Agent frameworks & infrastructure
Agent memory & context
4,032 agents · most starred first
Agent memory & context, page 8 of 41
12-layer memory architecture for OpenClaw agents — knowledge graph (3K+ facts), semantic search (multilingual, 7ms GPU), continuity + stability + graph-memory plugins, activation/decay system, domain RAG. Agents reconstruct themselves from files on every boot.
Open-source relational AI framework with identity persistence, memory, and MCP integration. Build relationship-aware AI agents that remember, grow, and maintain continuity. Built on Claude Agent SDK.
FastMemory is a topological representation of text data using concepts as the primary input. It helps in improving the RAG(by replacing embedding and vectors entirely), AI memory and LLM queries by upto 100% as in the huggingface benchmarks(22+ SOTA)
memweave is a zero-infrastructure, async-first Python library that gives AI agents persistent, searchable memory — stored as plain Markdown files
A Plug-and-Play, Cost-Efficient, and Lightweight Memory Plugin for Long-Horizon LLM Agents
💎 LaPis — Local AI memory for coding agents. Persistent SQLite memory for Pi, Claude Code, Hermes & MCP.
Memory operating system for AI agents — persistent memory, BM25 search, atomic decomposition, reflection engine, immune system. Zero API calls. 89% fewer tokens than alternatives. MCP + LangChain + OpenAI SDK.
Semantic retrieval for AI agents: high-recall memory and code search that cuts coding-agent token usage by nearly 50%
Builds up to a goal in a field you have no expertise in: splits it into the foundations that goal actually needs, mobilizes what you already hold, learns what is missing, and leaves every solved path behind for the next goal. Nothing is called done until a check passes.
MemXT is a Local long-term memory for Claude Code & coding agents. 100% local
🧠 𝑴𝒆𝒎𝒐𝒓𝒚-𝑷𝒍𝒖𝒔 is a lightweight, local RAG memory store for MCP agents. Easily record, retrieve, update, delete, and visualize persistent "memories" across sessions—perfect for developers working with multiple AI coders (like Windsurf, Cursor, or Copilot) or anyone who wants their AI to actually remember them.
Cloud-native, durable state for AI agents: WAL+snapshots, watch streams, idempotency, leases, TLS/mTLS, capability tokens, Python/TS SDKs, Helm.
Lightweight, local-first memory for AI agents. Hybrid vector + BM25 search, LLM-powered fact extraction, zero infrastructure — just pip install.
Agent-first Memory Bank skill pack for Codex CLI, Claude Code, and similar runtimes.
SQL Server connector for Semantic Kernel plugin and Kernel Memory
LLM using long-term memory through vector database
This project contains a step-by-step guide on how to design an advanced agentic memory for your LLM based applications.
It is a simple, fast, and hard-durable embedded database designed specifically for AI agent memory. It provides a single-file-like experience (no server required) but with native support for vectors, graphs, and temporal search.
RinAI agentic chat stack. Includes graphRAG memory, parallel tool calling, DeepSeek R1 powered web search, advanced LLM gateway and active context summarization.
蒸留蔵 — distilled long-term memory for agents: recall by meaning, writing gated by evidence, one kura per agent mode. Ships as a DeepSeek Harness plugin and an MCP server.
Branch from any answer and keep branching at any depth. Recursively isolated follow-ups that leave the main DSH task untouched.
Semantic memory system with knowledge graph, spreading activation, embedding-based recall, autonomous dream consolidation, and C++ LSTM+kNN pattern learning for any /MCP and the Hermes Agent.
Outcome-based persistent memory MCP server for Claude Code and OpenCode. Good advice promoted, bad advice demoted. pip install roampal.
Give any local LLM a billion-token memory. Open-source, local-first context + memory engine for Ollama and AI agents — runs on your own machine, free. 🧠⚡
Evidence-backed retrospectives for AI coding-agent histories. Offline indexing, scoped workflow improvements, and bilingual documentation.
Evaluate your agent memory on real-world dialogues, not LLM-simulated dialogues.
FlowState QMD — Anticipatory Memory for AI Agents. Hermes 2026 Hackathon Entry.
