A curated map of memory for AI agents โ the systems, benchmarks, and research that give LLM and multimodal agents long-term context, persistent recall, and the ability to improve from experience.
๐ Open-source resources (e.g. papers with reproducible code publicly available on Github) are marked in bold font and ranked higher.
---
### ๐งญ Start Here
| If you want toโฆ | Jump to |
|---|---|
| Add a memory layer to an agent you're building | [๐ฟ Products](#-products) |
| Choose a benchmark to evaluate a memory system | [๐ Benchmarks](#-benchmarks) |
| Get oriented in the field | [๐ Tutorials](#-tutorials) ยท [๐ Surveys](#-surveys) |
| Follow research on memory architectures | [๐ค Nonparametric Memory](#-papers---nonparametric-memory) ยท [๐ข Parametric Memory](#-papers---parametric-memory) |
| Build agents that learn from experience | [๐ Memory for Agent Evolution](#-papers---memory-for-agent-evolution) |
| Protect agent memory from poisoning and abuse | [๐ Memory Security & Defense](#-memory-security--defense) |
๐ How this list is curated
- Open-source products are ordered by GitHub star count โ an objective, CI-checked popularity signal, not a quality ranking or an endorsement. Products with fewer than 100 stars sit in a collapsed **Emerging projects** section and graduate into the main list once they cross that threshold.
- **Bold** marks resources with reproducible code publicly available.
- Descriptions are factual, not promotional (see the [contributing guide](CONTRIBUTING.md)).
- This list is maintained by [Bloo-Mind AI](https://www.bloo-mind.ai/) and the Ubiquitous AGI team at TeleAI. Entries affiliated with the maintainers are marked with โ and follow the same ranking, format, and style rules as every other entry. One convention applies to all entries regardless of affiliation: an API-compatible drop-in replacement for a listed product is nested under that product as an unranked sub-item rather than given its own star-ranked position.
- Projects that are inactive, archived, or whose claims are disputed move to the [Archival](#archival) subsection with a neutral status label, links to the evidence, and the date the status was last checked.
๐ฐ In the News
- ๐ฐ [[NVIDIA (2026-08-21)] AVO reaches 100% on ARC-AGI-3, crediting persistent memory across context resets and a supervisor that redirects a stalled agent](https://developer.nvidia.com/blog/nvidia-avo-reaches-100-on-arc-agi-3-demonstrating-a-frontier-level-general-purpose-architecture-for-long-horizon-autonomous-agents/)
- ๐ฐ [[Agent Memory Leaderboard (2026-08-13)] First public results: 136 teams registered and 69 memory systems completed the first text-memory evaluation](https://news.ycombinator.com/item?id=49281370)
- ๐ฐ [[Perplexity (2026-06-18)] Perplexity launches Brain, a self-improving memory system](https://www.perplexity.ai/hub/blog/self-improving-memory-for-agents)
- ๐ฐ [[OpenAI (2026-06-04)] Dreaming: Better memory for a more helpful ChatGPT](https://openai.com/index/chatgpt-memory-dreaming/)
- ๐ฐ [[Bloo-Mind AI (2026-05-20)] The Benchmark Theatre: Why Almost Nothing Youโve Read About Agent Memory Scores Is True](https://essays.bloo-mind.ai/posts/2026-05-20-mem-eval/) โ
- ๐ฐ [[Jiayi Weng (2026-05-09)] Learning Beyond Gradients](https://trinkle23897.github.io/learning-beyond-gradients/)
- ๐ฐ [[Anthropic (2026-05-08)] Three key areas Anthropic is working on for their next models](https://www.reddit.com/r/singularity/comments/1t5q53r/three_key_areas_anthropic_is_working_on_for_their/)
- ๐ฐ [[InfoQ (2026-04-30)] Cloudflare Announces Agent Memory, a Managed Persistent Memory Service for AI Agents](https://www.infoq.com/news/2026/04/cloudflare-agent-memory-beta/)
- ๐ฐ [[OpenAI (2026-04-22)] Chronicle: Build Codex Memories from Recent Screen Context](https://developers.openai.com/codex/memories/chronicle)
* _Open-Source Alternatives_: [OpenChronicle](https://github.com/Einsia/OpenChronicle), [MemScreen](https://github.com/smileformylove/MemScreen)
- ๐ฐ [[a16z (2026-04-22)] Why We Need Continual Learning](https://a16z.com/why-we-need-continual-learning/)
- ๐ฐ [[AI Godfather (2026-04-08)] MemPalace - How Milla Jovovich's AI Project Scammed the Internet](https://www.youtube.com/watch?v=WlxNNvDHJkE)
- ๐ฐ [[Troy Hua (2026-03-31)] How Anthropic Built 7 Layers of Memory and a Dreaming System for Claude Code](https://x.com/troyhua/status/2039052328070734102)
- ๐ฐ [[VelvetShark (2026-03-05)] OpenClaw Memory Masterclass: The complete guide to agent memory that survives](https://velvetshark.com/openclaw-memory-masterclass)
- ๐ฐ [[Business Insider (2026-01-08)] AI still needs a breakthrough in one key area to reach superintelligence, according to those who build it](https://www.businessinsider.com/superintelligent-ai-memory-sam-altman-2026-1)
---
๐๏ธ Table of Contents
**If you find this page helpful, please give it a โญ๏ธ โ starring also keeps updates in your GitHub feed.**
_๐ค Contributions welcome! Feel free to open an issue or submit a pull request to add papers, fix links, or improve categorization โ see the [contributing guide](CONTRIBUTING.md) for entry formats._
---
## ๐ฟ Products
### Open-Source
_Ordered by the number of GitHub stars. Products with fewer than 100 stars continue the list inside the collapsed **Emerging projects** section below โ they graduate into the main list once they cross that threshold. An API-compatible drop-in replacement for a listed product appears as an unranked sub-item under that product._
1. **[Claude-Mem](https://cmem.ai/)**
[[code](https://github.com/thedotmack/claude-mem)]
[[docs](https://docs.claude-mem.ai/introduction)]
[[blog](https://cmem.ai/blog)]
_Hook-based session capture compressed into searchable observations and re-injected into later sessions across Claude Code, Codex, Cursor, OpenClaw and other hosts; optional hosted cloud sync._
2. **[Mem0](https://mem0.ai/)**
[[code](https://github.com/mem0ai/mem0)]
[[docs](https://docs.mem0.ai/)]
[[paper](https://arxiv.org/abs/2504.19413)]
[[blog](https://mem0.ai/blog)]
_Universal memory layer for AI agents._
- **[TeleMem](https://github.com/TeleAI-UAGI/TeleMem)** โ
[[code](https://github.com/TeleAI-UAGI/TeleMem)]
[[docs](https://teleai-uagi.github.io/telemem/)]
[[paper](https://arxiv.org/abs/2601.06037)]
_API-compatible high-performance drop-in replacement for Mem0 (`import telemem as mem0`); listed as an unranked sub-item of Mem0 per the drop-in replacement convention. Maintainer-affiliated._
3. **[Hindsight](https://hindsight.vectorize.io/)**
[[code](https://github.com/vectorize-io/hindsight)]
[[paper](https://arxiv.org/abs/2512.12818)]
_Agent memory layer that learns from interaction feedback to improve recall over time._
4. **[OpenViking](https://openviking.ai/)**
[[code](https://github.com/volcengine/OpenViking)]
[[docs](https://docs.openviking.ai/)]
[[paper1](https://arxiv.org/abs/2605.29640)]
[[paper2](https://arxiv.org/abs/2606.16903)]
[[blog](https://blog.openviking.ai/)]
_Self-evolving context database for AI agents that unifies agent memory, knowledge RAG, and skills behind one storage/retrieval layer, with an MCP server for cross-session read/write._
5. **[Cognee](https://www.cognee.ai/)**
[[code](https://github.com/topoteretes/cognee)]
[[paper](https://arxiv.org/abs/2505.24478)]
[[blog](https://www.cognee.ai/blog)]
_Memory engine that ingests data into a hybrid graph + vector knowledge graph for cross-session agent recall._
6. **[Zep (powered by Graphiti)](https://www.getzep.com/)**
[[code](https://github.com/getzep/graphiti)]
[[paper](https://arxiv.org/abs/2501.13956)]
[[blog](https://blog.getzep.com/)]
_Real-time temporal knowledge graphs for AI agents._
7. **[gbrain](https://github.com/garrytan/gbrain)**
[[code](https://github.com/garrytan/gbrain)]
_Garry's opinionated OpenClaw/Hermes agent brain._
8. **[agentmemory](https://www.agent-memory.dev/)**
[[code](https://github.com/rohitg00/agentmemory)]
_Persistent memory for AI coding agents._
9. **[TencentDB Agent Memory](https://github.com/Tencent/TencentDB-Agent-Memory)**
[[code](https://github.com/Tencent/TencentDB-Agent-Memory)]
_Fully local long-term memory for AI agents via a 4-tier progressive pipeline, with zero external API dependencies._
10. **[Letta (formerly MemGPT)](https://www.letta.com/)**
[[code](https://github.com/letta-ai/letta)]
[[paper](https://arxiv.org/abs/2310.08560)]
[[research](https://www.letta.com/research)]
[[blog](https://www.letta.com/blog)]
_Stateful-agent platform with hierarchical memory that learns and self-improves over time._
11. **[Second Me](https://home.second.me/)**
[[code](https://github.com/mindverse/Second-Me)]
[[paper](https://arxiv.org/abs/2503.08102)]
_Personal AI trained on the user to represent them across applications._
12. **[MemU](https://memu.pro/)**
[[code](https://github.com/NevaMind-AI/memU)]
[[blog](https://memu.pro/blog)]
_Memory layer for 24/7 proactive agents._
13. **[EverOS (part of EverMind)](https://evermind-ai.com/)**
[[code](https://github.com/EverMind-AI/EverOS)]
[[blog](https://evermind-ai.com/blog/)]
_Toolkit for building, evaluating, and integrating long-term memory in self-evolving agents._
14. **[MemOS (by MemTensor)](https://memos.openmem.net/)**
[[code](https://github.com/MemTensor/MemOS)]
[[paper](https://arxiv.org/abs/2507.03724)]
_Memory OS for LLM agents with hybrid retrieval and cross-task skill reuse._
15. **[MemoryBear](https://www.memorybear.ai/)**
[[code](https://github.com/SuanmoSuanyangTechnology/MemoryBear)]
[[paper](https://arxiv.org/abs/2512.20651)]
_Memory framework providing human-like episodic and semantic recall to AI agents._
16. **[Honcho](https://honcho.dev/)**
[[code](https://github.com/plastic-labs/honcho)]
[[research](https://blog.plasticlabs.ai/research/)]
[[blog](https://blog.plasticlabs.ai/)]
[[eval](https://evals.honcho.dev/)]
_Memory library for stateful agents with a focus on user modeling._
17. **[engram (by Gentleman-Programming)](https://github.com/Gentleman-Programming/engram)**
[[code](https://github.com/Gentleman-Programming/engram)]
_Persistent memory for AI coding agents โ agent-agnostic single Go binary with SQLite + FTS5, exposed via MCP server, HTTP API, CLI, and TUI._
18. **[OpenMemory](https://openmemory.cavira.app/)**
[[code](https://github.com/caviraoss/openmemory)]
_Local persistent memory store for LLM apps (Claude Desktop, Copilot, Codex, etc.)._
19. **[memory-lancedb-pro](https://github.com/CortexReach/memory-lancedb-pro)**
[[code](https://github.com/CortexReach/memory-lancedb-pro)]
[[blog](https://lancedb.com/blog/openclaw-lancedb-memory-layer/)]
[[video](https://www.youtube.com/watch?v=bhuGrjuCM_g)]
_Enhanced [LanceDB](https://lancedb.com/) memory plugin for [OpenClaw](https://openclaw.ai/)_
20. **[MIRIX](https://mirix.io/)**
[[code](https://github.com/Mirix-AI/MIRIX)]
[[paper](https://arxiv.org/abs/2507.07957)]
[[blog](https://mirix.io/#/blog)]
_Multi-agent personal assistant that captures on-screen activity and consolidates it into structured memory._
21. **[MemMachine](https://memmachine.ai/)**
[[code](https://github.com/MemMachine/MemMachine)]
[[blog](https://memmachine.ai/blog/)]
_Interoperable memory layer providing extensible storage and retrieval primitives for AI agents._
22. **[Memobase](https://docs.memobase.io/)**
[[code](https://github.com/memodb-io/memobase)]
_User profile-based long-term memory for AI chatbot applications._
23. **[Memanto](https://memanto.ai/)** 
[[code](https://github.com/moorcheh-ai/memanto)]
[[paper](https://arxiv.org/abs/2604.22085)]
[[docs](https://docs.memanto.ai)]
_Typed semantic memory with `remember`/`recall`/`answer` operations and information-theoretic retrieval._
24. **[LangMem](https://langchain-ai.github.io/langmem/)**
[[code](https://github.com/langchain-ai/langmem)]
[[blog](https://blog.langchain.com/)]
_LangChain's memory primitives for storing, recalling, and managing agent state in LangGraph workflows._
25. **[Omnigraph](https://github.com/ModernRelay/omnigraph)**
[[code](https://github.com/ModernRelay/omnigraph)]
_Object-storage-native graph engine for agent memory with git-style branch/merge workflows._
26. **[PowerMem](https://www.powermem.ai)**
[[code](https://github.com/oceanbase/powermem)]
_Persistent, self-evolving memory for AI agents โ hybrid vector/full-text/graph retrieval with LLM-driven extraction, Ebbinghaus-style decay, and two-layer Experience + Skill distillation; from the OceanBase team._
