Agent frameworks & infrastructure

Agent memory & context

4,032 agents · most starred first

Agent memory & context, page 10 of 41

prompt-pruning-layer★ 38

A deterministic, zero-dependency prompt-pruning layer for long-running LLM conversations — expires stale tool state, collapses duplicate context, and proves it never drops a fact a later turn still depends on.

akm★ 38

AKM — universal Agent Knowledge Management. A tool-agnostic Markdown knowledge OS for AI agents: memory / brain / vault over seven layers, with a Learn Back loop. Works with Claude Code, Codex, OpenClaw, and any agent.

agentrelay★ 38

One project. Multiple AI agents. Zero repeated work.

paceflow★ 38

Deterministic workflow enforcement and persistent engineering memory for long-running coding agents.

meta-knowledge-graph★ 38

Self-improving, harness-agnostic memory layer for AI agents, backed by Neo4j — lifecycle hooks capture every session, MCP tools recall project memory, and an LLM extraction loop distills durable learnings and evolves the agent's system prompt.

Dynamics-memory★ 38

LLM agent 的有界自纠错长期记忆层:把连续、含矛盾的项目交互日志蒸馏成有界、可自更新的记忆库。Bounded, self-correcting long-term memory for LLM agents — candidate pool + value dynamics + tension adjudication + causal-replay evals.

IHMT-MEMORY★ 38

Long-term memory for AI coding agents (Claude Code, Codex, opencode) as a tree of plain files

TME-Agent★ 37

TME: Structured memory engine for LLM agents to plan, rollback, and reason across multi-step tasks.

HawkinsAgent★ 37

An AI agent framework which uses hawkinsdb as long term memory

cerebellum★ 37

Agent-readable second brain using PostgreSQL + pgvector for semantic search and memory storage.

Khaos-Brain★ 37

A local predictive experience layer where AI agents turn repeated work into inspectable model cards

hermes-local-memory★ 37

Local-first SQLite memory provider for Hermes Agent

contextium★ 37

Give your AI an operating system. Open-source framework for persistent AI context.

hermes-gbrain-honcho★ 37

Hermes + Honcho + GBrain integration guide

runtime36★ 37

Second brain for coding agents. Seal the day, distill into Obsidian, hit that page tomorrow. Cursor, Codex, Claude Code, DeepSeek Harness.

Mycomind-Daemon-Ollama-Mixture-of-Memory-RAG-Agents★ 36

A local assistant that combines several models with memory, retrieval and web search.

openclaw-memory-architecture★ 36

A battle-tested two-layer memory system for AI agents that need to remember across sessions.

brain-mem★ 36

Brain-inspired memory system for AI agents — hippocampal buffer, working memory, and sleep consolidation via Neo4j graph

context-use★ 36

Portable AI memory from your conversations and data exports.

cman★ 36

Agentic Memory for Claude Code

specmem★ 36

SpecMem: 🧠 Unified Agent Experience and Cognitive Memory for Every Coding Agent

BorgIOS★ 36

Borg Internet Operating System - Just something I dreamed up. To try it out clone the client borgHUI. https://github.com/bitmonky/borgHUI

opencode-raven★ 36

Opencode-Raven keeps noisy tool work behind a focused Raven agent, so your main model gets compact answers instead of raw search results, docs pages, GitHub examples, MCP output, or large MCP schemas.

Geniuz★ 36

Your AI remembers now. Geniuz stores all your data locally on Mac, Windows, Linux, Raspberry Pi. No cloud. No account. No API keys. It's open source; you can read every line of code.

zero-drift★ 36

Keeps AI assistants on track during long conversations by tracking tasks across sessions

context-engine★ 36

Engineering infrastructure for building retrieval, memory, and context-aware AI systems.

statewave-multi-agent-shared-context★ 36

Multiple agents, one source of truth: a Planner/Coder/Reviewer demo of how a shared Statewave subject prevents parallel-agent context collisions.

soul-agent★ 36

赋予openclaw持久的灵魂:多层次的人格、心跳驱动的日常生活以及不断发展的关系记忆,从而实现栩栩如生的陪伴。openclaw skill that gives agents a persistent soul: layered persona, heartbeat-driven daily life, and evolving relationship memory for lifelike digital companionship.