The open engine behind Aedelgard: a persistent, self-hosted AI agent with a verbatim memory palace and a model-agnostic brain. Swap Claude, Gemini, OpenAI, Mistral, DeepSeek or a local model without losing the mind.
The operating system for Claude Code. Memory, accountability, journaling, knowledge graphs, pattern recognition — one install, every session compounds.
铃音的记忆写入协议 - an informatics protocol: fragmented, cue-based, symbolic, highly compressed information for AI agents; usable for memory writing and transmission, decodable approximately or via retrieval.
Dragon Brain — persistent long-term memory for AI agents via MCP (Model Context Protocol). Knowledge graph (FalkorDB) + vector search (Qdrant) + CUDA GPU embeddings. Works with Claude, Gemini CLI, Cursor, Windsurf, VS Code Copilot. 30 tools, 1121 tests.
[ACL'26 Oral] AgentOCR is a token-efficient framework that compresses multi-turn agent history by rendering it into images and adopting RL-driven self-compression
Search and memory across your AI coding sessions — one index over Claude Code, Codex, and Pi. Fuzzy-find and resume from the CLI, give agents recall via MCP, and get usage reports.
This repository demonstrates how a simple memory file can be incorporated into an Agents routine using the AutoGen framework.
Cognitive engine for AI agents — semantic memory, observations, embeddings, dream consolidation. Cloud Run service + MCP tools.
File-based context resolution for AI agent DAG workflows. Three primitives, zero dependencies.
Implementation of Evo-Memory style learning for LLM agents. Agents learn from outcomes, refine strategies, and get smarter with every task. 🚀 Features: Experience-driven memory architecture Semantic search + context synthesis Self-improving agents
Local-first memory layer for LLM apps and agents. Runs offline on Ollama, sentence-transformers and FAISS; OpenAI and Anthropic optional. Importance scoring, contradiction resolution, confidence-aware retrieval.
[ACL26 Findings] LiCoMemory: Lightweight and Cognitive Agentic Memory for Efficient Long-Term Reasoning
Persistent memory system for agentic AI via MCP - remember, recall, forget with semantic search with knowledge graph
KalamDB — a lightweight, real-time, storage-efficient SQL database. Designed for per-user data isolation and scalable performance — ideal for the AI era.
Memory hygiene skills for Hermes Agent — dreaming (3-phase consolidation)
The native macOS & iOS app for browsing AI agent memory files — Claude Code, OpenClaw, Codex, Cursor, Gemini
Backup and restore tool for Hermes Agent + OpenClaw installations
Memory-native AI agent runtime: Rust gateway + React dashboard + Python Operator + Kumiho graph memory
Local-first memory governance for AI agents: shared, reviewable, auditable memory via SQLite and MCP.
Paper List of Agent Memory Systems along Four Axes
Agent skill package that gives your agent persistent memory via an [Obsidian](https://obsidian.md) knowledge vault. Your agent automatically orients itself at session start, navigates project architecture through graph traversal, writes discoveries back to the vault.
Read-only Dashboard and Desktop plugin for inspecting Hermes Agent memory.
Hindsight-lite: Agent Memory That Learns lightweight version. Get inspired by hindsight.vectorize.io
Keep your skills sharp. Intelligence for agent context.
Open-source, self-hostable, zero-knowledge memory for AI agents. End-to-end encrypted — the server stores only ciphertext and can't read your agent's memory. Works with any MCP agent .
USL is the storage-first reference implementation of the Agent Session Protocol (ASP). ACP handles the live present (editor↔agent interaction); ASP handles the durable past and cross-runtime migration; USL is the storage engine behind ASP.
PersonaMem-v2: Towards Personalized Intelligence via Learning Implicit User Personas and Agentic Memory
mnestic is a maintained fork of CozoDB (transactional relational-graph-vector DB with Datalog), tuned as a substrate for agentic memory.