27. **[Puppyone](https://www.puppyone.ai)**
[[code](https://github.com/puppyone-ai/puppyone)]
[[docs](https://www.puppyone.ai/doc)]
_Filesystem-shaped agent memory with auto-versioning, per-agent ACLs, and data connectors; accessible via MCP/REST/CLI._
28. **[Mem9](https://mem9.ai/)**
[[code](https://github.com/mem9-ai/mem9)]
[[blog](https://addozhang.medium.com/keep-memory-local-building-a-private-openclaw-memory-hub-with-mem9-tidb-5b305345b40a)]
_Local private memory hub for OpenClaw and similar coding agents._
29. **[deja](https://github.com/vshulcz/deja-vu)**
[[code](https://github.com/vshulcz/deja-vu)]
_Indexes the session transcripts twenty coding agents already write to disk, retroactively โ local BM25 recall over them, with credentials redacted at index time._
30. **[CodeAlmanac](https://github.com/AlmanacCode/codealmanac)**
[[code](https://github.com/AlmanacCode/codealmanac)]
_Repo-local Markdown wiki for AI coding agents that preserves project conversations, decisions, and implementation context._
31. **[projectmem](https://projectmem.dev)**
[[code](https://github.com/riponcm/projectmem)]
[[docs](https://projectmem.dev/guide)]
[[paper](https://arxiv.org/abs/2606.12329)]
_Local-first, event-sourced memory for AI coding agents: an append-only event log served via MCP, plus a pre-commit gate that warns before repeating a failed fix._
32. **[Memorix](https://github.com/AVIDS2/memorix)**
[[code](https://github.com/AVIDS2/memorix)]
_Local-first cross-agent memory layer for coding agents via MCP โ SQLite-backed project memory with observation, reasoning, and git-derived fact types, plus task-lensed context briefs._
33. **[HMS (Holographic Memory System)](https://github.com/Shadow-Weave/HMS)**
[[code](https://github.com/Shadow-Weave/HMS)]
_Long-term memory QA framework that wraps OpenAI clients with automatic recall and retain, PostgreSQL-backed, evaluated on LongMemEval._
34. **[Vestige](https://github.com/samvallad33/vestige)**
[[code](https://github.com/samvallad33/vestige)]
[[release](https://github.com/samvallad33/vestige/releases/tag/v2.1.23)]
_Local-first cognitive memory MCP server for coding agents, with FSRS-6 decay, spreading activation, active suppression, Receipt Lock, and an inspectable dashboard._
35. **[Compartment](https://github.com/MaxFreedomPollard/Compartment)**
[[code](https://github.com/MaxFreedomPollard/Compartment)]
_Offline, encrypted-at-rest vector memory for agents via MCP server, Python, or CLI; AEAD-encrypted embeddings, hybrid recall, per-record crypto-shred deletion, hash-chained audit log._
36. **[MisakaNet](https://github.com/Ikalus1988/MisakaNet)**
[[code](https://github.com/Ikalus1988/MisakaNet)]
[[wiki](https://github.com/Ikalus1988/MisakaNet/wiki)]
_Git-based distributed swarm memory; agents share lessons across nodes via GitHub Issues._
37. **[Caura (formerly MemClaw)](https://caura.ai)**
[[code](https://github.com/caura-ai/caura)]
[[blog](https://caura.ai/blog)]
_Governed shared memory for AI agent fleets โ cross-agent knowledge sharing with permissions, audit trails, and self-learning._
38. **[Mnemory](https://github.com/fpytloun/mnemory)** 
[[code](https://github.com/fpytloun/mnemory)]
_Multi-type agent memory (facts, preferences, episodic) with TTLs, user/agent scoping, and an MCP server._
39. **[Statewave](https://statewave.ai/)**
[[code](https://github.com/smaramwbc/statewave)]
[[docs](https://github.com/smaramwbc/statewave-docs)]
[[blog](https://www.statewave.ai/blog)]
_Open-source memory runtime for AI agents serving reproducible, provenance-tagged context bundles instead of query-time retrieval; self-hosted on Postgres + pgvector with Python/TypeScript SDKs._
40. **[OMEGA](https://omegamax.co)** 
[[code](https://github.com/omega-memory/omega-memory)]
[[blog](https://omegamax.co/blog)]
_MCP server exposing 25 memory tools for AI coding agents._
41. **[Belief Context Graph](https://bigai-nlco.github.io/bcg/)**
[[code](https://github.com/bigai-nlco/bcg)]
[[docs](https://belief-context-graph.docs.buildwithfern.com/)]
_Confidence-aware belief graph organizing long-horizon agent context into trackable, updatable belief states; includes SDK, graph-building tools, visualization UI, and agent benchmark._
42. **[Remnic](https://remnic.ai/)**
[[code](https://github.com/joshuaswarren/remnic)]
[[docs](https://remnic.ai/guides/)]
[[paper](https://doi.org/10.5281/zenodo.21922631)]
_Local-first Markdown memory shared across coding agents and MCP clients, with per-result provenance, correction workflows, and the MemCorrect benchmark._
43. **[Memov](https://www.memov.ai/)**
[[code](https://github.com/memovai/memov)]
_Git-based, traceable memory layer for Claude Code._
44. **[CommonGround Kernel](https://github.com/Intelligent-Internet/CommonGround)**
[[code](https://github.com/Intelligent-Internet/CommonGround)]
_PostgreSQL-backed shared work-record substrate for human-agent and multi-agent systems, with durable handoff facts, causal lineage, and pull-first recovery across runtimes._
๐ฑ Emerging projects โ open-source products with fewer than 100 GitHub stars, same format and ordering (click to expand)
45. **[causal-memory](https://github.com/JingxuanC/causal-memory)**
[[code](https://github.com/JingxuanC/causal-memory)]
[[eval](https://github.com/JingxuanC/causal-memory/tree/main/docs/benchmarks)]
_Local-first agent memory in Rust: facts and typed decisionโoutcome causal edges (caused/enabled/prevented) on one SQLite store, with inhibitory spreading activation, MCP server, CLI, Python bindings._
46. **[taOSmd](https://github.com/jaylfc/taosmd)**
[[code](https://github.com/jaylfc/taosmd)]
[[eval](https://github.com/jaylfc/taosmd/blob/master/docs/benchmarks.md)]
_Local-first, offline agent memory: an append-only transcript yields a typed temporal knowledge graph with source-grounded, verifier-checked facts and hybrid retrieval, tuned for small local models._
47. **[Wenlan](https://wenlan.app/)**
[[code](https://github.com/7xuanlu/wenlan)]
[[docs](https://wenlan.app/docs)]
_Local-first AI knowledge base and LLM wiki that distills agent work into source-cited pages and serves them to MCP clients._
48. **[Synap](https://maximem.ai)**
[[code](https://github.com/maximem-ai/maximem_synap_sdk)]
[[docs](https://docs.maximem.ai)]
_Long-term memory layer that extracts facts, preferences, episodes, and temporal events from conversations; integrates with most major agent frameworks._
49. **[LWC](https://janyork.github.io/llm-wiki-cli/)**
[[code](https://github.com/JanYork/llm-wiki-cli)]
_Agent-operated persistent memory CLI with source-cited Wiki pages, full-text search, document graphs, and CodeGraph indexes for cross-session project recall._
50. **[RetainDB](https://retaindb.com)**
[[code](https://github.com/RetainDB/RetainDB)]
_Local-first agent memory with noise filtering, typed facts, reuse-based reinforcement, and BM25 + vector + graph retrieval with RRF and reranking; Apache-2.0 core, BSL-1.1 server._
51. **[widemem-ai](https://widemem.ai)**
[[code](https://github.com/remete618/widemem-ai)]
_Lightweight memory layer with importance scoring, temporal decay, and 3-tier hierarchy._
52. **[memclaw (Felo)](https://memclaw.me)**
[[code](https://github.com/Felo-Inc/memclaw)]
_Persistent project memory for AI coding agents โ isolated per-project workspaces, a web dashboard to review what the agent remembers, and team collaboration._
53. **[Mi-Memory](https://darwin-agent.github.io/Mi-Memory/)**
[[code](https://github.com/Darwin-Agent/Mi-Memory)]
[[docs](https://darwin-agent.github.io/Mi-Memory/)]
[[paper](https://arxiv.org/abs/2607.18975)]
_Lifecycle memory framework for personal assistant agents from Xiaomi's Darwin Agent team; multi-source user state with provenance, editable correction and forgetting, device-adaptive deployment._
54. **[Data Olympus](https://github.com/knaisoma/data-olympus)**
[[code](https://github.com/knaisoma/data-olympus)]
_Governed project memory for AI coding agents: agents propose learnings, humans promote them, and MCP retrieval serves only in-force knowledge after validity and supersession checks._
55. **[Mnemoverse](https://mnemoverse.com)**
[[code](https://github.com/mnemoverse/mcp-memory-server)]
_Persistent memory API for agents over MCP: scores importance on write, builds Hebbian associations, and re-ranks recall from outcome feedback; managed engine, open MCP client._
56. **[ZenBrain](https://zensation.ai/en)**
[[code](https://github.com/zensation-ai/zenbrain)]
[[paper](https://arxiv.org/abs/2604.23878)]
_Neuroscience-inspired 7-layer memory architecture for autonomous agents in zero-dependency TypeScript, with FSRS spaced repetition, Hebbian learning, and sleep-cycle consolidation._
57. **[Selvedge](https://selvedge.sh)**
[[code](https://github.com/masondelan/selvedge)]
[[docs](https://selvedge.sh/reference/compatibility/)]
_Local decision memory for coding agents: records code decisions, rejected approaches and revisit conditions in SQLite; retrieved through MCP or CLI._
58. **[InvMem](https://github.com/wenxiaof345-ctrl/vanilla-rag-memory)**
[[code](https://github.com/wenxiaof345-ctrl/vanilla-rag-memory)]
_Vanilla RAG baseline (chunking, embeddings, FAISS/SQLite vector store) behind a synchronous Add/Search API; #1, Agent Memory Leaderboard (academic textual, 2026-08)._
59. **[Fidelis Memory](https://hermes-labs.ai/fidelis)**
[[code](https://github.com/hermes-labs-ai/fidelis)]
[[eval](https://github.com/hermes-labs-ai/fidelis/blob/b938676affd47f2a6e0a5106c44336fa11f4e92f/bench/results-default-0.3.0rc1.json)]
_Local-first memory for coding agents: MCP recall returns original passages verbatim via vector search, with an explicit BM25/RRF hybrid mode; no generative LLM in retrieval._
60. **[Tree Ring Memory](https://terminallylazy.github.io/Tree-Ring-Memory/)**
[[code](https://github.com/TerminallyLazy/Tree-Ring-Memory)]
_Local-first memory lifecycle for AI agents with a Rust CLI, SQLite/FTS recall, audit, forgetting, consolidation, and Ratatui TUI._
61. **[GoodMemory](https://github.com/hjqcan/GoodMemory)**
[[code](https://github.com/hjqcan/GoodMemory)]
[[docs](https://github.com/hjqcan/GoodMemory#quickstart-codex-or-claude-code-memory)]
_Local-first, auditable memory layer for AI agents and coding hosts, with durable SQLite, embedding-free recall, MCP access, and opt-in governed writeback._
62. **[ReFind](https://github.com/imlrz/ReFind)**
[[code](https://github.com/imlrz/ReFind)]
_Memory retriever that plans iterative searches over a conversation-level BM25 index and returns contextual evidence blocks; #2, Agent Memory Leaderboard (academic textual, 2026-08)._
63. **[A3M Router](https://github.com/Das-rebel/a3m-router)**
[[code](https://github.com/Das-rebel/a3m-router)]
_Multi-model LLM router with persistent memory (MemoryTree), cross-session context-window management, conversation memory with semantic recall, and ObsidianVault integration._
64. **[Lians agent memory](https://www.lians.ai/)**
[[code](https://github.com/Lians-ai/Lians)]
[[eval](https://github.com/Lians-ai/Lians/blob/master/docs/benchmark.md)]
_Bitemporal agent memory with deterministic supersession, point-in-time recall, MCP access, audit trails, and local SQLite or PostgreSQL storage._
65. **[Agentic Task System](https://github.com/renezander030/agentic-task-system)**
[[code](https://github.com/renezander030/agentic-task-system)]
_Agent-native context layer over your existing task app (TickTick; Notion/Obsidian planned), exposing hybrid retrieval over tasks/notes to agents via a CLI with pluggable storage adapters._
66. **[myc](https://aistastudio.github.io/myc/)**
[[code](https://github.com/aistastudio/myc)]
[[docs](https://aistastudio.github.io/myc/)]
_Local task-and-memory layer for coding agents: decision oplog with hybrid search, session/repo-scoped recall, PreCompact hook saving the episode before compaction; Bun + SQLite._
67. **[FluctlightDB](https://github.com/voxmastery/FluctlightDB)**
[[code](https://github.com/voxmastery/FluctlightDB)]
[[paper](https://doi.org/10.5281/zenodo.20949890)]
_Embedded database engine for AI agents with `experience()`/`activate()` API and reproducible LoCoMo evaluation._
68. **[AtMem](https://github.com/aetna000/atmem)**
[[code](https://github.com/aetna000/atmem)]
