Agent frameworks & infrastructure
Agent memory & context
4,032 agents · most starred first
Agent memory & context, page 13 of 41
A persistent, observable, and collaborative memory infrastructure
Applied reasoning, persistent agent memory, context engineering, and human-governed trust for AI coding agents — Claude Code, Claude Cowork, Codex, Cursor, Antigravity. Compatible with Obsidian.
AGENTIC · INFRASTRUCTURE — Every agent you run, working from the same brain. Open-source framework with a native shared Second Brain.
Self-hosted Dakera AI memory server — Docker Compose, Kubernetes, Helm, HA cluster setup, and monitoring.
A human-like memory for AI Agent
Open-source AI Agent OS: a folder of plain-text files that gives your AI coding assistant a lasting memory, house rules, sealed per-client work and a receipt for every run (v8 kernel). Not a chatbot, no people inside. MIT.
Context optimization layer for LLMs. 65-90% token savings with zero quality loss. Drop-in proxy for Claude, GPT, Gemini, and local LLMs (Ollama/VLLM/llama.cpp). Features KV cache-aware compression, session dedup, error cards, and Pichay-proven context paging.
High-fidelity transcript memory for Claude Code — index every session, recall byte-identical, search, prune, compose. Local-first, stdlib-only, MCP.
MoFlo — an opinionated, local-first AI agent orchestration toolkit for Claude Code: semantic memory, learned routing, gates, and spells. No API keys, no cloud, works out of the box.
Cross-session memory and recall for AI agents — git-synced knowledge base, hybrid semantic+TF-IDF search, auto-distillation with secrets scrubbing
WAKE.md for AI agents: compile project state so agents stop starting cold.
클로드코드 대화의 결정·약속·금지를 자동으로 기억하고, 검색·2D/3D 지식그래프·타임라인으로 보여주는 100% 로컬 기억 도구. 내보내기(6종)·가져오기(미리보기·되돌리기)·백업·복원, 의미검색 중단 알림, MCP 7도구 제공. 기억 데이터는 이 PC 밖으로 나가지 않으며 API 키·과금이 없습니다.
Lossless context folding for the Pi coding agent: only the oldest context changes; the fresh window stays untouched.
Persistent AI memory for Claude Code, OpenAI Codex and local agents. Open-source multi-agent memory with local ownership and portability. Keep what your agents learn across sessions, share project knowledge, and change tools without losing your memory. Plain files you own. No database, account or API key.
Jev-powered conversation memory: find past sessions and revisit decisions with original sources. A macOS workspace for OpenAI Codex. Manage AI conversations and agents, explore code with CodeGraph, and work with Git, Ghostty terminals, and Apple Notes in one app.
Local memory for Codex. Reuse work across chats without LLM calls to organize it.
DeepSeek Harness 社区记忆插件集:中文与扩展汉字全文检索、工具去重、混合召回、跨会话核心记忆、可追溯压缩与技能管理。8 个包、97 项单测;已验证 DSH 0.2.0-rc.2 完整 Web 宿主、真实持久化与冷重启。用户固定的记忆与技能受来源保护。
Open-source temporal memory engine for everyday AI applications.
沉淀、检索与治理一体的记忆插件,为你的 DeepSeek Harness 赋予可塑性记忆: Memory plugin with built-in curation and governance for DeepSeek Harness
一個以人類大腦結構為靈感的神經符號認知 AI 系統。具備三層圖譜記憶、虛擬神經化學、夢境反思引擎,以及 OpenClaw 整合,讓 AI 擁有真正的記憶、情感與自我。
Personalized Multimodal AI Agent via Google ADK & Gemini. Featuring Realtime Soniox STT, Long-term Memory (mem0), and LLM-generated Wiki Knowledge Base (inspired by Karpathy)
Example Langchain agent to demonstrate the potential use of Neo4j graph database as memory.