为 [DeepSeek](https://chat.deepseek.com) 网页版注入 Agentic 记忆系统、Skill 技能系统 、 系统提示词预设 、对话管理 和 Mcp 服务系统 的 Chrome 扩展。
a unified history for all your agents
This is the official repository for the paper "ACM: Agentic Context Management for Long-Horizon Tasks" [https://arxiv.org/abs/2607.23809]
Synthetic datasets, experiment protocols, and evaluation code for "Governed Memory: A Production Architecture for Multi-Agent Workflows"
double (dostoevsky reference) local memory system for agents
Tablestore for Agent Memory
Drop-in SDK for adding continual learning and long-term memory to any LLM agent.
Persistent memory for AI agents - Learn, remember, improve. Alternative to Mem0 with scoped learning, anti-patterns, multi-agent sharing, and MCP integration.
Agent memory built on chDB(ClickHouse embedded)
Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents
A markdown knowledge base for agents that's CLI and skill driven. Full featured with Google's OKF support and network vaults to share between users and systems.
(ACL 2026 SAC Highlight) Repo for "Agentic Memory: Learning Unified Long-Term and Short-Term Memory Management for Large Language Model Agents"
A brain-inspired, portable context database for AI agents
文件化经验库:让 AI agent 不再重复踩同一个坑。方法论 + 模板 + 可选工具。
Build Agents That Recall What Matters. Systematically engineer relevant context from chat history & business data. (Go SDK)
ambient memory agent, your machine's psyche
An experimental SDK for adding agentic memory and learning in a pluggable way
Claude Code's compaction engine, as a drop-in DeepAgents middleware
Persistent, governed memory for chat clients, IDEs, & agents. Share memory across clients without losing scope, provenance, or auditability.
An AI Multi-Agent memory system that remembers your code style, fixes, and preferences — for Hermes-Agent.OpenClaw,Opencode,Claude Code, etc.
Open source memory infrastructure for AI and humans. Turn scattered knowledge into AI-ready context across Claude, Cursor, ChatGPT, and any MCP client.
Longitudinal relational modeling for reflective agents: behavioral signals, evolving subject profiles, inspectable runtime governance.
Memory kernel stack for Hermes agent
User-owned memory and continuity layer for coding agents — provenance-aware semantic memory over PostgreSQL + pgvector.
hSUM: Stateful Understanding & Memory — local evidence for agents. Keep the source. Return the passage.
Open-source always-on Memory Agent for wearables. Turns your days into short- and long-term memory you can revisit and question, with an Agent that acts on it and builds new Skills from one sentence. The longer you wear it, the better it knows you.
Self-hosted second brain and operating engine for human and AI workspaces: topology, source evidence, memory, retrieval, agents, tools, workflows, and projections.
Psychology-grounded selective memory system for AI agents (Claude Code, OpenCode) with STM/LTM consolidation, decay curves, and learned importance weights
Agent skills for AI coding assistants (Antigravity)
Persistent memory for Claude Code — 3-5x longer sessions, 60-80% fewer wasted tokens. Branch-aware, self-healing, token-efficient.
Universal memory layer for AI agents. Remember, recall, consolidate.
Persistent memory system for Coding Agents (Claude, Opencode, Codex), infinite context, learning extraction, persona modeling and so much more!
Self learning agent skills
🦅 5-Layer Intelligent Skill Retrieval for Hermes Agent — hard triggers, FTS5, synonym dictionary, dense embedding, RRF fusion. Zero core modification.
Embedded, local-first agent memory. No server, no daemon, no mandatory cloud key.
Context File System for Agentic AI
A platform for hosting persistent AI agents that live, rather than stateless assistants that wake, perform, and vanish.
CliffCompaction — autocompaction for long-horizon coding agents
Mem-T: Densifying Rewards for Long-Horizon Memory Agents
Edge-first AI memory engine. Rust core, DuckDB single-file, sub-10ms search, 6 language bindings.
AI Workspace - Give your AI agents memory that spans across projects.
AGENTS.db is AGENTS.md on steroids. A file format and toolkit for deterministic context.
Neuroscience-grounded memory for AI agents — ACT-R activation, Ebbinghaus forgetting, Hebbian learning, cognitive consolidation. Rust core + Python/TS legacy. Published on crates.io as engramai.