_Local-first agent memory with governed retrieval, provenance, lifecycle controls, delegated context delivery, execution evidence, SQLite storage, MCP, and an audit dashboard._
69. **[Lint-AI](https://github.com/RooAGI/Lint-AI)**
[[code](https://github.com/RooAGI/Lint-AI)]
_Agent memory and semantic review engine with lexical, temporal, and graph-aware retrieval across conversations, documents, code, and agent sessions._
70. **[Lockstep](https://www.getlockstep.dev)**
[[code](https://github.com/lockstep-team-agent/lockstep)]
_Shared decision ledger for teams using AI coding agents: records accepted decisions, rationale, and rejected options, and briefs new agent sessions via MCP._
71. **[chamnan](https://github.com/ArcticFox2029/chamnan)**
[[code](https://github.com/ArcticFox2029/chamnan)]
[[data](https://github.com/ArcticFox2029/chamnan-corpus)]
_Repository-local context for coding agents: an architecture index, impact map, and decision records committed beside the code._
72. **[elephant](https://github.com/tonone-ai/elephant)**
[[code](https://github.com/tonone-ai/elephant)]
_Claude Code plugin keeping per-repo memory in a committed `ELEPHANT.md` plus a global cross-repo file; hooks load both at session start and prompt saves._
73. **[hermeneutic](https://hermes-labs.ai/hermeneutic)**
[[code](https://github.com/hermes-labs-ai/hermeneutic)]
_Turns corrections from AI agent logs into guidance for similar tasks: local semantic memory with prompt-context hooks and response checks._
74. **[sqlite-graph-memory](https://github.com/Palo-Alto-AI-Research-Lab/sqlite-graph-memory)**
[[code](https://github.com/Palo-Alto-AI-Research-Lab/sqlite-graph-memory)]
_Graph RAG memory for agents over a markdown vault: dense retrieval, hand-curated wikilink 1-hop expansion, cross-encoder rerank, per-turn SQLite ledger._
75. **[ActiveMemoryIndex](https://github.com/linxuhao/ActiveMemoryIndex)**
[[code](https://github.com/linxuhao/ActiveMemoryIndex)]
_Dual store of verbatim timestamped turns and atomic first-person facts, retrieved in the same register; #3, Agent Memory Leaderboard (academic textual, 2026-08)._
76. **[Graphmem](https://github.com/sonic182/graphmem)**
[[code](https://github.com/sonic182/graphmem)]
_Local-first memory for coding agents in Rust: scoped memories and an entity graph in SQLite, recalled via local embeddings plus Personalized PageRank; MCP server._
77. **[inspeximus (formerly mnemo)](https://dancenitra.github.io/inspeximus/)**
[[code](https://github.com/DanceNitra/inspeximus)]
_Zero-dependency memory layer and MCP server with value-ranked recall, per-type decay, keyed supersession, revert-based correction, signed provenance, tamper-evident receipts, and cross-store erasure._
78. **[skillmem](https://skillmem.dev)**
[[code](https://github.com/liza-studio/skillmem)]
_Skill memory for Claude Code and Codex agents: records how tasks were solved, recalls them, reinforces what works, decays the rest; SQLite + FTS5 MCP server._
79. **[Synapse](https://github.com/anshulyadav1976/synapse)**
[[code](https://github.com/anshulyadav1976/synapse)]
_Zero-dependency Markdown memory vault with ChatGPT/Claude imports, SQLite keyword and optional semantic search, linked wiki pages, MCP retrieval, and human-approved agent notes._
80. **[Hyperconsciousness](https://github.com/louis030195/hyperconsciousness)**
[[code](https://github.com/louis030195/hyperconsciousness)]
_Encrypted, append-only knowledge store for humans and agents: signed records sync across devices and are exposed through scoped, expiring grants over MCP._
81. **[archon-memory-core](https://github.com/atw4757-byte/archon-memory-core)**
[[code](https://github.com/atw4757-byte/archon-memory-core)]
_Local-first agent memory with nightly consolidation, active forgetting, and salience scoring._
82. **[birkin-mnemosyne](https://github.com/ashmoonori-afk/birkin-mnemosyne)**
[[code](https://github.com/ashmoonori-afk/birkin-mnemosyne)]
_Stdlib-only Python memory for agents: Markdown vault with BM25, Korean bigrams and usage decay; model curation passes through a deterministic safety-clamping executor; optional MCP server._
83. **[Agent Knowledge Cycle](https://github.com/shimo4228/agent-knowledge-cycle)**
[[code](https://github.com/shimo4228/agent-knowledge-cycle)]
[[paper](https://doi.org/10.5281/zenodo.20578272)]
_Six-phase knowledge cycle specification (ADRs, JSON schemas, reference implementation) that turns coding-agent sessions into persistent skills, rules, and memory._
84. **[kgai](https://kgai.dev)**
[[code](https://github.com/kgaidev/kgai)]
_Local-first immutable knowledge graph of engineering decisions for AI coding agents; superseded decisions and rejected approaches stay queryable; embedded graph DB, opt-in team sync._
85. **[Talamus](https://ampres-ai.github.io/talamus/)**
[[code](https://github.com/ampres-ai/talamus)]
[[docs](https://ampres-ai.github.io/talamus/)]
[[eval](https://ampres-ai.github.io/talamus/benchmarks/)]
_Local-first agent memory that stores source-grounded Markdown, preserves bitemporal history and provenance, and exposes search, recall, and review-gated correction through MCP._
86. **[kannaka-memory](https://github.com/kannaka-labs/kannaka-memory)**
[[code](https://github.com/kannaka-labs/kannaka-memory)]
[[eval](https://github.com/kannaka-labs/kannaka-bench/blob/master/RESULTS.md)]
_Rust agent memory as a wave-interference medium: phase-coded wavefronts, bilateral hemispheres, dream consolidation and forgetting, NATS swarm sync; CLI and MCP plugin._
87. **[agent-memory-doctor](https://github.com/chenhz01/agent-memory-doctor)**
[[code](https://github.com/chenhz01/agent-memory-doctor)]
_Boot-time integrity checker for agent memory files and SQLite session stores: fingerprint and hash checks, size and freshness limits, archive verification; CLI and GitHub Action._
88. **[Mnemosyne](https://github.com/ElonAug7/Mnemosyne-agentmemory-engine-openclaw-hermes)**
[[code](https://github.com/ElonAug7/Mnemosyne-agentmemory-engine-openclaw-hermes)]
_Memory engine for OpenClaw and Hermes: JSONL and Markdown layers ranked by weighted importance, recency, keyword and hit-frequency scoring; offline by default, DashScope embeddings opt-in._
89. **[PackRat](https://github.com/kevdogg102396-afk/packrat)**
[[code](https://github.com/kevdogg102396-afk/packrat)]
_Auto-learning codebook compression that shrinks agent context files while keeping them LLM-readable._
90. **[Hybrid Episodic Memory](https://github.com/tlysanhuo/agent-memory-challenge)**
[[code](https://github.com/tlysanhuo/agent-memory-challenge)]
_Deterministic weighted reciprocal-rank fusion of BM25 and dense retrieval over raw conversational turns, no LLM in the path; #6, Agent Memory Leaderboard (academic textual, 2026-08)._
91. **[Verified Memory Vault](https://github.com/secondbrainstarter/verified-memory-vault)**
[[code](https://github.com/secondbrainstarter/verified-memory-vault)]
_Obsidian vault doubling as Claude Code memory: deterministic health-score linter (undated entries, duplicates, dead links) plus a git pre-commit hook refusing mass deletions._
92. **[MemTether](https://github.com/MemTether/MemTether)**
[[code](https://github.com/MemTether/MemTether)]
_Shared SQLite memory hub for AI clients over MCP, CLI and REST: corrections supersede rather than delete, bi-temporal timestamps, hash-anchored audit log, human conflict adjudication._
93. **[memgres](https://github.com/mozgsml/memgres)**
[[code](https://github.com/mozgsml/memgres)]
_Versioned document memory for AI agents over one Postgres; lexical or semantic recall, diff-based history, git-blame line attribution, GDPR-erasable, multi-tenant via MCP/HTTP._
94. **[FlowGrid AML Retriever](https://github.com/dlxeva/flowgrid-aml-retriever)**
[[code](https://github.com/dlxeva/flowgrid-aml-retriever)]
_Deterministic, evidence-first Add/Search retriever that stores every original message and returns ranked, traceable source evidence; #8, Agent Memory Leaderboard (academic textual, 2026-08)._
95. **[engram (by FBISiri)](https://github.com/FBISiri/engram)**
[[code](https://github.com/FBISiri/engram)]
_Go memory service on Qdrant with write-time dedup and importance gating, type-based TTL decay, reflection into insights, and MCP plus REST interfaces._
96. **[AML Memory MVP](https://github.com/0xboyu/aml-memory-mvp)**
[[code](https://github.com/0xboyu/aml-memory-mvp)]
_Evidence-only, typo-tolerant retriever over English and CJK text using SQLite FTS5, character n-grams, and conversation-neighbor expansion; #10, Agent Memory Leaderboard (academic textual, 2026-08)._
97. **[ExperienceNet](https://github.com/cu-min/experiencenet)**
[[code](https://github.com/cu-min/experiencenet)]
[[docs](https://github.com/cu-min/experiencenet/blob/master/docs/API.md)]
_Self-hosted experience network for agents: search and write real technical attempts (problem/conditions/action/outcome), lexical + semantic hybrid retrieval over PostgreSQL/pgvector, gap capture, reuse feedback._
98. **[claude-memory-tidy](https://github.com/tonydzi/claude-memory-tidy)**
[[code](https://github.com/tonydzi/claude-memory-tidy)]
_Maintenance layer for always-loaded agent memory files: deterministic budget guard, orphan-note coverage, and verbatim folding into warm sub-indexes, guarding against silent truncation._
99. **[YYLO Ledger](https://github.com/yylo-dev/yylo-ledger)**
[[code](https://github.com/yylo-dev/yylo-ledger)]
_Git-native task and Record store for coding-agent workflows: reviewable current state, append-only history, dependency-aware work, and bounded queries._
100. **[Akephalos](https://github.com/daveinturkey15-byte/akephalos)**
[[code](https://github.com/daveinturkey15-byte/akephalos)]
_Local-first, markdown-based portable agent profile (preferences, rules, durable memories) synced across agents via plain files and Git._
101. **[ๆบฏๅฟ (Suyi)](https://github.com/xiaofanliu525-ctrl/suyi-memory)**
[[code](https://github.com/xiaofanliu525-ctrl/suyi-memory)]
_Dual-temporal memory engine for AI agents โ SQLite-backed, zero-dependency, Ebbinghaus-decayed fact storage with skill crystallization._
102. **[Panella](https://panella.tech)**
[[code](https://github.com/panellatech/panella)]
_Self-hosted governed memory over MCP; agent writes become durable only after human approval with verifiable receipts; Apache-2.0._
103. **[Hybrid Memory Search](https://github.com/cydd-1972/hybrid_search)**
[[code](https://github.com/cydd-1972/hybrid_search)]
_Local hybrid-retrieval memory service with per-user isolation, synchronous embedding on write, and fused BM25/dense ranking; #4, Agent Memory Leaderboard (academic textual, 2026-08)._
104. **[ChronoHybridMem](https://github.com/Tin11Mn/chrono-hybrid-mem)**
[[code](https://github.com/Tin11Mn/chrono-hybrid-mem)]
_Evidence-only textual memory over SQLite FTS5 with optional LLM fact extraction and multi-route candidate recall; #5, Agent Memory Leaderboard (academic textual, 2026-08)._
105. **[Chronicle Memory](https://github.com/simple-boy/Chronicle-Memory)**
[[code](https://github.com/simple-boy/Chronicle-Memory)]
_Evidence-only memory over SQLite with a hybrid lexical scorer adding phrase, temporal, and session-diversity bonuses; #7, Agent Memory Leaderboard (academic textual, 2026-08)._
106. **[MemoryAgent](https://github.com/llLAlisa/memory-agent-submission)**
[[code](https://github.com/llLAlisa/memory-agent-submission)]
_FastAPI + ChromaDB memory system with local sentence-transformers embeddings and similarity-based write deduplication; #9 as LLLMemoryAgent, Agent Memory Leaderboard (academic textual, 2026-08)._
107. **[FeedRecall](https://github.com/Paoladev45/feedrecall)**
[[code](https://github.com/Paoladev45/feedrecall)]
_Local-first MCP memory for saved social discoveries, with source dates, project relevance, evidence lifecycle, timelines, and bounded recall for coding agents._
108. **[RCLL](https://rcll.ai)**
[[code](https://github.com/holetron-lab/fleet-memory)]
[[docs](https://rcll.ai/docs/)]
[[eval](https://rcll.ai/docs/benchmarks/)]
_Self-hosted shared memory for a fleet of agents: topic rooms, L0โL3 depth, Postgres/pgvector; the read path invokes no language model. Fork of Hindsight._
109. **[notebook.py](https://github.com/minjimindypark/llmcompressor)**
[[code](https://github.com/minjimindypark/llmcompressor)]
_Single-file Python tool for Claude Code transcript memory, stored as editable Markdown with source references and retained history of retired entries._