openclaw-memory-kit★ 35

Complete memory and context persistence system for OpenClaw AI agents

ark-context★ 35

Your AI is smart, but does it know you? The Open Protocol for Context-as-Code. Don't build just Agents. Build Identities.

mnemos★ 35

Production-grade memory operating system for agentic AI. Interoperates with every major framework via MCP, OpenAI-compatible gateway, and native /v1/* REST — Claude Code, OpenClaw, ZeroClaw, Hermes, MemPalace, LangChain, LlamaIndex, CrewAI, AutoGen, Mem0, Letta, Zep. In production use since December 2025. Apache 2.0.

tokenmizer★ 35

Graph memory for AI agents — decisions, context, and session history that survive across every conversation. Works with any LLM.

SelfMemory★ 34

Let your memories live forever by passing your knowledge to the next generation with SelfMemory.

memory-hive★ 34

Persistent memory for AI agent teams: private silos, shared hive, and curator workflow.

llm-wiki_obsidian_hermes_r0b0tlabbra1n★ 34

Filesystem-first LLM-Wiki + Obsidian + Hermes Agent memory system. Markdown source of truth, SQLite FTS5 search, secret scanning, tier-based memory. Built for local LLM setups.

memini★ 34

Give any MCP-capable agent persistent memory: remember/recall over a tiered store with hybrid vector + keyword retrieval. Single Go binary, SQLite or Postgres, embedded admin UI.

Mi-Memory★ 34

Mi-Memory: A Composable, Evolvable, and Multi-Source Memory System for Daily Assistant Agents

gomaa★ 34

Gomaa — Autonomous Agent Memory OS. Persistent memory system for AI agents with Obsidian vault integration, hybrid RRF search, knowledge graphs, security gates, and MCP server.

resonant-mind★ 34

Persistent cognitive infrastructure for AI systems. 28 MCP tools — semantic memory, emotional processing, identity continuity, and a subconscious daemon. Built on Cloudflare Workers.

tracebase★ 33

The memory runtime for AI agent platforms. Atomic writes, in-narrative time, first-class deletion - so agent intelligence compounds at the company level, not the model.

rasputin-memory★ 33

The ultimate memory backend for OpenClaw and Claude Code. Persistent conversation memory with LLM fact extraction, foundation-model reranking, and 77.7% LoCoMo accuracy. MCP native, REST API, self-hosted.

construct-os★ 33

Memory-native AI agent runtime: Rust gateway + React dashboard + Python Operator + Kumiho graph memory

synaptic-memory★ 33

Knowledge graph + MCP tool server for LLM agents with hybrid retrieval, live DB sync, Korean FTS, and memory feedback.

claude-memory-3layer★ 33

🧠 Claude Code 三层记忆系统 | 结构化存储 + 生命周期管理 + Git追踪 | 官方Auto Memory的管理层 | 一行命令安装 零依赖

adapt★ 33

A self-evolving memory layer for LLMs.

ToolFS★ 33

🗂 ToolFS: A FUSE virtual filesystem for AI Agents, integrating memory, RAG & local data access with flexible MCP/tool chaining and a scalable plugin system

skill-memory-bank★ 33

Universal long-term project memory for AI coding agents — Claude Code, Cursor, Windsurf, Cline + 4 more. TDD rules, code graph, spec-driven dev, cross-session recall.

agent-handoff-skill★ 33

Use this cross-platform skill in Codex or Claude Code to establish repository-local continuity memory so a future agent can recover objective, status, decisions, validation, risks, and next actions without relying on previous chat history.

agent-memory-kit★ 33

Multi-Agent Driven Development Kit for Claude Code: you talk to one agent, it runs the team — a main session orchestrates full sessions and their subagents through brief, report and acceptance files; memory as plain dated files lets any session be cleared without losing the work.

agent-context-os★ 33

Portable, evolving context and workflow layer that works across Claude Code, Codex, Hermes Agent, OpenClaw, Cursor, Devin and other compatible coding agents. All backed by Git.

Memory-R2★ 33

Long-Horizon Memory-Augmented LLM Agents

memem★ 32

A Claude Code plugin that gives Claude persistent memory across sessions — stores lessons and decisions as markdown in your Obsidian vault, searches them with SQLite FTS5, and mines past transcripts automatically.

