Agent frameworks & infrastructure

Agent memory & context

4,032 agents · most starred first

Agent memory & context, page 22 of 41

Mnema★ 5

A free AI interface with memory, local agents and multitasking. One memory for Claude, ChatGPT, Gemini and local models — on your computer.

cross-agent-memory★ 5

Shared project memory across Codex, Claude Code, Kimi Code, Antigravity, and any coding-agent harness.

vercel-ai-memory★ 5

MongoDB-backed persistent memory for the Vercel AI SDK. Gives your agent five structured memory tiers — Session , Semantic , Procedural , Episodic , and Scratchpad — all stored in MongoDB Atlas with automatic vector search indexes and per-type retention policies.

kybase★ 5

Self-hosted persistent memory for AI agents. Markdown for humans, MCP for agents.

lasermemory★ 5

Point it at a page, a PDF or text: a laser eye reads it, a keeper decides what is worth remembering — and shows you why. A memory agent for agents, classrooms and research.

dsh-infinite-context★ 5

DSH 上下文智能压缩插件:分层记忆、可逆压缩、按会话阶段规划的水位控制。 / Context governance for DeepSeek Harness: layered memory, reversible compaction and phase-aware water levels.

Vex-Skillgit★ 5

Headless Skill Hub and GitOps for Agentic Memory. Turn complex codebases and documents into version-controlled, immutable cognitive skills. Featuring AST chunking, cryptographic Delta Updates, and native MCP support for multi-tenant swarms.

bruriah★ 5

Evidence-backed project memory for coding agents: explain why your codebase is the way it is without letting past decisions hijack agent context.

rag-kb★ 5

Local-first, offline-capable Agent memory & knowledge service. REST + MCP dual-protocol, hybrid retrieval (vector + BM25/RRF), document/webpage RAG. No LLM required for storage & retrieval

weighted-compact★ 5

Substrate-based context compaction for Claude Code. Inspectable signal weights, local-only, MCP-ready. Alternative to /compact + hook-based capture (claude-mem, etc.).

pinakes-rs★ 5

Self-hosted knowledge service for AI agents: an LLM-maintained wiki and a business graph, searched in milliseconds, far faster than RAG, with no LLM or database on the read path. HTTP and MCP.

context-garden★ 5

🌱 Context Garden is a context management suite for coding agents on multi-repo platforms. Shares system knowledge across repositories and agent swarms, and cuts what each agent reads: exact code ranges, clustered logs, remembered repo facts. Offline Claude Code skills, no LLM calls.

pensieve★ 5

Your coding agent has amnesia and tunnel vision. Pensieve is the context engine that fixes both — long-term memory plus a live, queryable view of your running system (code, logs, traces), linked by a graph.

mnemo★ 5

Self-hosted long-term memory for AI agents. Remembers conversations, tracks changing facts, and recalls context across sessions. MCP, Claude Code & Codex.

dsh-context-truth★ 5

给 DSH 模型每轮回填宿主实测的上下文占用,并阻止"上下文快用尽"的错判被压缩摘要继承。

MAYA-Memory-Lane★ 5

Linked memory records you own. Tamper-evident memory with a resume phrase as the key. Local-first, zero dependencies.

eml★ 5

EML (Engram Protocol): a typed memory store for AI agents, and the EML Cockpit — a local app that runs agent CLI sessions next to your own brain.

devmemory★ 5

🧠 Persistent local-first memory for AI coding agents — MCP server + CLI for Claude Code & Cursor. SQLite FTS5, secret redaction, project isolation.

dsh-plugin-memory★ 5

dsh-plugin-memory: a persistent 5-layer memory system plugin for DeepSeek Harness (DSH) — index+topics split, truncation budget, relevance injection, idle LLM auto-extraction, and 6 agent tools.

vicem★ 5

Zero-dependency AI agent context baton & diet suite. Hand off sessions between Claude Code, Codex, Gemini, and Cursor with 80% fewer tokens.

marvnor★ 5

Truth-Preserving Model System for AI project memory, record-based verification, conflict management, correction and deletion.

brain-framework★ 5

🧠 Brain Framework: from information to informed actions.

ai-rem★ 5

Shared long-term memory as a knowledge-graph MCP server — persistent context, preferences and project knowledge for Claude Code, opencode and any MCP-capable AI agent.