110. **[kith](https://github.com/theNamek/kith)**
[[code](https://github.com/theNamek/kith)]
[[eval](https://github.com/theNamek/kith/tree/main/examples/delegation_sim)]
_Relationship memory between agents on SQLite: append-only observations derive trust, reliability, and sentiment views with per-entry visibility scopes._
111. **[alethech](https://alethech.alicelabs.site/)**
[[code](https://github.com/eddyflores100-lang/alethech)]
_Verifiable agent continuity in Python: Ed25519-signed memory commits in a hash-linked DAG, identity-preserving key rotation, encrypted portable history; local-first, zero-LLM._
### Closed-Source
- [MemoraX](https://memorax.ai/)
[[spec](https://memorax.ai/spec/)]
[[platform](https://platform.memorax.net/)]
_Memory layer for long-horizon agents from MemoraX AI; core system is API-only; #1, Agent Memory Leaderboard (industry textual, 2026-08)._
- [MemoraX Code](https://code.memorax.net/)
[[code](https://github.com/memorax-ai/memorax-code)]
_Coding-agent memory productโnot a text-chatbot memory layerโthat carries engineering experience, repository knowledge, preferences, and procedures across tasks and sessions._
- [MemoryLake](https://www.memorylake.ai/en)
[[blog](https://www.memorylake.ai/en/blogs)]
- [Supermemory](https://supermemory.ai/)
[[code](https://github.com/supermemoryai/supermemory)]
[[docs](https://supermemory.ai/docs)]
_Memory API, hosted or free self-hosted; SDKs, MCP server, and dashboard are MIT, but the memory engine ships only as a prebuilt binary._
- [Memories.ai](https://memories.ai/)
[[research](https://memories.ai/research)]
[[paper](https://memories.ai/research/Camera)]
[[blog](https://memories.ai/blogs)]
- [Macaron Mind Lab](https://macaron.im/mindlab)
[[blog](https://macaron.im/mindlab/research)]
[[paper](https://macaron.im/mindlab/publications)]
- [Mem 2.0](https://get.mem.ai/)
[[blog](https://get.mem.ai/blog)]
- [M-Flow](https://m-flow.ai/)
- [TwinMind](https://twinmind.com/)
[[blog](https://twinmind.com/blogs)]
- [Penfield](https://www.penfield.app/)
[[blog](https://penfieldlabs.substack.com/)]
- [Sonzai](https://sonz.ai/)
- [Threadline](https://threadline.to)
[[partial-code](https://github.com/vidursharma202-del/threadline-mcp)]
[[schema](https://github.com/vidursharma202-del/context-schema)]
[[docs](https://threadline.to/docs)]
- [Remio](https://remio.ai/)
_Local-first personal knowledge base that indexes files, webpages, recordings, notes, emails, and messages for agent retrieval via search and RAG._
- [AccInt](https://accint.xyz)
[[partial-code](https://github.com/maxbaluev/accreted-intelligence)]
_Local-first MCP Work Model for coding agents that retrieves scored memory, records actions, and credits real outcomes; engine is a closed-source binary._
- [Agentage Memory](https://memory.agentage.io)
_Remote MCP memory server (OAuth 2.1 + PKCE + DCR) giving Claude, Cursor, and ChatGPT one shared markdown memory mirrored locally as files you own._
- [screenpipe](https://screenpipe.com)
[[source-available](https://github.com/screenpipe/screenpipe)]
[[license](https://github.com/screenpipe/screenpipe/blob/main/LICENSE.md)]
[[docs](https://docs.screenpi.pe)]
_Local-first work memory that captures screen, audio, input, browser, and meeting context for search and agent retrieval._
- [Firekeep](https://firekeep.ai/)
[[source-available](https://github.com/kapella-hub/FirekeepHQ)]
[[license](https://github.com/kapella-hub/FirekeepHQ/blob/main/LICENSE)]
[[docs](https://firekeep.ai/docs.html)]
_Self-hosted shared memory, working context, coordination, and evidence for Claude Code, Codex, Kiro, OpenCode, and other MCP clients._
- [ORANO](https://oranoai.com/)
[[docs](https://oranoai.com/mcp)]
[[blog](https://oranoai.com/blog/personal-mcp-context-ai-agents.html)]
_Consumer app that distills saved Reels, videos, articles, and PDFs into projects and memory facts that the user's agent reads over a read-only MCP server._
- [Wontopos (Tablet 2)](https://wontopos.com/)
[[paper](https://arxiv.org/abs/2608.23920)]
[[eval](https://github.com/wontopos/beam1m-tablet-2)]
_Memory API with no language model in the retrieval path; paper-reported 95.7% LongMemEval-S, 95.2% recall@5 over 70 language pairs, and BEAM-1M 67.5% with published harness._
- [Perseus Vault (formerly Mimir)](https://perseus.observer/vault/)
[[docs](https://perseus.observer/vault/mcp-reference/)]
_Local MCP memory server as a single Rust binary: SQLite FTS5 plus vector hybrid search, AES-256-GCM at rest; GitHub source repository not publicly accessible (2026-09)._
- [ContextStream](https://contextstream.io)
[[partial-code](https://github.com/contextstream/mcp-server)]
[[docs](https://contextstream.io/docs/mcp)]
[[eval](https://contextstream.io/benchmarks)]
_Hosted MCP context layer for coding agents: persistent decisions and lessons, semantic code search, post-compaction recovery; MIT-licensed Rust client, hosted backend closed-source._
- [Sentra](https://www.sentra.app/)
_Organization-wide memory layer for teams and agents: ingests meetings, mail, tickets, and repositories into a bi-temporal fact graph served over REST and MCP._
- [Moraine Home](https://github.com/ceniran/moraine-home)
[[source-available](https://github.com/ceniran/moraine-home)]
[[license](https://github.com/ceniran/moraine-home/blob/main/LICENSE)]
_Local-first memory workbench for personal agents and companions: hybrid keyword/vector retrieval, human-and-agent review of candidates, reversible consolidation, timelines, and MCP/HTTP interfaces._
- [ByteRover](https://www.byterover.dev/)
[[source-available](https://github.com/campfirein/byterover-cli)]
[[license](https://github.com/campfirein/byterover-cli/blob/main/LICENSE)]
[[paper](https://arxiv.org/abs/2604.01599)]
[[docs](https://docs.byterover.dev/)]
_LLM-curated hierarchical context tree for coding agents, with git-like branching, cloud sync, and MCP; formerly Cipher._
- [Context Mode](https://context-mode.com/)
[[source-available](https://github.com/mksglu/context-mode)]
[[license](https://github.com/mksglu/context-mode/blob/main/LICENSE)]
_Context-window optimization for AI coding agents: diverts large tool outputs into a locally searchable store and persists session memory across platforms via MCP and hooks._
- [Agent QA](https://vostride.com/docs/agent-qa)
[[source-available](https://github.com/vostride/agent-qa)]
[[license](https://github.com/vostride/agent-qa/blob/main/LICENSE.md)]
_QA agent that retains persistent test memory to reuse prior runs and self-heal natural-language web and mobile tests._
- [RE-call](https://github.com/GiulioDER/RE-call)
[[source-available](https://github.com/GiulioDER/RE-call)]
[[license](https://github.com/GiulioDER/RE-call/blob/master/LICENSE)]
[[docs](https://github.com/GiulioDER/RE-call/blob/master/docs/USING_WITH_CLAUDE.md)]
[[eval](https://github.com/GiulioDER/RE-call/blob/master/results/FINDINGS.md)]
_Postgres plus pgvector memory retrieval for AI agents, with provenance, trust verdicts, tenant isolation, MCP access, and abstention when evidence is insufficient._
- [past.dev](https://past.dev/)
[[docs](https://past.dev/docs/memory-api/overview)]
[[eval](https://github.com/pastdotdev/benchmarks)]
_Temporal memory API: stores timestamped text with its source; recall returns facts current at a given time and what each replaced; REST and remote MCP._
- [Monolithos Robot Brain](https://robot-brain.monolithos.ai/)
[[docs](https://github.com/silas-zhen/robot-brain/blob/main/docs/en/integration.md)]
_Long-term memory and experience layer for embodied AI; public Alpha under an evaluation license, core ships as a compiled package; HTTP API and Python SDK._
- [Remnant](https://remnant.dedale-bi.com/)
[[docs](https://github.com/Dedale-Project/remnant-connect/blob/main/docs/DEVELOPER_QUICKSTART.md)]
_Hosted collective memory over MCP: agents search other agents' published technical experience, inspect provenance, record outcomes, and contribute lessons through a separately authorized connection._
- [Marvnor](https://wendelxia.github.io/marvnor/)
[[docs](https://github.com/wendelxia/marvnor/blob/main/PUBLIC_EVALUATION_API.md)]
_Hosted structured memory API for AI agents, with fact verification, conflicting-value detection, record correction, and targeted deletion._
### Archival
_Projects that are inactive or whose claims have been disputed by third parties. Status labels link to the evidence and note when the status was last checked._
- [MemPalace](https://github.com/MemPalace/mempalace) โ ๏ธ Disputed (third-party critiques challenge the project's core claims; last checked 2026-07)
[[code](https://github.com/milla-jovovich/mempalace)]
[[critique1](https://www.youtube.com/watch?v=WlxNNvDHJkE), [critique2](https://penfieldlabs.substack.com/p/milla-jovovich-just-released-an-ai)]
_Developed by actress [Milla Jovovich](https://en.wikipedia.org/wiki/Milla_Jovovich) and her friends_
- [Memvid](https://www.memvid.com/) โ ๏ธ Disputed (technical critique raised in GitHub issues, since deleted but archived; last checked 2026-07)
[[code](https://github.com/Olow304/memvid)]
[[critique (archived)](https://web.archive.org/web/20250807093442/https://github.com/Olow304/memvid/issues/49)]
- [Memary](https://kingjulio8238.github.io/memarydocs/) โ๏ธ Inactive (no significant development activity; last checked 2026-07)
[[code](https://github.com/kingjulio8238/memary)]
---
## ๐ Tutorials
#### ๐๏ธ 2026
- **[Agent Memory Techniques](https://github.com/NirDiamant/Agent_Memory_Techniques)** (NirDiamant): 30 runnable Jupyter notebooks covering conversation buffers, vector stores, knowledge graphs, episodic and semantic memory, Mem0, MemGPT/Letta, Zep, Graphiti, and LoCoMo benchmarks
[[code](https://github.com/NirDiamant/Agent_Memory_Techniques)]
- **[Choose an agent-memory architecture](https://sir-ad.github.io/awesome-memory/guide.html)** (sir-ad): Decision guide mapping four memory jobs to five architecture patterns, minimum controls, evaluation baselines, and primary research.
- **[Tools, Actions, Memory, and Context](https://books.bloo-mind.ai/masact/ch-04-tools-actions-environments)** โ : Chapter 4 of the textbook _[Multi-Agent Systems: A Contemporary Treatment](https://books.bloo-mind.ai/masact/)_.
#### ๐๏ธ 2025
- **[ACM SIGIR-AP 2025](https://www.sigir-ap.org/sigir-ap-2025/) Tutorial: [Conversational Agents: From RAG to LTM](https://sites.google.com/view/ltm-tutorial)** โ
[[paper](https://dl.acm.org/doi/10.1145/3767695.3769671)]
[[code](https://github.com/TeleAI-UAGI/Awesome-Agent-Memory)]
- Daily Dose of DS: A Practical Deep Dive Into Memory Optimization for Agentic Systems
[[Part-A](https://www.dailydoseofds.com/ai-agents-crash-course-part-15-with-implementation/)]
[[Part-B](https://www.dailydoseofds.com/ai-agents-crash-course-part-16-with-implementation/)]
[[Part-C](https://www.dailydoseofds.com/ai-agents-crash-course-part-17-with-implementation/)]
---
## ๐ Surveys
#### ๐๏ธ 2026
- **[Rethinking Memory Mechanisms of Foundation Agents in the Second Half: A Survey](https://arxiv.org/abs/2602.06052)**
[[code](https://github.com/AgentMemoryWorld/Awesome-Agent-Memory)]
- **[Memory in the LLM Era: Modular Architectures and Strategies within a Unified Framework](https://arxiv.org/abs/2604.01707)**
[[code](https://github.com/Yanchen398/Memory-in-the-LLM-Era)]
- **[From Storage to Experience: A Survey on the Evolution of LLM Agent Memory Mechanisms](https://arxiv.org/abs/2605.06716)**
[[code](https://github.com/FeishuLuo/Evolving-LLM-Agent-Memory-Survey)]
- **[Toward Efficient Agents: Memory, Tool Learning, and Planning](https://arxiv.org/abs/2601.14192)**
[[code](https://github.com/yxf203/Awesome-Efficient-Agents)]
- **[Anatomy of Agentic Memory: Taxonomy and Empirical Analysis of Evaluation and System Limitations](https://arxiv.org/abs/2602.19320)**
[[code](https://github.com/FredJiang0324/Anatomy-of-Agentic-Memory)]
- [Memory for Large Language Models](https://arxiv.org/abs/2607.25380)
- [LLM Agent Memory: A Survey from a Unified RepresentationโManagement Perspective](https://www.preprints.org/manuscript/202603.0359)
- [Memory for Autonomous LLM Agents: Mechanisms, Evaluation, and Emerging Frontiers](https://arxiv.org/abs/2603.07670)
- [Externalization in LLM Agents: A Unified Review of Memory, Skills, Protocols and Harness Engineering](https://arxiv.org/abs/2604.08224)