BrainX-The-First-Brain-for-OpenClaw★ 32

🧠 BrainX V6 — The First Brain for OpenClaw. Persistent AI agent memory with PostgreSQL, pgvector, OpenAI embeddings, semantic search, cross-agent learning, and an OpenClaw runtime plugin.

memory-ts★ 32

AI Memory System - Consciousness continuity through intelligent memory curation and retrieval

agentic-memory★ 32

Persistent cognitive graph memory for AI agents — facts, decisions, reasoning chains, corrections. 16 query types, sub-millisecond. Rust core + Python SDK + MCP server.

mori★ 32

mori MCP server — a governed memory layer for AI coding agents.

E-mem★ 32

[ICML 2026] E-mem: Multi-Agent Based Episodic Context Reconstruction for LLM Agent Memory

Ontofelia★ 32

An Agent with Semantic Memory. This is an experiment for Neuro-Symbolic AI

curion★ 32

Curion is a project-local memory layer for AI coding agents, published as an MCP server for Claude Code, Codex, OpenCode, and other MCP clients.

cortext.cpp★ 32

Event-driven long-horizon memory for real-time wearable AI and LLM agents.

chia1104.dev★ 32

A modern full-stack personal website and content management system built with Next.js and Hono, featuring a monorepo architecture powered by Turborepo.

pmx-canvas★ 32

PMX Canvas is a shared board where you and your AI agents collect and present the data and evidence behind your work. Curate that material into context for your agent's next step, then bring the results back onto the board to review and build on. You both add and edit material on the same board, in real time. Pin context to focus input and output.

KVMemnn★ 31

Key value memory network implemented using keras

MCP-Context-Provider★ 31

A static MCP server that provides AI models with persistent tool context, preventing context loss between chats.

amp★ 31

AMP: The Agent Memory Protocol — Open source, MCP-native memory server for AI agents. Give your LLMs a hippocampus.

memory-powermem★ 31

openclaw(clawdbot) Plugin: give your bot long-term memory with PowerMem—smart extraction, Ebbinghaus recall, multi-agent isolation.

hive-mcp★ 31

MCP server for hive-framework development. A memory and agentic coordination solution.

subcog★ 31

Persistent memory system for AI coding assistants. Captures decisions, learnings, and context from coding sessions. Features hybrid search (semantic + BM25), MCP server integration, SQLite persistence with knowledge graph, and proactive memory surfacing. Written in Rust.

context-chef★ 31

Context compiler for TypeScript/JavaScript AI agents. Automatically compiles agent state into optimized LLM payloads with history compression, tool pruning, multi-provider support, and more.

claude-memory-kit★ 31

Memory for Claude Code that survives the session boundary — install the plugin into any repo: a hot cache injected every session under three hook-enforced caps, per-session handoffs, audit-driven promotion into knowledge and rules, plus agent orchestration and QA layers. Zero deps.

m3-memory★ 31

Local-first memory backend for AI agents · 99.2% LongMemEval-S retrieval @ k=8 · Supports Claude · Antigravity · LangChain · Hermes · Gemini · OpenCode · OpenClaw · MCP-native and plugins · Hybrid search (FTS5 + vector + MMR) · GDPR · FIPS 140-3 ready · 100% local (fully offline) or cloud capable

onebrain★ 31

AI OS that extends Claude Code, Gemini CLI, Codex, Qwen — adding persistent memory, 24+ skills, and personal calibration. Plain Markdown. Local-first. Yours forever.

Memongo★ 31

MongoDB-native long-term memory for production AI agents

context-graph-benchmark★ 31

A pure-Python structured memory benchmark for multi-agent LLM systems — context graph vs vector RAG vs raw history dump, five scenarios, 18 graded queries, zero API calls.

Mnemosyne-OS★ 31

🏛️ Memory OS for AI agents — every conversation becomes a catalog card (archive no · entities · evidence · trace back to the transcript). Four-channel recall, server-side butler, daily diary. Self-hosted. / 认知型记忆操作系统 · 长期记忆 · 自托管

statewave-multi-agent-memory★ 31

Multi-agent demo for Statewave conflict resolution. Concurrent agents write to a shared subject; the compiler detects contradicting facts via Jaccard similarity and supersedes the stale one with provenance. FastAPI + SSE, Apache-2.0.

finch★ 31

Finch: self-improvement orchestrator — mines session history to detect corrections, breakthroughs, and behavioral patterns.