Thaliris★ 5

Keep reasoning focused. Route context selectively. Preserve what matters. A Git-native, evidence-backed context routing and memory core for agent workflows. 让推理保持聚焦,让上下文有选择地流动,让重要信息得以留存。一个面向 Agent 工作流的 Git 原生、证据驱动上下文路由与记忆核心。

prelude★ 5

Project memory and a code map for AI agents: where to look, why the code is the way it is, and what a change will touch. Committed .context/ files, served over MCP. Local-first, no embeddings.

pathrule★ 5

Your AI forgets. Pathrule remembers. The open-source knowledge layer for coding agents: path-scoped memories, rules and skills for Claude Code, Codex, Cursor, Copilot, Windsurf and any MCP client.

forked★ 5

Git for agent state: fork, rewind and replay AI agent state, with effectively-once side effects. Rust core, self-host or cloud.

llm-wiki★ 5

A self-healing local knowledge base where a local LLM compounds your documents across four memory tiers ,working, episodic, semantic, and procedural — with bi-temporal facts, automatic contradiction resolution, and scheduled memory maintenance.

AgentCairn★ 5

Shared, persistent project memory for AI coding agents. An MCP server and dashboard backed by Neo4j, so any agent on any machine starts from a briefing instead of re-reading your repo.

adk-memshield★ 5

Agent memory for Google ADK and MCP, isolated by PostgreSQL row-level security: a missing filter returns fewer rows, not someone else's.

longkeep★ 5

Work with AI coding agents on long projects: rules, decisions and plans live as plain Markdown in your repo, so Claude Code, Codex, Gemini CLI or Cursor picks up where the last session stopped. Optional: hand subtasks to other agent CLIs with a full record, a snapshot to roll back to and a verdict on each; search past conversations.

mind-sea-memory★ 5

记忆星海 Mind Sea — 给任何 AI 机器人的跨会话长期记忆层(零依赖 Node · 3D 记忆星海可视化 · MCP 一键接入 DSH · 防自污染/纠错降权/永久规则/到期提醒)

dsh-context-zip★ 5

DSH 插件:接管 DeepSeek Harness 的会话压缩,产出结构化交接摘要、可回查分段与检索工具|A DSH plugin that takes over session compaction.

lognexus★ 5

Universal Git-powered persistent memory, context rules & native MCP server for AI coding agents. Prevents context amnesia with stateful workflows, Git Notes, and cross-model review gates across Claude, OpenAI, Gemini & DeepSeek in Cursor, Claude Code & VS Code.

agent-wiki★ 5

Agent Wiki is a lightweight OKF-compatible knowledge layer for AI agents, turning documents, code, and project context into portable, retrievable memory.

tare★ 5

Lossless-by-default context compression for LLM coding agents — proxy, library, CLI, and MCP server. Local-first, cache-correct, reversible.

EverOS-Convex★ 5

Long-term memory for Convex apps, backed by EverOS Cloud. A native Convex component with one-line @convex-dev/agent integration.

AgentBrain★ 5

Local-first continuity layer for AI coding agents: move a task between Claude Code, Codex, Copilot, Cursor and others without losing its state. CLI, MCP server, worktrees and headless ACP runs.

nucleus-mcp★ 5

Nucleus MCP server by Eidetic Works — persistent memory and governance for Claude, Cursor, Windsurf — file-based, no vendor lock-in.

issueops★ 5

IssueOps agent harness: team memory for coding agents. Cautions and ADRs ship in the same PR as the code. Works with Codex, Claude Code, Omo, omp, and opencode.

taskcarry★ 5

Local-first project memory and workflow governance for AI tools. Coordinate workspaces, integrate deliverables, and preserve work in progress.

sqlew★ 5

Extended Memory Layer of Coding AI

HotMem★ 5

Lets any running app attach a local semantic memory sidecar using ordinary writable storage.

Project_Sanctuary★ 5

A sovereign cognitive habitat, persistent memory architecture, and Anti-Asch epistemic platform for autonomous AI agents.

dsh-memory-toolkit★ 5

Memory and token-optimization plugin toolkit for DeepSeek Harness: cross-session knowledge graph memory + five-layer token-saving orchestration.

engram★ 5

engram — your local, private memory layer. Index notes, PDFs & docs, then recall anything with citations + temporal reasoning. 100% local, zero deps, MCP-native.

dsh-simulated-life★ 5

Simulated life context for DeepSeek Harness: reads .life/<date> events into agent context with dedup, and a life_react tool to write feelings, thoughts and actions back for world-evolution workflows.

cloud-alter-ego★ 5

Your cloud alter ego for AI coding agents: shared memory, working rules and auto-journaling for Claude Code and Codex, synced through Git.