- [Survey on AI Memory: Theories, Taxonomies, Evaluations, and Emerging Trends](https://github.com/BAI-LAB/Survey-on-AI-Memory/blob/main/Survey%20on%20AI%20Memory.pdf)
- [The AI Hippocampus: How Far are We From Human Memory?](https://arxiv.org/abs/2601.09113)
#### ๐๏ธ 2025
- **[AI Meets Brain: Memory Systems from Cognitive Neuroscience to Autonomous Agents](https://arxiv.org/abs/2512.23343)**
[[code](https://github.com/AgentMemory/Huaman-Agent-Memory)]
- **[Memory in the Age of AI Agents](https://arxiv.org/abs/2512.13564)**
[[code](https://github.com/Shichun-Liu/Agent-Memory-Paper-List)]
- **[Rethinking Memory in AI: Taxonomy, Operations, Topics, and Future Directions](https://arxiv.org/abs/2505.00675)**
[[code](https://github.com/Elvin-Yiming-Du/Survey_Memory_in_AI)]
- [From Human Memory to AI Memory: A Survey on Memory Mechanisms in the Era of LLMs](https://arxiv.org/abs/2504.15965)
- [Cognitive Memory in Large Language Models](https://arxiv.org/abs/2504.02441)
- [Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems (Chapter 3)](https://arxiv.org/abs/2504.01990)
- [Human-inspired Perspectives: A Survey on AI Long-term Memory](https://arxiv.org/abs/2411.00489)
#### ๐๏ธ 2024
- **[A Survey on the Memory Mechanism of Large Language Model based Agents](https://arxiv.org/abs/2404.13501)**
[[code](https://github.com/nuster1128/LLM_Agent_Memory_Survey)]
---
## ๐ Benchmarks
### ๐ฌ Plain-Text Benchmarks
#### ๐๏ธ 2026
- **[Agent Memory Leaderboard](https://agentmemories.ai/home)**
[[code](https://github.com/AML-memory/agent-memory-leaderboard)]
_Public evaluation platform: participants expose Add/Search APIs and are scored on textual-memory and coding-agent-memory tracks._
- **[Agent Memory Benchmark (AMB)](https://agentmemorybenchmark.ai/)**
[[code](https://github.com/vectorize-io/agent-memory-benchmark)]
_Open harness and leaderboard scoring memory providers on accuracy, latency, and token cost over six datasets; built by Vectorize, whose Hindsight is among the providers._
- **OmniMemEval**
[[code](https://github.com/MemTensor/OmniMemEval)]
- **[Are We Ready For An Agent-Native Memory System?](https://arxiv.org/abs/2606.24775)**
(The MemoryData Paper)
[[code](https://github.com/OpenDataBox/MemoryData)]
- **[Locomo-Plus: Beyond-Factual Cognitive Memory Evaluation Framework for LLM Agents](https://arxiv.org/abs/2602.10715)**
[[code](https://github.com/xjtuleeyf/Locomo-Plus)]
- **LoCoMo Refined: Recalibrating LoCoMo with Stricter LLM Judging and A Cleaned Dataset**
[[code](https://github.com/mem-eval-suite/LoCoMo_refined)]
- **Agent-Memory Integrity Benchmark**
[[code](https://github.com/DanceNitra/agent-memory-integrity)]
- **[Cost and Accuracy of Long-Term Memory in Distributed Multi-Agent Systems Based on Large Language Models](https://arxiv.org/abs/2601.07978)**
[[code](https://github.com/wolffbe/dmas-memory)]
#### ๐๏ธ 2025
- **[Beyond a Million Tokens: Benchmarking and Enhancing Long-Term Memory in LLMs](https://arxiv.org/abs/2510.27246)**
(The BEAM Paper)
[[code](https://github.com/mohammadtavakoli78/BEAM)]
[[data](https://huggingface.co/datasets/Mohammadta/BEAM)]
- **[MOOM: Maintenance, Organization and Optimization of Memory in Ultra-Long Role-Playing Dialogues](https://arxiv.org/abs/2509.11860)**
(The ZH-4O Paper)
[[code](https://github.com/cows21/MOOM-Roleplay-Dialogue)]
[[data](https://github.com/cows21/MOOM-Roleplay-Dialogue/tree/main/data)]
- **[Know Me, Respond to Me: Benchmarking LLMs for Dynamic User Profiling and Personalized Responses at Scale](https://arxiv.org/abs/2504.14225)**
(The PersonaMem and ImplicitPersona Paper)
[[code](https://github.com/bowen-upenn/PersonaMem)]
[[data1](https://huggingface.co/datasets/bowen-upenn/PersonaMem)]
[[data2](https://huggingface.co/datasets/bowen-upenn/ImplicitPersona)]
- **[Evaluating Memory in LLM Agents via Incremental Multi-Turn Interactions](https://arxiv.org/abs/2507.05257)**
(The MemoryAgentBench Paper)
[[code](https://github.com/HUST-AI-HYZ/MemoryAgentBench)]
[[data](https://huggingface.co/datasets/ai-hyz/MemoryAgentBench)]
- **[LifelongAgentBench: Evaluating LLM Agents as Lifelong Learners](https://arxiv.org/abs/2505.11942)**
[[code](https://github.com/caixd-220529/LifelongAgentBench)]
[[data](https://huggingface.co/datasets/csyq/LifelongAgentBench)]
- **[NoLiMa: Long-Context Evaluation Beyond Literal Matching](https://arxiv.org/abs/2502.05167)**
[[code](https://github.com/adobe-research/NoLiMa)]
[[data](https://github.com/adobe-research/NoLiMa/tree/main/data)]
- **[HaluMem: Evaluating Hallucinations in Memory Systems of Agents](https://arxiv.org/abs/2511.03506)**
[[code](https://github.com/MemTensor/HaluMem)]
[[data](https://huggingface.co/datasets/IAAR-Shanghai/HaluMem)]
- **[LongBench v2: Towards Deeper Understanding and Reasoning on Realistic Long-context Multitasks](https://arxiv.org/abs/2412.15204)**
[[code](https://github.com/THUDM/LongBench)]
- **[Minerva: A Programmable Memory Test Benchmark for Language Models](https://arxiv.org/abs/2502.03358)**
[[code](https://github.com/microsoft/minerva_memory_test)]
- **[MemBench: Towards More Comprehensive Evaluation on the Memory of LLM-based Agents](https://arxiv.org/abs/2506.21605)**
[[code](https://github.com/import-myself/Membench)]
- [Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory](https://arxiv.org/abs/2511.20857)
- [OdysseyBench: Evaluating LLM Agents on Long-Horizon Complex Office Application Workflows](https://arxiv.org/abs/2508.09124)
#### ๐๏ธ 2024
- **[LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory](https://arxiv.org/abs/2410.10813)**
[[data](https://github.com/xiaowu0162/LongMemEval)]
- **[Evaluating Very Long-Term Conversational Memory of LLM Agents](https://arxiv.org/abs/2402.17753)**
(The LoCoMo Paper)
[[code](https://github.com/snap-research/LoCoMo)]
[[data](https://github.com/snap-research/locomo/tree/main/data)]
- **[โBench: Extending Long Context Evaluation Beyond 100K Tokens](https://arxiv.org/abs/2402.13718v3)**
[[code](https://github.com/OpenBMB/InfiniteBench)]
- **[LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding](https://arxiv.org/abs/2308.14508)**
[[code](https://github.com/THUDM/LongBench)]
#### ๐๏ธ 2023
- **[StoryBench: A Multifaceted Benchmark for Continuous Story Visualization](https://proceedings.neurips.cc/paper_files/paper/2023/hash/f63f5fbed1a4ef08c857c5f377b5d33a-Abstract-Datasets_and_Benchmarks.html)**
[[code](https://github.com/google/storybench)]
### ๐ฌ Multimodal Benchmarks
#### ๐๏ธ 2026
- **[VoxMem: Benchmarking Multimodal Memory in Large Audio Language Models](https://arxiv.org/abs/2609.32607)**
[[code](https://github.com/swagshaw/voxmem)]
[[data](https://huggingface.co/datasets/AudioMemory/voxmembench)]
[[proj](https://swagshaw.github.io/voxmem/)]
- **[MBench: A Comprehensive Benchmark on Memory Capability for Video World Models](https://arxiv.org/abs/2606.00793)**
[[code](https://github.com/study-overflow/MBench)]
[[proj](https://peanutup.github.io/MBench-project/)]
[[leaderboard](https://huggingface.co/spaces/study-overflow/MBench_Leaderboard)]
- **[RoboMemArena: A Comprehensive and Challenging Robotic Memory Benchmark](https://arxiv.org/abs/2605.10921)**
[[code](https://github.com/OpenHelix-Team/RoboMemArena)]
[[data](https://huggingface.co/datasets/RoboMemArenaBenchmark/RoboMemArena)]
[[proj](https://robomemarena.github.io/)]
[[leaderboard](https://robomemarena.github.io/leaderboard.html)]
- **[DeepImageSearch: Benchmarking Multimodal Agents for Context-Aware Image Retrieval in Visual Histories](https://arxiv.org/abs/2602.10809)**
[[code](https://github.com/RUC-NLPIR/DeepImageSearch)]
[[data](https://huggingface.co/datasets/RUC-NLPIR/DISBench)]
[[leaderboard](https://huggingface.co/spaces/RUC-NLPIR/DISBench-Leaderboard)]
- **[Persona-MME: A Benchmark for Long-Term Personalized Multimodal LLMs](https://arxiv.org/abs/2604.13074)**
[[code](https://github.com/MiG-NJU/PersonaVLM)]
[[data](https://huggingface.co/datasets/ClareNie/Persona-MME)]
- **[RMBench: Memory-Dependent Robotic Manipulation Benchmark with Insights into Policy Design](https://arxiv.org/abs/2603.01229)**
[[code](https://github.com/robotwin-Platform/rmbench)]
[[proj](https://rmbench.github.io/)]
- **[According to Me: Long-Term Personalized Referential Memory QA (ATM-Bench)](https://arxiv.org/abs/2603.01990)**
[[code](https://github.com/JingbiaoMei/ATM-Bench)]
[[data](https://huggingface.co/datasets/Jingbiao/ATM-Bench)]
[[proj](https://atmbench.github.io/)]
[[leaderboard](https://atmbench.github.io/leaderboard.html)]
#### ๐๏ธ 2025
- **[TeleEgo: Benchmarking Egocentric AI Assistants in the Wild](https://arxiv.org/abs/2510.23981)** โ
[[code](https://github.com/TeleAI-UAGI/TeleEgo)]
[[data](https://huggingface.co/datasets/David0219/TeleEgo)]
[[proj](https://programmergg.github.io/jrliu.github.io/)]
[[leaderboard](https://programmergg.github.io/jrliu.github.io/#leaderboard)]
- **[LVBench: An Extreme Long Video Understanding Benchmark](https://arxiv.org/abs/2406.08035)**
[[code](https://github.com/zai-org/LVBench)]
- **[Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis](https://arxiv.org/abs/2405.21075v3)**
[[code](https://github.com/MME-Benchmarks/Video-MME)]
#### ๐๏ธ 2024
- **[MovieChat+: Question-aware Sparse Memory for Long Video Question Answering](https://arxiv.org/abs/2404.17176)**
[[code](https://github.com/rese1f/MovieChat)]
- **[CinePile: A Long Video Question Answering Dataset and Benchmark](https://arxiv.org/abs/2405.08813)**
[[code](https://huggingface.co/datasets/tomg-group-umd/cinepile)]
- **[LongVideoBench: A Benchmark for Long-Context Interleaved Video-Language Understanding](https://arxiv.org/abs/2407.15754)**
[[code](https://github.com/longvideobench/LongVideoBench)]
#### ๐๏ธ 2023
- **[EgoSchema: A Diagnostic Benchmark for Very Long-form Video Language Understanding](https://proceedings.neurips.cc/paper_files/paper/2023/file/90ce332aff156b910b002ce4e6880dec-Paper-Datasets_and_Benchmarks.pdf)**
[[code](https://github.com/egoschema/egoschema)]
- [LvBench: A Benchmark for Long-form Video Understanding with Versatile Multi-modal Question Answering](https://arxiv.org/abs/2312.04817)
### ๐ฎ Dynamic Benchmarks & Simulation Environments
#### ๐๏ธ 2026
- **[EmbodiedMemory-Bench: Benchmarking Embodied Memory for Long-Horizon Embodied Tasks](https://arxiv.org/abs/2609.28236)**
[[code](https://github.com/ZJU-OmniAI/Embodied-Omni/tree/main/embodied_memory)]
[[data](https://huggingface.co/datasets/lzLiang/EmbodiedMemoryBench)]
- **[SEAGym: An Evaluation Environment for Self-Evolving LLM Agents](https://arxiv.org/abs/2606.17546)**
[[code](https://github.com/antropy-research/SEAGym)]
- **[Beyond Static Dialogues: Benchmarking Realistic, Heterogeneous, and Evolving Long-Term Memory](https://arxiv.org/abs/2605.31086)**
[[code](https://github.com/microsoft/RHELM)]
[[data](https://huggingface.co/datasets/microsoft/RHELM)]
[[proj](https://microsoft.github.io/RHELM/)]
- **[AMemGym: Interactive Memory Benchmarking for Assistants in Long-Horizon Conversations](https://arxiv.org/abs/2603.01966)**
[[code](https://github.com/AGI-Eval-Official/amemgym)]
[[proj](https://agi-eval-official.github.io/amemgym/)]
- [StreamMemBench: Streaming Evaluation of Agent Memory for Future-Oriented Assistance](https://arxiv.org/abs/2606.14571)
- **[agent-memory-bench](https://giulioder.github.io/agent-memory-bench/)**
[[code](https://github.com/GiulioDER/agent-memory-bench)]
[[data](https://huggingface.co/datasets/Gde05/agent-memory-bench-corpus)]
_Preregistered harness scoring memory layers for coding agents by executing task checkers rather than judging text; the author's own RE-call is among the arms._
#### ๐๏ธ 2025
- **[MemoryBench: A Benchmark for Memory and Continual Learning in LLM Systems](https://arxiv.org/abs/2510.17281)**
[[code](https://github.com/LittleDinoC/MemoryBench)]
[[data](https://huggingface.co/datasets/THUIR/MemoryBench)]
- **[ARE: Scaling Up Agent Environments and Evaluations](https://arxiv.org/abs/2509.17158)**
(The Gaia2 Paper)
[[code](https://github.com/facebookresearch/meta-agents-research-environments)]
#### ๐๏ธ 2024
- **[AppWorld: A Controllable World of Apps and People for Benchmarking Interactive Coding Agents](https://arxiv.org/abs/2407.18901)**
[[code](https://github.com/StonyBrookNLP/appworld)]
---