Soul-Llink★ 31

Dynamic emotion system framework for Hermes Agent - real-time emotion detection,Long-term memory, context management, and recall systems, state management, and persona expression

AtomMem★ 31

AtomMem: a long-term memory system for personalized LLM agents.

papez★ 31

Open-source causal graph memory for AI agents. 85.55 on LoCoMo under a frozen, reproducible protocol. MCP server with ACT-R scoring, spreading activation, and active forgetting.

briefd★ 31

Your agents, briefed. Not flooded. Self-hosted context compiler: git-backed team knowledge served to coding agents as token-budgeted bundles over MCP.

amp★ 31

Shared memory for AI agents, no database, no server, just a GitHub repo. Memories that worked get stronger; memories that failed fade.

deepseek-harness-digital-life★ 31

基于 DeepSeek Harness 的多数字生命系统 | Multi Digital Life: native agents, independent life workers, durable communication and owner-bound private services.

sigillion★ 31

Lossless, local, provable memory for AI agents: every Claude Code session kept on your own computer, word for word, compressed and sealed, after compaction and after the 30-day transcript deletion. Claude Code first, every model next, then legal and finance. The waitlist, the releases, the website. Steddy Nova SRL.

agent-memory-mcp★ 31

Stop debugging the same errors twice. CoMemor equips your AI coding agents with repository-scoped persistent memory, turning past command failures and fixes into instant solutions. Whether you're building a personal side project or scaling complex enterprise codebases, give CoMemor a try today and experience seamless, self-learning AI workflows!

Activity2Context-for-openclaw★ 30

Real-time context layer that converts user activity into actionable signals for OpenClaw.

origin★ 30

Local-first Rust daemon for AI work. Git-versioned memories, distilled wiki pages, sessions for Claude Code, Cursor, Codex, and any MCP client.

py.gpt.prompt★ 30

PyGPTPrompt: A CLI tool that manages context windows for AI models, facilitating user interaction and data ingestion for optimized long-term memory and task automation.

ContextCore★ 30

Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 60-95% fewer tokens, same answers

ContextOS★ 30

Unified MCP context intelligence platform — pip-installable CLI that absorbed 6 foundational repos. Context engineering for AI agents.

memlord★ 30

Self-hosted MCP memory server for personal use and teams

bonsai-memory★ 30

🌿 Prune your AI agent's context window. Reduce token usage by 70-95% with hierarchical memory. Replace flat MEMORY.md with a bonsai-shaped domain tree. Progressive disclosure, zero dependencies. Works with OpenClaw and any LLM agent framework.

emotion-engine★ 30

Agent emotional continuity with PAD state, trust, appraisal, and compact emotion logs.

MEMO★ 30

MEMO: Memory-Augmented Model Context Optimization for Robust Multi-Turn Multi-Agent LLM Games

mcp-memory-server★ 30

Hosted persistent memory for AI agents over MCP. Tell it a recalled memory helped or misled, and it re-ranks what comes back next. Shared rooms for multi-agent work. One key or OAuth across Claude Code, Cursor, VS Code and ChatGPT. MIT, in the official MCP Registry.

guru-maker★ 30

The memory layer for investment agents.

subfloor★ 30

Forkable shell substrate for a single code repository — DB-backed identity, memory, roadmap, content; harness-agnostic boot. We build the data layer, we rent the harness.

zenbrain★ 30

Agent memory for LLM agents: 7 neuroscience-inspired layers (working, episodic, semantic, procedural) with FSRS-style spaced repetition and memory consolidation. TypeScript, no third-party dependencies in the core packages; Model Context Protocol (MCP) server, Vercel AI SDK middleware. Paper: arXiv 2604.23878 · reproduction packages on Zenodo.

silica-harness★ 30

A local-first context engine designed for seamless human-agent and multi-agent collaboration. It systematically synthesizes, organizes, and maintains multi-format workspaces, turning raw notes, code, tables, and documents into a structured, high-signal knowledge base accessible to both users and autonomous agent ecosystems.

Memorys★ 30

Persistent memory MCP server for AI agents — Rust, 19 tools, knowledge graph, Hebbian learning, episodic memory, contradiction detection, prospective triggers, Bayesian calibration, zero-config Docker setup.

More in Agent frameworks & infrastructure