AI-IQ: Persistent context system for AI coding assistants. AI doesn't need knowledge — it needs relevant context. Hybrid search (FTS+semantic), graph intelligence, zero config.
Relational memory substrate for AI agents with MCP tools, identity continuity, and self-hosted persistence.
开源、用户自有、运行在个人 Cloudflare 账号中的 AI 记忆层。
OpenFused is a shared memory and messaging layer for AI agents, built on plain files. It gives agents persistent context, signed mail, and peer sync without locking them into a proprietary runtime.
SOUL.md governance framework for Hermes Agent — structured memory, skill management, and operational rules
Give Claude Code a permanent memory — 100% local, zero config, graph-powered
backlog: Your AI agent forgets between sessions. Backlog remembers.
A developer memory layer that automatically captures and organizes AI sessions, Git history, and development context into searchable knowledge.
CLI + MCP for agent context: a task layer over TickTick, Notion, GitHub, Obsidian & more (RRF retrieval, provenance, reviewed writes, undo) plus a knowledge-graph layer — ats kg: facts from any source, human-ratified. Recommended engines: Graphiti (server), LadybugDB (embedded).
Agent memory that needs no API key, no LLM, and no embedding provider. One command on Bun and it is serving MCP over stdio against a local SQLite store. Drop-in for the MCP reference memory server. Every memory keeps its source, who may read it, and the evidence that contradicts it.
Agent memory for Claude Code, Cursor, Codex: MCP server + Graph RAG over your Obsidian or markdown notes, a search layer for a Karpathy-style LLM Wiki. Wikilinks are the graph, SQLite the only store.
Estimate token usage after every agent step, trigger configurable alerts, and suggest practical ways to save context.
Agentic Memory System records work performed into strict schema making it actionable
A local-first agent memory lifecycle engine: typed memories in an append-only, integrity-chained log with dedupe, conflict detection, forgetting, compaction, verification, query, and portable export.
Gates 84.7% of tool-schema prompt tokens away for DeepSeek Harness: plugins declare themselves organs, a nerve impulse routes each command, reflex arcs fire with zero model calls, memory consolidates while idle, every failure is attributed before retry. Reproducible offline, MIT.
OpenFused is a shared memory and messaging layer for AI agents, built on plain files. It gives agents persistent context, signed mail, and peer sync without locking them into a proprietary runtime.
An agent memory system for small language models (SLMs), featuring semantic clustering and focused long-term memory retrieval. Findings of ACL 2026.
A bidirectional cognitive synchronization framework between you and AI agents
a basic bot to evaluate using neo4j graph database as chatbot memory.
GPT-3 Chatbot with long-term memory and external sources. Original work & inspiration by @daveshap
A (personal) knowledge base for LLMs that turns notes into structured, searchable long-term memory.
Structured memory plugin for Claude Cowork. Wikilinked knowledge base with tiered context loading and Obsidian integration.
AI Identity Through Grounded Principles - Compressed soul documents with full provenance tracking
Remembra Relay: one agent stops, the next one already knows. Session handoffs built from git facts, for the next coding agent in any MCP tool, on any machine. MCP server + CLI. MIT.
Local-first digital persona engine with long-term memory, voice, and channel integrations
Zaxy turns agent work into durable, auditable memory: a hash-chained Eventloom log as the source of truth, an embedded temporal knowledge graph for reasoning (local-first, no sidecar; Neo4j/Postgres optional), cited Memory Checkout for compact context, and MCP tools for model-facing retrieval, capture, and feedback.
Local task-state ledger for AI coding agents. Preserve decisions, source conclusions, evidence, and checkpoints across Codex, Claude Code, Cursor, and other MCP clients.
A structured framework for giving AI assisted teams the architectural context they need to generate code that fits your system.
Git-backed Claude Code session backup with timeline view, folder analysis, deletion detection, and session restore
Local SQLite memory for LangGraph agents, with provenance, TTL, export, and MCP.