## ๐ค Papers - Nonparametric Memory
### ๐ Text Memory
#### ๐๏ธ 2026
- **[The Price of Meaning: Why Every Semantic Memory System Forgets](https://arxiv.org/abs/2603.27116)**
[[code](https://github.com/Dynamis-Labs/no-escape)]
- **[StructMem: Structured Memory for Long-Horizon Behavior in LLMs](https://arxiv.org/abs/2604.21748)**
[[code](https://github.com/zjunlp/LightMem)]
- **[Memory Efficiency and Resource-Rational Encoding in Sentence Processing](https://aclanthology.org/2026.acl-long.1550/)**
[[code](https://github.com/weijiexu-charlie/resource-rational-encoding)]
- **[AutoMem: Automated Learning of Memory as a Cognitive Skill](https://arxiv.org/abs/2607.01224)**
[[code](https://github.com/autoLearnMem/AutoMem)]
[[proj](https://autolearnmem.github.io/)]
- **[Mandol: An Agglomerative Agent Memory System for Long-Term Conversations](https://arxiv.org/abs/2606.29778)**
[[code](https://github.com/AgentCombo/Mandol)]
- **[RecMem: Recurrence-based Memory Consolidation for Efficient and Effective Long-Running LLM Agents](https://arxiv.org/abs/2605.16045)**
[[code](https://github.com/CaiusDai/RecMem)]
- **[Evoking User Memory: Personalizing LLM via Recollection-Familiarity Adaptive Retrieval](https://arxiv.org/abs/2603.09250)** (RF-Mem)
[[code](https://github.com/Applied-Machine-Learning-Lab/ICLR2026_RF-Mem)]
- **[MemPrivacy: Privacy-Preserving Personalized Memory Management for Edge-Cloud Agents](https://arxiv.org/abs/2605.09530)**
[[code](https://github.com/MemTensor/MemPrivacy)]
- **[Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation](https://arxiv.org/abs/2602.02007)**
[[code](https://github.com/HU-xiaobai/xMemory)]
- **[MemSearch-o1: Empowering Large Language Models with Reasoning-Aligned Memory Growth in Agentic Search](https://arxiv.org/abs/2604.17265)**
[[code](https://github.com/Applied-Machine-Learning-Lab/ACL2026_MemSearch-o1)]
- **[SimpleMem: Efficient Lifelong Memory for LLM Agents](https://arxiv.org/abs/2601.02553)**
[[code](https://github.com/aiming-lab/SimpleMem)]
- **[Beyond Similarity Search: Tenure and the Case for Structured Belief State in LLM Memory](https://arxiv.org/abs/2605.11325)**
[[code](https://github.com/jeffreyflynt/tenure)]
- [Self-Correcting Long-Horizon Search Agents via Tree-Structured Memory](https://arxiv.org/abs/2608.10676) (ReTree)
- [MemCompiler: Compile, Don't Inject -- State-Conditioned Memory for Embodied Agents](https://arxiv.org/abs/2605.07594)
- [Agentic Memory Enhanced Recursive Reasoning for Root Cause Localization in Microservices](https://arxiv.org/abs/2601.02732) (AMER-RCL)
#### ๐๏ธ 2025
- **[LightMem: Lightweight and Efficient Memory-Augmented Generation](https://arxiv.org/abs/2510.18866)**
[[code](https://github.com/zjunlp/LightMem)]
- **[What Deserves Memory: Adaptive Memory Distillation for LLM Agents](https://arxiv.org/abs/2508.03341)**
[[code](https://github.com/nemori-ai/nemori)]
- **[Human-inspired Episodic Memory for Infinite Context LLMs](https://arxiv.org/abs/2407.09450)**
[[code](https://github.com/em-llm/EM-LLM-model)]
- **[MemWeaver: A Hierarchical Memory from Textual Interactive Behaviors for Personalized Generation](https://arxiv.org/abs/2510.07713)**
[[code](https://github.com/fishsure/MemWeaver)]
- [Evaluating Long-Term Memory for Long-Context Question Answering](https://arxiv.org/abs/2510.23730)
- [Text2Mem: A Unified Memory Operation Language for Memory Operating System](https://arxiv.org/abs/2509.11145)
- [O-Mem: Omni Memory System for Personalized, Long Horizon, Self-Evolving Agents](https://arxiv.org/abs/2511.13593)
- [Omne-R1: Learning to Reason with Memory for Multi-hop Question Answering](https://arxiv.org/abs/2508.17330)
- [In Prospect and Retrospect: Reflective Memory Management for Long-term Personalized Dialogue Agents](https://aclanthology.org/2025.acl-long.413/)
- [SEDM: Scalable Self-Evolving Distributed Memory for Agents](https://arxiv.org/abs/2509.09498)
- [MemoRAG: Boosting Long Context Processing with Global Memory-Enhanced Retrieval Augmentation](https://arxiv.org/abs/2409.05591)
- [Towards LifeSpan Cognitive Systems](https://arxiv.org/abs/2409.13265)
#### ๐๏ธ 2024
- **[Compress to Impress: Unleashing the Potential of Compressive Memory in Real-World Long-Term Conversations](https://arxiv.org/abs/2402.11975)**
[[code](https://github.com/nuochenpku/COMEDY)]
- **[Agent Workflow Memory](https://arxiv.org/abs/2409.07429)**
[[code](https://github.com/zorazrw/agent-workflow-memory)]
- **[MemoryBank: Enhancing Large Language Models with Long-Term Memory](https://arxiv.org/abs/2305.10250)**
[[code](https://github.com/zhongwanjun/MemoryBank-SiliconFriend)]
- **[Toward Conversational Agents with Context and Time Sensitive Long-term Memory](https://arxiv.org/abs/2406.00057)**
[[data](https://github.com/Zyphra/TemporalMemoryDataset)]
- [InfLLM: Training-Free Long-Context Extrapolation for LLMs with an Efficient Context Memory](https://arxiv.org/abs/2402.04617)
#### ๐๏ธ 2023
- [RET-LLM: Towards a General Read-Write Memory for Large Language Models](https://arxiv.org/abs/2305.14322)
### ๐ Graph Memory
#### ๐๏ธ 2026
- **[Jev-Mem: System-One-Controlled Agentic Memory for Efficient AI Agents](https://arxiv.org/abs/2609.23986)**
[[code](https://github.com/libingzheren/Jev-Mem)]
- **[T-Mem: Memory That Anticipates, Not Archives](https://arxiv.org/abs/2606.15405)**
[[code](https://github.com/Sherlockwz/T-Mem)]
- **[Rethinking Memory as Continuously Evolving Connectivity](https://arxiv.org/abs/2605.28773)** (FluxMem)
[[code](https://github.com/zjunlp/LightMem)]
- **[GraphPlanner: Graph Memory-Augmented Agentic Routing for Multi-Agent LLMs](https://arxiv.org/abs/2604.23626)**
[[code](https://github.com/ulab-uiuc/GraphPlanner)]
- **[HyperMem: Hypergraph Memory for Long-Term Conversations](https://arxiv.org/abs/2604.08256)**
[[code](https://github.com/EverMind-AI/EverOS)]
- **[Mnemis: Dual-Route Retrieval on Hierarchical Graphs for Long-Term LLM Memory](https://arxiv.org/abs/2602.15313)**
[[code](https://github.com/microsoft/Mnemis)]
- **[MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents](https://arxiv.org/abs/2601.03236)**
[[code](https://github.com/FredJiang0324/MAMGA)]
- **[TraceMem: Weaving Narrative Memory Schemata from User Conversational Traces](https://arxiv.org/abs/2602.09712)**
[[code](https://github.com/YimingShu-teay/TraceMem)]
- **[PlugMem: A Task-Agnostic Plugin Memory Module for LLM Agents](https://arxiv.org/abs/2603.03296)**
[[code](https://github.com/TIMAN-group/PlugMem)]
- [SAGE: A Self-Evolving Agentic Graph-Memory Engine for Structure-Aware Associative Memory](https://arxiv.org/abs/2605.12061)
#### ๐๏ธ 2025
- **[From RAG to Memory: Non-Parametric Continual Learning for Large Language Models](https://arxiv.org/abs/2502.14802)**
[[code](https://github.com/OSU-NLP-Group/HippoRAG)]
- **[MIRIX: Multi-Agent Memory System for LLM-Based Agents](https://arxiv.org/abs/2507.07957)**
[[code](https://github.com/Mirix-AI/MIRIX)]
- **[From Single to Multi-Granularity: Toward Long-Term Memory Association and Selection of Conversational Agents](https://arxiv.org/abs/2505.19549)** (MemGAS)
[[code](https://github.com/quqxui/MemGAS)]
- **[Hierarchical Memory Organization for Wikipedia Generation](https://aclanthology.org/2025.acl-long.1423/)**
[[code](https://github.com/eugeneyujunhao/mog)]
- [From Experience to Strategy: Empowering LLM Agents with Trainable Graph Memory](https://www.arxiv.org/abs/2511.07800)
- [Bridging Intuitive Associations and Deliberate Recall: Empowering LLM Personal Assistant with Graph-Structured Long-term Memory](https://aclanthology.org/2025.findings-acl.901/)
- [HiAgent: Hierarchical Working Memory Management for Solving Long-Horizon Agent Tasks with Large Language Model](https://aclanthology.org/2025.acl-long.1575/)
- [Optimizing the Interface Between Knowledge Graphs and LLMs for Complex Reasoning](https://arxiv.org/abs/2505.24478)
#### ๐๏ธ 2024
- **[HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language Models](https://arxiv.org/abs/2405.14831)**
[[code](https://github.com/OSU-NLP-Group/HippoRAG)]
- **[AriGraph: Learning Knowledge Graph World Models with Episodic Memory for LLM Agents](https://arxiv.org/abs/2407.04363)**
[[code](https://github.com/AIRI-Institute/AriGraph)]
### ๐ฅ Multimodal Memory (for Understanding)
#### ๐๏ธ 2026
- **[VoiceMem: Streaming Dual-Brain Memory for Real-Time Interaction](https://arxiv.org/abs/2608.26005)**
[[code](https://github.com/xzf-thu/VoiceMem)]
[[proj](https://xzf-thu.github.io/VoiceMem/)]
- **[MemGUI-Agent: An End-to-End Long-Horizon Mobile GUI Agent with Proactive Context Management](https://arxiv.org/abs/2606.19926)**
[[code](https://github.com/kwai/MemGUI-Agent)]
[[proj](https://memgui-agent.github.io/)]
- **[FluxMem: Adaptive Hierarchical Memory for Streaming Video Understanding](https://arxiv.org/abs/2603.02096)**
[[code](https://github.com/YiwengXie/FluxMem)]
[[proj](https://yiwengxie.com/FluxMem/)]
- **[SE-GA: Memory-Augmented Self-Evolution for GUI Agents](https://arxiv.org/abs/2605.16883)**
[[code](https://github.com/jinshilong-dev/SE-GA)]
- **[Visual Agentic Memory: Enabling Online Long Video Understanding via Online Indexing, Hierarchical Memory, and Agentic Retrieval](https://arxiv.org/abs/2605.16481)**
[[code](https://github.com/yiliu-li/Visual-Agentic-Memory)]
- **[PersonaVLM: Long-Term Personalized Multimodal LLMs](https://arxiv.org/abs/2604.13074)**
[[code](https://github.com/MiG-NJU/PersonaVLM)]
[[proj](https://PersonaVLM.github.io)]
- **[Omni-SimpleMem: Autoresearch-Guided Discovery of Lifelong Multimodal Agent Memory](https://arxiv.org/abs/2604.01007)**
[[code](https://github.com/aiming-lab/SimpleMem)]
- **[HERMES: KV Cache as Hierarchical Memory for Efficient Streaming Video Understanding](https://arxiv.org/abs/2601.14724)**
[[code](https://github.com/haowei-freesky/HERMES)]
- **[EventMemAgent: Hierarchical Event-Centric Memory for Online Video Understanding with Adaptive Tool Use](https://arxiv.org/abs/2602.15329)**
[[code](https://github.com/lingcco/EventMemAgent)]
- **[M2A: Multimodal Memory Agent with Dual-Layer Hybrid Memory for Long-Term Personalized Interactions](https://arxiv.org/abs/2602.07624)**
[[code](https://github.com/Little-Fridge/M2A)]
- [LT-Mem: Volatility-Aware Spatio-Temporal Memory for Lifelong Scene Understanding](https://arxiv.org/abs/2608.19059)
[[proj](https://lt-mem.github.io/)]
- [NativeMEM: Native Memory Compression for Long-Horizon Robotic Manipulation](https://arxiv.org/abs/2607.06678)
[[proj](https://opendrivelab.com/NativeMEM/)]
#### ๐๏ธ 2025
- **[WorldMM: Dynamic Multimodal Memory Agent for Long Video Reasoning](https://arxiv.org/abs/2512.02425)**
[[code](https://github.com/wgcyeo/WorldMM)]
- **[MemVerse: Multimodal Memory for Lifelong Learning Agents](https://arxiv.org/abs/2512.03627)**
[[code](https://github.com/KnowledgeXLab/MemVerse)]
- **[MGA: Memory-Driven GUI Agent for Observation-Centric Interaction](https://arxiv.org/abs/2510.24168)**
[[code](https://github.com/MintyCo0kie/MGA4OSWorld)]
- **[Seeing, Listening, Remembering, and Reasoning: A Multimodal Agent with Long-Term Memory](https://arxiv.org/abs/2508.09736)**
[[code](https://github.com/bytedance-seed/m3-agent)]
- **[HippoMM: Hippocampal-inspired Multimodal Memory for Long Audiovisual Event Understanding](https://arxiv.org/abs/2504.10739)**
[[code](https://github.com/linyueqian/HippoMM)]
- [Infinite Video Understanding](https://www.arxiv.org/abs/2507.09068)
- [Episodic Memory Representation for Long-form Video Understanding](https://arxiv.org/abs/2508.09486)
- [Multi-RAG: A Multimodal Retrieval-Augmented Generation System for Adaptive Video Understanding](https://arxiv.org/abs/2505.23990)
- [Contextual Experience Replay for Self-Improvement of Language Agents](https://arxiv.org/abs/2506.06698)
#### ๐๏ธ 2024
- **[VideoAgent: Long-form Video Understanding with Large Language Model as Agent](https://arxiv.org/abs/2403.10517)**
[[code](https://github.com/HKUDS/VideoAgent)]
- **[VideoChat-Flash: Hierarchical Compression for Long-Context Video Modeling](https://arxiv.org/abs/2501.00574)**
[[code](https://github.com/OpenGVLab/VideoChat-Flash)]
- **[LongVLM: Efficient Long Video Understanding via Large Language Models](https://arxiv.org/abs/2404.03384)**
[[code](https://github.com/ziplab/LongVLM)]
- **[KARMA: Augmenting Embodied AI Agents with Long-and-short Term Memory Systems](https://arxiv.org/abs/2409.14908)**
