Agent frameworks & infrastructure
Agent memory & context
4,032 agents · most starred first
Agent memory & context, page 5 of 41
MemPrivacy is a privacy-preserving personalized memory management framework for edge-cloud agents.
Markdown based AI agent memory with semantic search, hybrid retrieval, and offline-first sync between agents.
Markdown-only hierarchical memory for long-running AI agents. Zero dependencies, 7 months in production. Copy-paste install prompt for any agent.
Local, git-versioned memory for AI coding agents. No RAG, no Docker, no external service. Capture, compile, recall over a local LLM wiki with on-device embeddings and an MCP server.
Your durable workspace across AI agents.
S3, PostgreSQL, Azure Blob, GCS, MongoDB, and Redis remote backends for LangChain Deep Agents
Epistemological database for agent and human efforts, beliefs, and findings.
The best-benchmarked open-source memory system for AI coding assistants
(WIP) Open Knowledge Layer(OpenKL): The Advance Knowledge and Memory for Personal Agents.
An agentic memory database that cuts session tokens by 82–99%. One portable SQLite file — your agent's memory, anywhere.
DreamGraph is a graph-governed conceptual development environment (CDE) that turns plans, architecture decisions, and project knowledge into auditable execution through a persistent cognitive graph.
🎉 Congratulations! This work has been accepted for publication as a Findings paper at ACL 2026! Mistake Notebook Learning (MNL) is a novel, training-free memory framework that enables Large Language Model (LLM) agents to systematically learn from their mistakes
让 AI 主动记错题、自我迭代的协作机制 · A framework that makes your AI reflect on its own mistakes and stop repeating them
MADRE · Codex, Claude Code, Gemini CLI and OpenCode in one local chat on your machine, with one memory every agent shares. Home for all your projects and a Coordinator above them. Read-only by default; you unlock more per message. In Spanish and English. Public beta.
Inspired by traditional kernels — a lightweight, flexible, general-purpose kernel for Agent OS.
Memory that learns what works.
A chat interface that uses the REMO memory system with LangFlow
Portable Memory Harness for Agents. Grounding the Autonomous Era
A cognition aware database engine for AI agent memory. Purpose built in Rust with WAL, HNSW, knowledge graphs, and speculative context pre assembly. Not a wrapper, a ground up storage engine that thinks.
Git for AI Agent Knowledge. A persistent, versioned memory layer for AI systems. Hybrid GraphRAG retrieval. Runs entirely offline.
[NeurIPS 2026] xMemory: Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation
the agent experience layer · observability + memory for AI coding agents (Claude Code + Codex) · local-first, typed, yours
A Node.js AI chatbot with unlimited context and chat history.
Own your AI memory — import ChatGPT, Claude and Gemini exports, see what each AI knows about you. Checkpoint/restore and cross-model continuity for agents.
The simple yet powerful long-term memory manager between AI and you💕
Local-first RAG memory for AI agents: hybrid + GraphRAG search, MCP server with OAuth 2.1, plugin for Claude Code / OpenCode / Codex.
Agentic development environment for your coding agents, built in public
Recent memories stay detailed. Older ones grow concise. Important ones live on.近期记忆保持详细,旧记忆逐渐精简,重要记忆长期存在。
AI coding agents forget. They forget across sessions, and they forget across tools. A decision you reached in Claude Code at midnight is invisible to Cursor the next morning. Honeycomb fixes that. Learn something once; recall it everywhere,
A memory layer for Claude Code that learns as you work. Persistent, per-context memory as plain markdown in your own git repo. No server, no telemetry.
This repository holds enhanced Agents, built for the Microsoft AutoGen Framework. Debuting with a MemoryEnabledAgent with improvements in context/token control, portability, and PnP functionality.
A Comprehensive Library for Memory of LLM-based Agents.
Bounded, typed, ablatable memory contract for long-horizon LLM agents — Slay the Spire 2 testbed
ACL 2026 Code and Skills for “AMA: Adaptive Memory via Multi-Agent Collaboration”.
Drop-in AI memory layer with 2x faster retrieval and 10x lower cost. Fully compatible with Mem0 API. Migrate in 5 minutes without any code changes. Self-host for free.
A curated list of awesome open-source libraries for context engineering (Long-term memory, MCP: Model Context Protocol, Prompt/RAG Compression, Multi-Agent)
Deterministic memory layer for your agents
dsh-sieve: context engineering & token optimization plugin for DeepSeek Harness (DSH) — tool output filtering, context pruning, progressive skill disclosure. 36% smaller payload in offline replay. DSH 上下文管理与 token 优化节省插件。
Flexible storage solutions for agent sessions.
Skardi is a self-improving context framework for ai agents
SRA-Bench and SR-Agents: a benchmark and toolkit for skill-retrieval-augmented LLM agents.
Causal memory layer for AI agents — MCP server that records decision→outcome relationships. Survives compaction.
Turn isolated Claude sessions into a continuous working relationship
LLM based agents with proactive interactions, long-term memory, external tool integration, and local deployment capabilities.
🧠 Brain-inspired cognitive architecture plugin for AI agents — persistent memory, evolving personality, neurochemistry-driven mood, learning from mistakes. Fully local on OpenClaw.
Vector memory that gets smarter every time your agent uses it.
具备自主经验涌现与上下文治理能力的 Agent 长期记忆系统
Agent-native persistent knowledge management — compile knowledge once, query forever.