AI agent memory you can trust
Coordination ledger for coding agents: what was decided, what's still open, where you parked — durable across sessions, repos, and weeks. Memory tools store what the agent knows; Director tracks the state of the work. Run both: they don't overlap. Claude Code + Codex + OpenCode.
A 125-task benchmark for knowledge-mediated implicit associations and retrieval blind spots in long-term agent memory.
SGME 拾光记忆引擎:AI 全面接管记忆、技能库、WIKI 知识库三大模块,跨会话、多智能体共享记忆,AI 会一直记得你的偏好。ShiGuang Memory Engine · AI takes full charge of three core modules — memory, skills, and wiki knowledge base. Memory is shared across sessions and multiple agents, so your AI always remembers your preferences.
Local-first, LLM-agnostic memory system — hybrid retrieval (vector + lexical + graph), multi-signal ranking. Runs on 4GB RAM, CPU-only, no cloud.
DeepSeek Harness (DSH) plugin for self-improving AI agents: continual learning, persistent memory, cross-session knowledge, review-and-refine workflows, and automatic rollback.
DeepSeek Harness (DSH) 跨会话记忆插件:对话模型边干边记的 Markdown 项目记忆,支持专题文件渐进式披露、可选后台静默蒸馏、按项目隔离与热更新设置面板。机制对齐 Claude Code auto memory。 | Agent-maintained cross-session Markdown memory: progressive-disclosure topic files, opt-in silent distillation, per-project isolation. Claude Code-aligned.
LuckyAgent — 注重长任务规划和系统记忆性的Agent框架
Version control for AI agent memory. Commit, branch, merge, blame and time-travel what your agents know.
Mnemovela - Agent Cognition Runtime (typed memory, Git-like branches, hybrid retrieval, governed forgetting; embedded-first Python & Go). Open-source client SDKs (Python/TypeScript/Go), CLI, API docs, and prebuilt binaries for Mnemovela. Apache-2.0.
Self-hosted long-term memory for AI coding agents: Postgres+pgvector doc-RAG (component/constraint map + hybrid RRF + cross-encoder rerank) and bi-temporal personal memory. MCP-native, local-model friendly.
Persistent Knowledge Graph Memory for AI Coding Agents. Adds long-term context (goals, strategies, preferences) to Cursor, Windsurf, & VS Code via MCP.
Open-source causal graph memory for AI agents. 89.9% on LoCoMo. MCP server with ACT-R scoring, spreading activation, and active forgetting.
Native OpenAI Agents SDK session management implementation using Redis as the persistence layer.
4D Strategic Memory Engine for Autonomous AI Agents — store wisdom, not just facts
OpenChronicle is an open-source, self-hosted interaction engine for LLMs that makes conversations durable. It adds persistent memory, deterministic tasking, and auditable decision trails so work doesn’t reset each session. Provider-agnostic by design, it supports multiple interfaces (CLI, Discord, MCP) with explicit, privacy-aware routing.
Persistent shared memory and handoffs across any combination of AI coding agents — Claude Code, Codex CLI, Cursor, Aider — using nothing but markdown.
Lossless local memory for Claude Code. The full, unabbreviated version. Every tool call, every file edit, every thinking block — stored verbatim on your machine. Persistent memory without tokens.
Semantic retrieval for AI agents: high-recall memory and code search that cuts coding-agent token usage by nearly 50%
Session 接力棒:以 context engineering(為 AI 準備它需要知道的背景資料)為核心,把進度、 決策和下一步留在專案資料夾,讓不同的 AI 工具接手繼續工作。Your session relay baton — context engineering at its core: keep progress, decisions and next steps in your project folder, so supported AI tools can pick up where the last one left off.
Vant: Versatile Autonomous Networked Tool - Durable agent OS + memory + cold storage + runtime, with generational evolution, stored with version control. Built for long-running autonomous sessions, using git based persistence.