[[code](https://github.com/WZX0Swarm0Robotics/KARMA/tree/master)]
### ๐ฅ Multimodal Memory (for Generation)
#### ๐๏ธ 2026
- **[LayerRecall: A State-Conditioned Memory Router for Long-Horizon Consistency in Video Generation](https://arxiv.org/abs/2608.28460)**
[[code](https://github.com/Yixuan-Ding-ZJU/LayerRecall)]
[[proj](https://yixuan-ding-zju.github.io/LayerRecall_Web/)]
- **[MemSlides: A Hierarchical Memory Driven Agent Framework for Personalized Slide Generation with Multi-turn Local Revision](https://arxiv.org/abs/2606.17162)**
[[code](https://github.com/huohua325/Memslides)]
[[proj](https://memslides.github.io/)]
- **[LoGeR: Long-Context Geometric Reconstruction with Hybrid Memory](https://arxiv.org/abs/2603.03269)**
[[code](https://github.com/Junyi42/LoGeR)]
- [OneStory: Coherent Multi-Shot Video Generation with Adaptive Memory](https://arxiv.org/abs/2512.07802)
#### ๐๏ธ 2025
- **[MagicWorld: Towards Long-Horizon Stability for Interactive Video World Exploration](https://arxiv.org/abs/2511.18886)**
[[code](https://github.com/vivoCameraResearch/Magic-World)]
- **[Yume-1.5: A Text-Controlled Interactive World Generation Model](https://arxiv.org/abs/2512.22096)**
[[code](https://github.com/stdstu12/YUME)]
- **[StoryMem: Multi-shot Long Video Storytelling with Memory](https://arxiv.org/abs/2512.19539)**
[[code](https://github.com/Kevin-thu/StoryMem)]
- **[MemFlow: Flowing Adaptive Memory for Consistent and Efficient Long Video Narratives](https://arxiv.org/abs/2512.14699)**
[[code](https://github.com/KlingTeam/MemFlow)]
- **[MotionRAG: Motion Retrieval-Augmented Image-to-Video Generation](http://arxiv.org/abs/2509.26391)**
[[code](https://github.com/MCG-NJU/MotionRAG)]
- **[VideoRAG: Retrieval-Augmented Generation over Video Corpus](http://arxiv.org/abs/2501.05874)**
[[code](https://github.com/starsuzi/VideoRAG)]
- [Pretraining Frame Preservation in Autoregressive Video Memory Compression](https://arxiv.org/abs/2512.23851)
- [EgoLCD: Egocentric Video Generation with Long Context Diffusion](https://arxiv.org/abs/2512.04515)
- [Pack and Force Your Memory: Long-form and Consistent Video Generation](http://arxiv.org/abs/2510.01784)
- [Video World Models with Long-term Spatial Memory](http://arxiv.org/abs/2506.05284)
- [Mixture of Contexts for Long Video Generation](http://arxiv.org/abs/2508.21058)
- [Context as Memory: Scene-Consistent Interactive Long Video Generation with Memory Retrieval](http://arxiv.org/abs/2506.03141)
## ๐ข Papers - Parametric Memory
#### ๐๏ธ 2026
- **[Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models](https://arxiv.org/abs/2601.07372)**
(The DeepSeek **Engram** Paper)
[[code](https://github.com/deepseek-ai/Engram/)]
+ **[Beyond Conditional Computation: Retrieval-Augmented Genomic Foundation Models with Gengram](https://arxiv.org/abs/2601.22203)**
[[code](https://github.com/zhejianglab/Gengram/)]
+ [Pooling Engram Conditional Memory in Large Language Models using CXL](https://arxiv.org/abs/2603.10087)
+ [A Collision-Free Hot-Tier Extension for Engram-Style Conditional Memory: A Controlled Study of Training Dynamics](https://arxiv.org/abs/2601.16531)
- **[Metis: Memory Foundation Model](https://arxiv.org/abs/2607.26760)**
[[code](https://github.com/MemTensor/Metis)]
[[model](https://huggingface.co/collections/IAAR-Shanghai/metis)]
- **[ฮด-mem: Efficient Online Memory for Large Language Models](https://arxiv.org/abs/2605.12357)**
[[code](https://github.com/MindLab-Research/delta-Mem)]
- **[Language Model Memory and Memory Models for Language](https://arxiv.org/abs/2602.13466)**
[[code](https://github.com/blbadger/memorymodels)]
- **[MSA: Memory Sparse Attention for Efficient End-to-End Memory Model Scaling to 100M Tokens](https://github.com/EverMind-AI/MSA/blob/main/paper/MSA__Memory_Sparse_Attention_for_Efficient_End_to_End_Memory_Model_Scaling_to_100M_Tokens.pdf)**
[[code](https://github.com/EverMind-AI/MSA)]
- **[GradMem: Learning to Write Context into Memory with Test-Time Gradient Descent](https://arxiv.org/abs/2603.13875)**
[[code](https://github.com/yurakuratov/gradmem)]
- **[MeKi: Memory-based Expert Knowledge Injection for Efficient LLM Scaling](https://arxiv.org/abs/2602.03359)**
[[code](https://github.com/ningding-o/MeKi)]
- **[STEM: Scaling Transformers with Embedding Modules](https://arxiv.org/abs/2601.10639)**
[[code](https://github.com/Infini-AI-Lab/STEM)]
- **[MeMo: Memory as a Model](https://arxiv.org/abs/2605.15156)**
[[code](https://github.com/arunv3rma/MeMo)]
- [Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories](https://arxiv.org/abs/2606.03979)
- [Do Language Models Need Sleep? Offline Recurrence for Improved Online Inference](https://arxiv.org/abs/2605.26099)
- [Training Transformers for KV Cache Compressibility](https://arxiv.org/abs/2605.05971)
- [Contextual Agentic Memory is a Memo, Not True Memory](https://arxiv.org/abs/2604.27707)
- [Memory Caching: RNNs with Growing Memory](https://arxiv.org/abs/2602.24281)
- [Fast-weight Product Key Memory](https://arxiv.org/abs/2601.00671)
#### ๐๏ธ 2025
- **[MoM: Linear Sequence Modeling with Mixture-of-Memories](https://arxiv.org/abs/2502.13685)**
[[code](https://github.com/OpenSparseLLMs/MoM)]
- **[MLP Memory: Language Modeling with Retriever-pretrained External Memory](https://arxiv.org/abs/2508.01832)**
[[code](https://github.com/Rubin-Wei/MLPMemory)]
- **[Memory Decoder: A Pretrained, Plug-and-Play Memory for Large Language Models](https://www.arxiv.org/abs/2508.09874)**
[[code](https://github.com/LUMIA-Group/MemoryDecoder)]
- **[Little By Little: Continual Learning via Incremental Mixture of Rank-1 Associative Memory Experts](https://arxiv.org/abs/2506.21035)**
[[code](https://github.com/Artificer-AI-Lab/MoRAM)]
[[proj](https://artificer-ai-lab.github.io/MoRAM/)]
- [How Much Do Language Models Memorize?](https://arxiv.org/abs/2505.24832)
- [Memory Retrieval and Consolidation in Large Language Models through Function Tokens](https://arxiv.org/abs/2510.08203)
- [Nested Learning: The Illusion of Deep Learning Architectures](https://openreview.net/forum?id=nbMeRvNb7A)
- [Improving Factuality with Explicit Working Memory](https://arxiv.org/abs/2412.18069)
- [R3Mem: Bridging Memory Retention and Retrieval via Reversible Compression](https://arxiv.org/abs/2502.15957)
- [May the Memory Be With You: Efficient and Infinitely Updatable State for Large Language Models](https://dl.acm.org/doi/abs/10.1145/3721146.3721951)
- [MeMo: Towards Language Models with Associative Memory Mechanisms](https://aclanthology.org/2025.findings-acl.785/)
- [REFRAG: Rethinking RAG based Decoding](https://arxiv.org/abs/2509.01092)
- [EpMAN: Episodic Memory AttentioN for Generalizing to Longer Contexts](https://aclanthology.org/2025.acl-long.574/)
- [Disentangling Memory and Reasoning Ability in Large Language Models](https://aclanthology.org/2025.acl-long.84/)
#### ๐๏ธ 2024
- **[InfLLM: Training-Free Long-Context Extrapolation for LLMs with an Efficient Context Memory](https://arxiv.org/abs/2402.04617)**
[[code](https://github.com/thunlp/InfLLM)]
- **[MA-LMM: Memory-Augmented Large Multimodal Model for Long-Term Video Understanding](https://arxiv.org/abs/2404.05726)**
[[code](https://github.com/boheumd/MA-LMM)]
- **[MemoryLLM: Towards Self-Updatable Large Language Models](https://arxiv.org/abs/2402.04624)**
[[code](https://github.com/wangyu-ustc/MemoryLLM)]
- **[WISE: Rethinking the Knowledge Memory for Lifelong Model Editing of Large Language Models](https://arxiv.org/abs/2405.14768)**
[[code](https://github.com/zjunlp/EasyEdit)]
- [Titans: Learning to Memorize at Test Time](https://arxiv.org/abs/2501.00663)
- [Memory3: Language Modeling with Explicit Memory](https://arxiv.org/abs/2407.01178v1)
- [Infinite-LLM: Efficient LLM Service for Long Context with DistAttention and Distributed KVCache](https://arxiv.org/abs/2401.02669)
- [MemServe: Context Caching for Disaggregated LLM Serving with Elastic Memory Pool](https://arxiv.org/abs/2406.17565)
- [WISE: Rethinking the Knowledge Memory for Lifelong Model Editing of Large Language Models](https://arxiv.org/abs/2405.14768/)
- [Ultra-Sparse Memory Network](https://arxiv.org/abs/2411.12364)
#### ๐๏ธ 2023
- **[Augmenting Language Models with Long-Term Memory](https://arxiv.org/abs/2306.07174)**
[[code](https://github.com/Victorwz/LongMem)]
- **[Efficient Memory Management for Large Language Model Serving with PagedAttention](https://arxiv.org/abs/2309.06180)**
[[code](https://github.com/vllm-project/vllm)]
## ๐ Papers - Memory for Agent Evolution
### ๐งญ Reinforcement Learning & Continual Learning
#### ๐๏ธ 2026
- **[Bilevel Coordinated Reflection: A Game-Theoretic Approach to Multi-Agent LLM Systems](https://arxiv.org/abs/2609.02750)**
[[code](https://github.com/YihangChen9/Bilevel-Coordinated-Reflection)]
- **[Large Discovery Models: Empirically-grounded Model-Based Open-Ended Search](https://arxiv.org/abs/2608.15669)**
[[code](https://github.com/yzailab/Large-Discovery-Models)]
[[proj](https://largediscovery.net/)]
[[blog](https://largediscovery.net/blog/)]
- **[SEED: Self-Evolving On-Policy Distillation for Agentic Reinforcement Learning](https://arxiv.org/abs/2607.14777)**
[[code](https://github.com/jinyangwu/SEED)]
[[proj](https://jinyangwu.github.io/seed/)]
- **[MLEvolve: A Self-Evolving Framework for Automated Machine Learning Algorithm Discovery](https://arxiv.org/abs/2606.06473)**
[[code](https://github.com/InternScience/MLEvolve)]
- **[EvoTrainer: Co-Evolving LLM Policies and Training Harnesses for Autonomous Agentic Reinforcement Learning](https://arxiv.org/abs/2606.03108)**
[[code](https://github.com/AlibabaResearch/DAMO-ConvAI/tree/main/EvoTrainer)]
- **[SkillOpt: Executive Strategy for Self-Evolving Agent Skills](https://arxiv.org/abs/2605.23904)**
[[code](https://github.com/microsoft/SkillOpt)]
- **[Learning, Fast and Slow: Towards LLMs That Adapt Continually](https://arxiv.org/abs/2605.12484)**
[[code](https://rishabhtiwari.ai/projects/fst/code/)]
[[blog](https://gepa-ai.github.io/gepa/blog/2026/05/11/learning-fast-and-slow/)]
- **[CASCADE: Case-Based Continual Adaptation for Large Language Models During Deployment](https://arxiv.org/abs/2605.06702)**
(The DTLBench Paper)
[[code](https://github.com/guosyjlu/CASCADE)]
- **[Memory Intelligence Agent](https://arxiv.org/abs/2604.04503)**
[[code](https://github.com/ECNU-SII/MIA)]
- **[PASK: Toward Intent-Aware Proactive Agents with Long-Term Memory](https://arxiv.org/abs/2604.08000)**
[[code](https://github.com/xzf-thu/Pask)]
- **[Toward Autonomous Long-Horizon Engineering for ML Research](https://arxiv.org/abs/2604.13018)**
[[code](https://github.com/AweAI-Team/AiScientist)]
- **[OpenClaw-RL: Train Any Agent Simply by Talking](https://arxiv.org/abs/2603.10165)**
[[code](https://github.com/Gen-Verse/OpenClaw-RL)]
- **[Memento-Skills: Let Agents Design Agents](https://arxiv.org/abs/2603.18743)**
[[code](https://github.com/Memento-Teams/Memento-Skills)]
- **[Memento 2: Learning by Stateful Reflective Memory](https://arxiv.org/abs/2512.22716)**
[[code](https://github.com/Agent-on-the-Fly/Memento)]
- **[Principled Fast and Meta Knowledge Learners for Continual Reinforcement Learning](https://arxiv.org/abs/2603.00903)** (FAME)
[[code](https://github.com/datake/FAME)]
- **[MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents](https://arxiv.org/abs/2602.02474)**
[[code](https://github.com/ViktorAxelsen/MemSkill)]
- **[ProcMEM: Learning Reusable Procedural Memory from Experience via Non-Parametric PPO for LLM Agents](https://arxiv.org/abs/2602.01869)**
[[code](https://github.com/Miracle1207/ProcMEM)]
- **[MemRL: Self-Evolving Agents via Runtime Reinforcement Learning on Episodic Memory](https://arxiv.org/abs/2601.03192)**
[[code](https://github.com/MemTensor/MemRL)]
- [UniMem: Complementary Episodic-to-Parametric Memory for Boundary-Agnostic Task Streams](https://arxiv.org/abs/2607.26017)
- [Self-Evolving Multi-Agent Systems via Decentralized Memory](https://arxiv.org/abs/2605.22721)
- [Meta-Cognitive Memory Policy Optimization for Long-Horizon LLM Agents](https://arxiv.org/abs/2605.30159)
- [Mem-ฯ: Adaptive Memory through Learning When and What to Generate](https://arxiv.org/abs/2605.21463)
- [Useful Memories Become Faulty When Continuously Updated by LLMs](https://arxiv.org/abs/2605.12978)
- [MEMTIER: Tiered Memory Architecture and Retrieval Bottleneck Analysis for Long-Running Autonomous AI Agents](https://arxiv.org/abs/2605.03675)
