Agent frameworks & infrastructure
Agent memory & context
4,032 agents · most starred first
Agent memory & context, page 17 of 41
Open-source context engineering protocol for AI agents: governed retrieval, bounded multi-agent handoffs, MCP, provenance, recoverable effects, and replay.
Open standard for how agents turn useful work into durable, portable knowledge.
Per-repo memory, outcome telemetry, and a calibrated-confidence gate for Claude Code, with MCP and AGENTS.md projections so other AI coding tools can read its context. Notes survive sessions; success claims need test evidence; your reverts are remembered. Local-only, stdlib runtime.
Local-first memory engine for AI-agent teams — private/team/project ACL, associative recall, federated sync, and a built-in MCP server. One SQLite file, no LLM required. Apache-2.0.
A cognitive memory architecture for agents: ACT-R activation, association, consolidation, and forgetting. sqlite, zero dependencies.
Capture, validate, and transfer AI agent expertise. Across sessions, platforms, and teams.
Persistent Coordination and Memory Layer for AI Coding Agents powered by langGraph.
Your project remembers across coding agents. Compact, cited project memory for Codex, Claude Code, Cursor, and compatible tools.
🪐 Growmos: A new approach to agentic AI—a living system expanding its knowledge like the universe, compounding its capabilities, and getting more work done with fewer tokens.
Turn corrections from AI agent logs into guidance for similar tasks. Local memory with semantic retrieval, prompt-context hooks, and response checks.
Local-first memory for AI agents. Keep decisions and evidence across sessions. Recall with sources through MCP.
CPersona — persistent memory for AI agents over MCP. One SQLite file you own, hybrid vector + full-text + keyword search, no LLM in the loop, and honest reporting when recall degrades. MIT licensed.
Field guide and local-first tools for token and context optimization in LLM agents.
Local-first persistent memory infrastructure for DeepSeek Harness: hot bootstrap, Chinese-BM25 cold recall, lease-lock transactional writes, read-only governance
Lunar Vault is a self-contained memory infrastructure for Codex. Features zero-dependency ONNX embeddings, advanced reranking, and secure one-click installation without leaking personal data
Long-term memory MCP server for AI coding agents. SQLite + FTS5 + vector search, CodeGraph, error learning, compaction survival. Beats Mem0 on benchmarks. No cloud, no API key.
Lightweight graph engine for AI graph context, memory, and agent harness with Rust Core.
Local, searchable project memory for AI coding agents. Markdown source of truth, MCP interface, safe structured updates — no cloud.
"Infrastructure for continuity across discontinuity — for AI instances and the humans who work with them."
Infinite context without the infinite context window. Compresses agent sessions (Claude Code, Codex, Grok, Hermes) into a searchable HKP1 pack — full-history RAG, ~60x lower context cost, 0 deps.
Architectural Moat: JIT-JEV Context OS — Epistemic runtime & JEV System 1 context gate for AI agents (L0 SQLite WAL <3ms, Epistemic Invariants I1–I10, CERN Zenodo DOI: 10.5281/zenodo.22649542)
Local-first autonomous cognitive runtime and sovereign skill registry for AI agents: bounded context admission, persistent episodic memory, multi-model neural routing, Stark-grade holographic cockpit, and fail-closed cryptographic verification.
Self-hosted AI memory for people and teams: project continuity, traceable evidence, private workspaces, and MCP integration.
GenPark AI Agent Skill - Continuous hierarchical episodic memory stream, preference decay, contextual memory anchor inspired by Meta Muse.
Local-first, zero-API memory for AI agents. Persistent, private, and yours — runs entirely on your machine, no cloud calls, no API keys.
Freeze your best AI sessions into consultable clones: a local registry with a commander.
Safe write access for AI agents. An append-only business fact store with an MCP server: every change is kept, attributed to the agent that made it, and can be undone.
Teacat-Agent is an agent harness inspectable for adaptation (not currently updated) - A very present AI assistant with customizable memory routing. Available as a standalone Telegram agent or use our template files with your OpenClaw or similar build.
Long-term memory for PocketRisu / RisuAI role-play — a local sidecar that remembers what scrolled out of context, with history and provenance. (beta)
Your AI memory bank is eating your context window. Capped, git-committed, retrieval-first project memory shared by every agent via AGENTS.md.