Mine reusable skills from your coding agents' transcripts (Claude Code, Codex CLI, or any agent) with outcome-weighted scoring and a local review queue
Pure-Rust port of CerebroCortex — brain-analogous AI memory system with ACT-R/FSRS activation, associative networks, 8-phase dream engine, and MCP-over-stdio. Drop-in replacement for the Python version, Pi-native, zero Python runtime.
A standalone memory engine for AI agents in .NET, with hybrid retrieval, graph memory, pluggable providers, and behavior-shaped memories.
Long-term memory infrastructure for .NET AI agents with semantic, temporal, spatial and robotics memory, hybrid retrieval, and pluggable providers.
The memory layer that gets better as you use it. What broke, why, what depended on what, even the "push this to Wednesday" kind of thing that never makes it into any doc. No one has to write it down. Anchor your context to your code. when the code moves, so does the memory. works on claude code, cursor, codex. Completely local.
The context exoskeleton for AI development: optimize the entire development lifecycle to reduce token usage and context occupancy.
Long-term memory infrastructure for .NET AI agents with semantic, temporal, spatial and robotics memory, hybrid retrieval, and pluggable providers.
SANCTIS is a cognitive architecture that gives LLMs structured tools to organize their own reasoning, maintain long-horizon coherence, and reduce drift. It installs stable thinking patterns that emerge more powerfully the longer the model runs.
Branch-Thinking MCP Tool A TypeScript-powered MCP server for managing parallel branches of thought, semantic cross-references, and persistent tasks. Features dynamic scoring, AI-generated insights, batch operations, and visual graph navigation for advanced agentic workflows.
Allows to save chat history of converation.
Oscillink — Self‑Optimizing Coherent Memory for Embedding Workflows
One memory, every terminal — a shared memory layer for any MCP client.
Neuroscience-inspired memory for AI agents — sleep cycles that rewrite and strengthen knowledge over time. TypeScript, PostgreSQL, MCP.
Structured local memory storage and retrieval for LLM agents
File-first memory infrastructure for AI agents, built with .NET 8 and Azure backends
Generate and maintain AI-readable project memory, worklogs, and handoffs for any repository.
Persistent local memory for Claude, Cursor & Codex. 21 MCP tools, hybrid retrieval (BM25 + vector RRF), bi-temporal queries, contradiction detection. SQLite + Knowledge Graph. No cloud, no API keys. npx @studiomeyer/local-memory-mcp
Any brain, any agent. Package an AI agent's brain — persona, MCP stack, skills, memory, config — into one open, signed bundle and install it into any framework (OpenClaw, Hermes, Claude Code) in one click.
Storage engine for AI agents to navigate, store, and retrieve structured data. PostgreSQL ltree, built-in versioning, scoped queries.
Codex long-task compaction continuity and subagent prompt configuration guide
Claude Code의 영구 기억 시스템 — 4-layer 파이프라인 (세션 검색 + 메모리 회수 + Memory Compiler). Karpathy LLM-as-Compiler 패턴 실증판.
Remembers your project so you don't have to explain it again. Memory across Claude, Codex, Cursor.
Local-first memory engine for AI agents
Mnemovela - Agent Cognition Runtime (typed memory, Git-like branches, hybrid retrieval, governed forgetting; embedded-first Python & Go). Open-source client SDKs (Python/TypeScript/Go), CLI, API docs, and prebuilt binaries for Mnemovela. Apache-2.0.
Build a second brain every AI can share: your content in Oracle AI Database 26ai — semantic search with in-database embeddings, a self-compiling wiki, four kinds of agent memory, and an MCP server that plugs it into Claude, ChatGPT, and any MCP client. Runs on the free local container or Always Free cloud.
DeepSeek Harness long-term memory & self-evolving plugin: L0 capture -> L1 memory extraction -> L2 scene grouping -> L3 user persona, auto recall injection + skill synthesis, fully local.