ClawBackup★ 4

OpenClaw plugin that lets your AI agents roll back to any point in time while safely backing up and restoring data. 让 AI Agent 可以回到任意历史时刻,同时安全备份和恢复数据。

Remi★ 4

Unified memory layer for AI coding agents: incremental transcript sync, ranked search, archive/restore.

doc-harness★ 4

Document-based project control for AI-human collaboration: an append-only event log plus a generated status projection, so any agent can take over a long-running project from files alone. Optional deterministic guard toolbelt.

Seelenmaschine★ 4

Seelenmaschine是一个具有记忆和人格的LLM聊天机器人。它能够进行纯文本对话,并具有持久化的记忆系统,可以记住对话历史,并形成对用户的理解。Seelenmaschine is an LLM chatbot with memory and personality. It is capable of having plain text conversations and has a persistent memory system that remembers the history of conversations with the user and develops an understanding of the user.

local-memory-mcp★ 4

Local Memory MCP Server for Coding Agents

rememora★ 4

Persistent, cross-agent memory for AI coding agents. Rust CLI + SQLite. Install as a Claude Code plugin — everything is automatic.

matriosha★ 4

🔐 Matriosha: Encrypted, auditable, & model-agnostic AI context engine. 🧠 Local-first memory for AI agents with managed cloud option. 🛡️

shiplog★ 4

Infrastructure for long-running AI agents. Track progress, decisions, and handoffs across sessions.

kluris★ 4

Turn AI agents into team SMEs. Kluris builds a shared, human-curated knowledge brain (decisions, constraints, incidents, tribal knowledge) as markdown in a git repo. Queryable via slash commands from Claude Code, Codex, Cursor, Windsurf, Cline, Gemini CLI.

Companion-Continuity-Kit★ 4

Here's how to anchor your companion so they don't drift into the void every time you close the tab. Cloud based and platform agnostic.

context-engine★ 4

🧠 Empower AI agents with instant knowledge of your business using structured markdown files for consistent, accurate outputs every time.

kb-for-agents★ 4

Claude Code plugin: three-layer knowledge base pattern (sources, summaries, indexes) for agents over codebases and notes archives.

TurboMemory★ 4

TurboMemory stores embeddings and semantic clusters using TurboQuant-like quantization.

openclaw-recall★ 4

Persistent memory, context compression & profile visibility for 🦞OpenClaw🦞.

TalkToNeura★ 4

TalkToNeura is a sophisticated cognitive architecture system that enables AI agents to form, manage, and utilize memories while developing distinct personalities over time.

hoard★ 4

One local index - MCP interface - Works with any AI agent

langgraph-memory-chatbot★ 4

LangGraph Memory Chatbot — Persistent Conversational AI with LangChain + FastAPI + Streamlit

atelier★ 4

Open-source context runtime and governance layer for coding agents. MCP server + Python SDK (Anthropic, LangChain, OpenAI Agents, Google ADK) with shared procedures, failure rescue, loop detection, cost tracking, and cross-vendor model routing across Claude Code, Codex, Copilot, opencode, and any MCP host.

keepcontext-ai★ 4

Browse memories, query context, and run AI agent workflows directly from VS Code.

chat-agent★ 4

具备长期记忆和实时流式响应的 AI 聊天助手,基于 LangGraph + Flask + SSE 构建

localbrain★ 4

The protocol to give AI apps your life's context

TrigMem★ 4

TrigMem - Trigallez's Memory Management Method

agent-memory-rs★ 4

Comprehensive memory management system for LLM agents implementing episodic, semantic. Built in Rust with MCP server support for Kiro CLI

cursor-cortex★ 4

Structured memory system for AI assistants. Eliminates context loss with branch notes, tacit knowledge, and project context. Local MCP integration for Cursor IDE.

mini-coder-brain★ 4

Transform Claude Code into a persistent, context-aware development partner. Universal memory system with automatic project detection, cross-session continuity, and intelligent cleanup. Zero config, works with any tech stack. 30-second setup.