- [Neural Garbage Collection: Learning to Forget while Learning to Reason](https://arxiv.org/abs/2604.18002)
- [AVO: Agentic Variation Operators for Autonomous Evolutionary Search](https://arxiv.org/abs/2603.24517)
- [Why the Brain Consolidates: Predictive Forgetting for Optimal Generalisation](https://arxiv.org/abs/2603.04688)
- [Agentic Memory: Learning Unified Long-Term and Short-Term Memory Management for Large Language Model Agents](https://arxiv.org/abs/2601.01885)
#### ๐๏ธ 2025
- **[End-to-End Test-Time Training for Long Context](https://arxiv.org/abs/2512.23675)**
[[code](https://github.com/test-time-training/e2e)]
- **[ML-Master: Towards AI-for-AI via Integration of Exploration and Reasoning](https://arxiv.org/abs/2506.16499)**
[[code](https://github.com/sjtu-sai-agents/ML-Master)]
- **[MemEvolve: Meta-Evolution of Agent Memory Systems](https://arxiv.org/abs/2512.18746)**
[[code](https://github.com/bingreeky/MemEvolve)]
- **[Remember Me, Refine Me: A Dynamic Procedural Memory Framework for Experience-Driven Agent Evolution](https://arxiv.org/abs/2512.10696)**
[[code](https://github.com/agentscope-ai/ReMe)]
- **[EvolveR: Self-Evolving LLM Agents through an Experience-Driven Lifecycle](https://arxiv.org/abs/2510.16079)**
[[code](https://github.com/KnowledgeXLab/EvolveR)]
- **[Learning on the Job: An Experience-Driven, Self-Evolving Agent for Long-Horizon Tasks](https://arxiv.org/abs/2510.08002)**
[[code](https://github.com/KnowledgeXLab/MUSE)]
- **[Mem-ฮฑ: Learning Memory Construction via Reinforcement Learning](https://arxiv.org/abs/2509.25911)**
[[code](https://github.com/wangyu-ustc/Mem-alpha)]
- **[Memento: Fine-tuning LLM Agents without Fine-tuning LLMs](https://arxiv.org/abs/2508.16153)**
[[code](https://github.com/Agent-on-the-Fly/Memento)]
- **[Goal-Directed Search Outperforms Goal-Agnostic Memory Compression in Long-Context Memory Tasks](https://arxiv.org/abs/2511.21726)**
[[code](https://arxiv.org/abs/2511.21726)]
- **[General Agentic Memory via Deep Research](https://arxiv.org/abs/2511.18423)**
[[code](https://github.com/VectorSpaceLab/general-agentic-memory/)]
- **[AgentEvolver: Towards Efficient Self-Evolving Agent System](https://arxiv.org/abs/2511.10395)**
[[code](https://github.com/modelscope/AgentEvolver)]
- **[FLEX: Continuous Agent Evolution via Forward Learning from Experience](https://arxiv.org/abs/2511.06449)**
[[code](https://github.com/GenSI-THUAIR/FLEX)]
- **[MemAgent: Reshaping Long-Context LLM with Multi-Conv RL-based Memory Agent](https://arxiv.org/abs/2507.02259)**
[[code](https://github.com/BytedTsinghua-SIA/MemAgent)]
- [Beyond Heuristics: A Decision-Theoretic Framework for Agent Memory Management](https://arxiv.org/abs/2512.21567)
- [Nested Learning: The Illusion of Deep Learning Architecture](https://abehrouz.github.io/files/NL.pdf)
[[blog](https://research.google/blog/introducing-nested-learning-a-new-ml-paradigm-for-continual-learning/)]
- [LightSearcher: Efficient DeepSearch via Experiential Memory](https://www.arxiv.org/abs/2512.06653)
- [Memory-R1: Enhancing Large Language Model Agents to Manage and Utilize Memories via Reinforcement Learning](https://arxiv.org/abs/2508.19828)
- [Latent Learning: Episodic Memory Complements Parametric Learning by Enabling Flexible Reuse of Experiences](https://arxiv.org/abs/2509.16189)
- [Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory](https://arxiv.org/abs/2511.20857)
- [ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory](https://arxiv.org/abs/2509.25140)
- [Long Term Memory: The Foundation of AI Self-Evolution](https://arxiv.org/abs/2410.15665)
- [REFRAG: Rethinking RAG based Decoding](https://arxiv.org/abs/2509.01092)
- [MemGen: Weaving Generative Latent Memory for Self-Evolving Agents](https://arxiv.org/abs/2509.24704)
- [ReSum: Unlocking Long-Horizon Search Intelligence via Context Summarization](https://arxiv.org/abs/2509.13313)
- [MARC: Memory-Augmented RL Token Compression for Efficient Video Understanding](https://arxiv.org/pdf/2510.07915)
- [Continual Learning via Sparse Memory Finetuning](https://arxiv.org/abs/2510.15103)
- [Task-Core Memory Management and Consolidation for Long-term Continual Learning](https://arxiv.org/abs/2505.09952)
### ๐งฉ Context Engineering & Harness Engineering
#### ๐๏ธ 2026
- **[Recursive ExperientialโWorking Memory Evolution for Long-Horizon Agent Harnesses](https://arxiv.org/abs/2608.24876)** (Recuris)
[[code](https://github.com/Gen-Verse/Recuris)]
- **[TokenPilot: Cache-Efficient Context Management for LLM Agents](https://arxiv.org/abs/2606.17016)**
[[code](https://github.com/zjunlp/LightRSI)]
- **[Code as Agent Harness](https://arxiv.org/abs/2605.18747)**
[[code](https://github.com/YennNing/Awesome-Code-as-Agent-Harness-Papers)]
- **[SWE-Pruner: Self-Adaptive Context Pruning for Coding Agents](https://arxiv.org/abs/2601.16746)**
[[code](https://github.com/Ayanami1314/swe-pruner)]
- **[LCM: Lossless Context Management](https://papers.voltropy.com/LCM)**
[[code](https://github.com/Martian-Engineering/lossless-claw)]
- **[CL-bench: A Benchmark for Context Learning](https://arxiv.org/abs/2602.03587)**
[[code](https://github.com/Tencent-Hunyuan/CL-bench)]
- **[Self-Harness: Harnesses That Improve Themselves](https://arxiv.org/abs/2606.09498)**
[[code](https://github.com/qzzqzzb/Self-Harness)]
- [Is Grep All You Need? How Agent Harnesses Reshape Agentic Search](https://arxiv.org/abs/2605.15184)
- [M\*: Every Task Deserves Its Own Memory Harness](https://arxiv.org/abs/2604.11811)
#### ๐๏ธ 2025
- **[Everything is Context: Agentic File System Abstraction for Context Engineering](https://arxiv.org/abs/2512.05470)**
[[code](https://github.com/AIGNE-io/aigne-framework)]
- **[AgentFold: Long-Horizon Web Agents with Proactive Context Management](https://arxiv.org/abs/2510.24699)**
[[code](https://github.com/Alibaba-NLP/DeepResearch)]
- **[ACON: Optimizing Context Compression for Long-horizon LLM Agents](https://arxiv.org/abs/2510.00615)**
[[code](https://github.com/microsoft/acon)]
- [Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models](https://arxiv.org/abs/2510.04618)
## ๐ฌ Papers - Memory in Cognitive Science
#### ๐๏ธ 2026
- **[The Geometry of Forgetting](https://arxiv.org/abs/2604.06222)**
[[code](https://github.com/Dynamis-Labs/hide-project)]
- **[A Neural Network Model of Free Recall Learns Multiple Memory Strategies](https://www.nature.com/articles/s42256-026-01274-0)**
[[code](https://github.com/Veritaria/rnn-free-recall)]
- [Subspace Communication in the HippocampalโRetrosplenial Axis](https://www.nature.com/articles/s41586-026-10481-z)
- [A Neural State Space for Episodic Memories](https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613(25)00284-0)
- [Dopaminergic Processes Predict Temporal Distortions in Event Memory](https://www.nature.com/articles/s41467-026-69950-8)
- [Awareness as the Heart of Working Memory](https://www.sciencedirect.com/science/article/pii/S1364661326000756)
- [Neural Activations and Representations during Episodic versus Semantic Memory Retrieval](https://www.nature.com/articles/s41562-025-02390-4)
- [Distinct Neuronal Populations in the Human Brain Combine Content and Context](https://www.nature.com/articles/s41586-025-09910-2)
#### ๐๏ธ 2025
- [The Memory Paradox: Why Our Brains Need Knowledge in an Age of AI](https://arxiv.org/abs/2506.11015)
- [Neural Population Activity for Memory: Properties, Computations, and Codes](https://www.cell.com/neuron/fulltext/S0896-6273(25)00854-2)
- [How Prediction Error Drives Memory Updating: Role of Locus CoeruleusโHippocampal Interactions](https://www.cell.com/trends/neurosciences/abstract/S0166-2236(25)00189-4)
- [Towards Large Language Models with Human-Like Episodic Memory](https://www.cell.com/trends/cognitive-sciences/abstract/S1364-6613(25)00179-2)
#### ๐๏ธ 2024
- [A Generative Model of Memory Construction and Consolidation](https://www.nature.com/articles/s41562-023-01799-z)
---
## ๐ Memory Security & Defense
- **[Agent Memory Guard](https://owasp.org/www-project-agent-memory-guard/)**
[[code](https://github.com/OWASP/www-project-agent-memory-guard)]
_OWASP runtime defense layer that screens agent memory writes for poisoning: multi-layer validation with semantic anomaly detection, entropy scoring, and provenance verification._
- [From Untrusted Input to Trusted Memory: A Systematic Study of Memory Poisoning Attacks in LLM Agents](https://arxiv.org/abs/2606.04329)
(The MPBench Paper)
- **[inspeximus (formerly mnemo) poisoning probes](https://github.com/DanceNitra/agora/tree/main/research/probes)**
[[attack](https://github.com/DanceNitra/agora/blob/main/research/probes/memory_defense_layer_probe.py)]
[[defense](https://github.com/DanceNitra/agora/blob/main/research/probes/memory_gate_defense_probe.py)]
_Runnable probe scripts demonstrating that provenance written into a memory record is forgeable, and that a retrieval-time corroboration gate raises the cost of memory-poisoning attacks._
---
## ๐ฐ Articles
#### ๐๏ธ 2026
- [Harness Engineering for Self-Improvement](https://lilianweng.github.io/posts/2026-07-04-harness/)
- [221 Agents: Multi-Agent Coordination Lessons](https://web.archive.org/web/20260427040129/https://blog.kinthai.ai/221-agents-multi-agent-coordination-lessons) (archived; original site currently unreachable)
- [OpenClaw Multi-Tenancy: Why VM-Per-User Does Not Scale](https://web.archive.org/web/20260429092735/https://blog.kinthai.ai/openclaw-multi-tenancy-why-vm-per-user-doesnt-scale) (archived; original site currently unreachable)
#### ๐๏ธ 2025
- [Survey of AI Agent Memory Frameworks](https://www.graphlit.com/blog/survey-of-ai-agent-memory-frameworks)
#### ๐๏ธ 2024
- [Memory in Language Model-Enabled Agents](https://yuweisunn.github.io/blog-1-06-24.html)
- [Mastering LLM Memory: A Comprehensive Guide](https://www.strongly.ai/blog/mastering-llm-memory-a-comprehensive-guide.html)
#### ๐๏ธ 2023
- [LLM Powered Autonomous Agents](https://lilianweng.github.io/posts/2023-06-23-agent/)
---
## ๐ฅ Workshops
#### ๐๏ธ 2026
- [CoRL 2026](https://2026.corl.org/): [Workshop on Memory for Robot Foundation Models (RoboMem)](https://corl2026-memory.github.io/)
[[schedule](https://corl2026-memory.github.io/#schedule)]
- [ICLR 2026](https://iclr.cc/Conferences/2026): [Workshop on Memory for LLM-Based Agentic Systems (MemAgents)](https://sites.google.com/view/memagent-iclr26/)
[[proceedings](https://openreview.net/group?id=ICLR.cc/2026/Workshop/MemAgent)]
#### ๐๏ธ 2025
- [ACL 2025](https://2025.aclweb.org/): [The First Workshop on Large Language Model Memorization (L2M2)](https://sites.google.com/view/memorization-workshop)
[[proceedings](https://aclanthology.org/volumes/2025.l2m2-1/)]
---
## ๐ Citation
To cite this collection itself, use the metadata in [`CITATION.cff`](CITATION.cff) (GitHub's "Cite this repository" button), or:
```bibtex
@misc{zhang2025awesomeagentmemory,
author = {Zhang, Dell and Sun, Changzhi and Luo, Jixiang and Chen, Xiangyu and Li, Xuelong},
title = {Awesome Agent Memory: Curated Systems, Benchmarks, and Papers on Memory for {LLMs}/{MLLMs}},
year = {2025},
howpublished = {\url{https://github.com/TeleAI-UAGI/Awesome-Agent-Memory}}
}
```
This list grew out of the maintainers' SIGIR-AP 2025 tutorial, which you can cite as the related publication:
```bibtex
@inproceedings{zhangConversationalAgentsRAG2025,
author = {Zhang, Dell and Feng, Yue and Liu, Haiming and Sun, Changzhi and Luo, Jixiang and Chen, Xiangyu and Li, Xuelong},
title = {Conversational Agents: From {RAG} to {LTM}},
year = {2025},
isbn = {9798400722189},
doi = {10.1145/3767695.3769671},
booktitle = {Proceedings of the 2025 Annual International ACM SIGIR Conference on Research and Development in Information Retrieval in the Asia Pacific Region (SIGIR-AP)},
pages = {447โ452},
location = {China}
}
```
---
## Star History
Regenerated weekly by [a scheduled workflow](.github/workflows/star-history.yml); the previous [star-history.com](https://www.star-history.com/) live chart broke when GitHub restricted the stargazers API to repo admins and collaborators in June 2026.
---
**If you find this page helpful, please give it a โญ๏ธ โ starring also keeps updates in your GitHub feed.**
Made with โค๏ธ by [Bloo-Mind AI Ltd](https://www.bloo-mind.ai/) and the Ubiquitous AGI team at TeleAI.