Self-hosted AI agent memory — MCP memory server with durable recall for agents
Local archive of your Claude Code and Codex sessions, plus reviewed long-term memory for coding agents and a doctor for their skill folders. Runs offline; no accounts.
A caching solution for asyncio
🧠 Active memory management system enabling functional infinite context for LLMs through cognitive workspace architecture
The explainable, local-first memory engine for AI agents. One ~9 MB binary fuses vector + graph + columnar under VelesQL; why() returns the evidence path behind every recall. No cloud, no glue code — runs on server, browser, mobile and desktop.
🧸 给AI加长期记忆的轻量网关 / A lightweight gateway that adds long-term memory to any LLM
Context Compression in Long-Horizon LLM-based Agents
Hybrid memory system for OpenClaw using Graphiti temporal knowledge graph
Local-first persistent memory for OpenAI Codex: governed recall, progressive disclosure, no hosted vector database.
Give every agent your team's brain on day one. The memory layer for AI agents. Everything your team knows. Cited, and always up to date.
Augment AI agents with long-term memory through knowledge graph 🧠
Claude Code / Codex 共用的长期记忆 MCP:本地 Markdown/Obsidian 存储,跨工具记住你的项目与偏好|Local-first shared memory MCP for Claude Code & Codex
Mission control for your agent fleet. One local-first board for Claude Code, Codex, Goose, Cursor and any MCP harness: durable tasks, live activity, cost attribution and audit exports.
Mnemosyne is a agentic memory and orchestration system designed to provide Claude Code with persistent semantic memory across sessions.
High-performance embedded off-heap storage engine for Java, built on memory-mapped files. Map/List/Set/Queue beyond the JVM heap, plus an AI memory layer with hybrid vector + BM25 search.
Re-anchor long conversations that drift through repeated local corrections.
Shared memory for AI coding tools: hybrid search, cited Q&A, harness memory discovery and a built-in Web console. Local index, MCP + HTTP.
Universal memory runtime for AI agents
Self-hosted semantic memory for AI agents, served over MCP. Persistent sessions, experience scoring, a graveyard for dead ends and skills compiled from what worked, plus knowledge from the RSS feeds you read. One Rust binary on SQLite: a tray app for macOS, Windows and Linux, or a server.
An example chatbot application built on the Letta API, which makes each chatbot a stateful agent (agent with memory) under the hood.
(ACL 2026 Findings) Learning on the Job: An Experience-Driven, Self-Evolving Agent for Long-Horizon Tasks
The source code of Mem-Gallery: Benchmarking Multimodal Long-Term Conversational Memory for MLLM Agents (ACL2026 Main).
[CVPR 2026 Highlight] WorldMM: Dynamic Multimodal Memory Agent for Long Video Reasoning
The Letta Agent SDK is the SDK for stateful agents. Each agent has its own identity and long-term memory, and keeps both across conversations, models, and the computers it runs on.
ReflectWorld-MM: An Entity-Oriented Multimodal Memory System for Open-Ended Video Stream
基本可用 Universal context-compression proxy for AI coding agents
An advanced, UI-powered AI second brain Agent Skill for OpenClaw, Hermes Agent, WorkBuddy, TRAE, and QClaw. ZhiGui uses long-term memory to help you make better decisions, automatically plan tomorrow, resolve conflicting notes, and deliver each day's plan through your existing agent channels.
A memory governance engine for multi-agent cloud knowledge bases — one knowledge base shared by every agent and every device, with consistency guaranteed. Pure Markdown, nine governance automations, zero dependencies.
Portable, verifiable work state for coding agents
Collaborative Memory-Augmented Agentic Recommender System
Memory layer for AI agents — SDKs, CLI and MCP server for Membase
Cortex: Advanced Memory System for AI Agents
A unified agent memory workspace for human and agent
MemoHarness: Agent Harnesses That Learn from Experience
Compiles AI agent traces and truns them into reusable context.
Model-controlled context blocks for DeepSeek Harness PTC mode
Memora is an agent that aims to replicate the human memory for AI assistants / agents.
Staged self-improvement engine for Hermes-style memory, skill, and fact updates with explicit review and apply/discard gates.
transformers safetensors qwen3 text-generation conversational arxiv:2507.18071
Official implementation of paper "Cache Engineering for Agents"
A selective learning and memory substrate for agentic systems — typed, revisable, decayable memory with competence learning and trust-aware retrieval.
Implementation of contextual engineering pipeline with LangChain and LangGraph Agents
YantrikDB memory provider for NousResearch/hermes-agent — self-maintaining memory with benchmarked recall, self-tuning ranking, contradiction tracking, proactive hygiene, and explainable recall.
Compiles AI agent traces and truns them into reusable context.
Atlas — Agent Memory on Elasticsearch. Three indices, hybrid recall with a reranker, supersession, decay, and per-user DLS isolation.
Navigational memory for coding agents: chart discoveries as you explore, cache-bust when files change
The Context Engineering framework for building smarter, more reliable AI Agents.
Compiles AI agent traces and truns them into reusable context.
Local-first AI companion daemon — an agent harness with its own long-term memory. Remembers you, grows with you, extends and repairs itself. 跑在你自己电脑上的 AI 搭档守护进程:会记住你、随你成长、能自加技能、能自我修复。
Mac app to create context for agents: writing, reading, voice, terminal, bookmarks, and clipboard.
Zero-LLM cognitive memory architecture for AI agents
Memory 🧠 to your Personal ChatBot 🤖| LangChainAI and Databutton