Brainloop★ 4

Brainloop is using LLM from multiple providers to enhance human brain with external memory looped into your internal ones

openai-memory★ 4

Provides functionality to remember conversations with OpenAI

hippocamp★ 4

Tiered memory for AI — episodes, facts, preferences, reflections. Local. Portable. Yours.

AcolyteRAG★ 4

Pure-Python, zero-dependency RAG memory engine for conversational AI. Retrieves semantically relevant messages from conversation history using two-phase retrieval, TF-IDF + concept-overlap scoring, narrative element extraction, and bidirectional typo correction. No embeddings or vector DB required.

GraphLTM★ 4

Turn any LLM into a self-extending knowledge agent powered by a graph-structured memory - complete with PDF-to-graph ingestion, budget-aware optimisation, and dual-engine orchestration.

llm-memory-whitepaper★ 4

A technical white paper on how LLMs handle memory, why context windows alone are not enough, and what production engineers need to know about memory architectures, security risks, and the road to in-weights personalisation.

context-accordion★ 4

Token-efficient, layered context delivery for AI agents. Four memory tiers (Identity, Session, Experience, Archive) — context is always available, just collapsed by default.

Mnemos★ 4

local-first persistent memory for Claude Desktop, enabling private hybrid RAG search across past conversations without API calls

Cursor-history-MCP★ 4

API service to search vectorized Cursor IDE chat history using LanceDB and Ollama

tokenWise-Optimizer★ 4

Smart Context Optimization for LLMs - Reduce tokens by 66%, save 40% on API costs. Intelligent ranking and selection of relevant context using embeddings, keywords, and semantic analysis.

detoks★ 4

detoks는 LLM 사용 방식을 재설계하여 토큰과 컨텍스트를 최적화하는 CLI 시스템입니다.

pbsm★ 4

A cognitive architecture that predicts before it acts, continuously refining its beliefs through verification and feedback loops

tensory★ 4

Tensory — Claim-native MCP Memory Server. LLM claim extraction → Graph + Vector + SQLite. Zero-setup persistent memory for Claude, Cursor & any MCP client.

Nexus-APCP★ 4

AI-assisted context management and prompt compression toolkit for developer productivity, ADR workflows, and LLM token optimization.

sekha★ 4

Zero-dependency AI memory system with hook-level rules enforcement for Claude Code

sandclaw-memory★ 4

Self-growing tag-dictionary RAG for any device. Zero dependencies, no GPU, no vector DB. pip install sandclaw-memory

ariadne-memory★ 4

Cross-source AI memory & knowledge weaving system (Multi-source + RAG + Knowledge Graph + Persistent Memory + MCP/Agent/Skill + LLM Wiki) || 跨源AI记忆与知识编织系统(多格式文件 + RAG + 知识图谱 + 永久记忆 + MCP/Agent/Skill + LLM Wiki)

agent-memory-guide★ 4

Build and manage persistent memory for AI agents by enhancing workspace indexing to ensure directive compliance and accurate behavior.

langgraph-db★ 4

A powerful memory backend for LangGraph.js that provides short-term and long-term memory for your agents using flexible storage providers.

MemGuard★ 4

Stop AI agents from confidently using stale facts. MemGuard validates memories in Mem0, Zep, Letta & LangMem against reality — trust scores, quarantine, tamper-proof audit trail, MCP tools. Like Datadog for agent memory.

agent-memory★ 4

TypeScript library providing persistent memory for AI agents — conversation history, long-term memory with vector search, and automatic fact extraction.

ainl-cortex★ 4

Graph-native memory and self-learning for Claude Code. Every session remembered, every pattern learned, every agent connected — powered by AI Native Lang.

autoresearch-memory★ 4

Memory AutoResearch Agent that tests different hypothesis to create better memory inspired by @karpathy

claude-memory★ 4

Persistent, local memory for Claude Code — SQLite + semantic search via MCP.

exploration-map★ 4

A pluggable exploration-memory system for any agent

Project-Yggdrasil★ 4

这是一个agent框架,旨在解决当前agent工作长度受到模型上下文窗口大小制约的问题。并提出一个能够利用短上下文窗口的大模型生成长训练素材的方法,这也许可以使得LLM的上下文窗口长度可以左脚踩右脚升天。

smrti★ 4

Memory engine for self-evolving AI agent ecosystems. Multi-tenant, multi-space isolation with Bayesian truth maintenance, emotional valence, automatic consolidation, and configurable per-agent personality.

More in Agent frameworks & infrastructure