@misc{diamant2026agentmemory,\n    title={Agent Memory Techniques: A Comprehensive Collection},\n    author={Nir Diamant},\n    year={2026},\n    url={https:\u002F\u002Fgithub.com\u002FNirDiamant\u002FAgent_Memory_Techniques\n}\n```\n\n---\n\n**Built with care by [Nir Diamant](https:\u002F\u002Fgithub.com\u002FNirDiamant)**, making advanced AI accessible to everyone.\n\n---\n\n![](https:\u002F\u002Feurope-west1-amt-views-tracker.cloudfunctions.net\u002Famt-tracker?notebook=main-readme)\n","该项目提供了一套针对大型语言模型（LLM）代理的记忆技术教程，包括30个可运行的Jupyter笔记本。核心功能涵盖了对话缓冲、向量存储、知识图谱、情景与语义记忆等主要记忆模式，并且引入了MemGPT、Mem0、Letta、Zep、Graphiti等多种工具和技术。这些内容不仅适用于研究者深入了解AI代理如何处理和存储信息，也适合开发者在实际项目中应用先进的记忆管理策略。通过本项目的学习路径，用户可以系统地掌握从基础到高级的各种记忆技术实现方式。","2026-06-11 03:31:12","CREATED_QUERY"]