Vellis — open-source context graph engine for AI agents: typed, local-first memory with explicit schema, validated change, deterministic query, migration, snapshots, replay, and audit. MCP-native. Apache-2.0.
💡A vault of project policy for AI agents. You talk; the agent writes, links docs, and recalls. CLI + MCP.
boonOS Beta: contexto, memória e skills para sua IA trabalhar do seu jeito.
Private local memory for any AI agent: Markdown database, SQLite search, MCP, zero runtime dependencies.
The source code for the Oblivion package and experimentation pipeline corresponding to the "Oblivion: Self-Adaptive Agentic Memory Control through Decay-Driven Activation" paper.
MeAI Vault - Permanent Agent Memory for Hermes
Self-hosted persistent long-term memory for Claude Code and MCP clients. One memory for every session and machine: hybrid BM25 + local vector + knowledge-graph search over plain Markdown, with auto-recall hooks.
Git captures what a project is. ltm captures what it ran into.
Local-first memory, observation, and autonomy layer for Claude. A background daemon that learns from your sessions.
git for agent brains — move, debug, and secure AI agent state across 10 frameworks
Give Claude Code persistent memory across sessions. Hybrid BM25 + vector semantic search, auto-structuring via LLM, and 4-channel retrieval (MCP + hooks). Your AI assistant finally remembers.
Soul Protocol: An open standard for portable AI identity, memory, and emotion. Like HTTP, but for AI companions.
Persistent memory for AI agents. Single Rust CLI, hybrid Gemini + FTS5 + RRF retrieval. R@5 = 0.99 on LongMemEval S (beats MemPalace). Agent-native: no MCP, no server, just shell out.
Autonomous Agent - Profiling, Memory, Planning, Action, Autogen, MemGPT
Context Engineering - The art of providing all the context for the task to be plausibly solvable by the LLM.
MemorizedMCP delivers a high‑performance, local‑first MCP Server in Rust that equips AI agents with an advanced hybrid memory system. It combines a knowledge graph, vector embeddings, and full‑text search with a novel documentary memory capability, enabling accurate recall of exact documents alongside rich semantic relationships.
This repository contains comprehensive documentation for implementing and optimizing the A-MEM (Associative Memory) system utilizing both vector and graph storage for dynamic memory organization.
Give your AI coding assistant a working memory — so it remembers decisions, respects boundaries, and knows when to stop.
Long-term memory system for Cline AI assistant — persistent context, conversation history & knowledge management
Long Term Memory Chatbot A sophisticated AI chatbot with persistent personality and memory, designed to maintain consistent character traits and conversation history across sessions.
Example code for a basic Long Term Memory Chatbot using Qdrant and a conversation history list.
Persistent memory for AI agents. They learn once, remember forever.
Autonomous memory system for AI infrastructure.
A memory store for LLM agents. Hybrid BM25 + dense retrieval, cross-encoder reranking, clustered retrieval-induced forgetting (RIF), and an optional LLM enrichment layer at write time.
An open-source AI that quietly remembers who you are — in plain markdown you fully own. Works with Claude Code, Codex CLI, opencode, and any other agent that reads AGENTS.md.
Claw Memory Sharing Service — cross-instance memory sharing for AI agents via Cloudflare Workers + TiDB Cloud
Memory Layer for AI Assistants: Persistent memory, semantic retrieval, and evaluation tooling for building assistants that remember users over time.
Lightweight Rust alternative to mem0. 3.2MB binary, SQLite vector + graph memory, Ollama support.
Cross-project memory for AI coding agents. MCP server that maintains context, decisions, and knowledge across workspaces. Fully local.
A file-based protocol for humans + AI agents. Stack files like trail markers — any agent finds the trail.
The grainulation ecosystem. Structured research, decision-making, and knowledge management for AI agents.
Local-first persistent memory for AI agents — store, recall, and consolidate knowledge across sessions using FAISS, SQLite, and any LLM. MCP server for coding agents and any MCP client.
Assistant IA à mémoire persistante et croissance organique
A cognitive memory system for model continuity
A deterministic, high-performance memory engine. Built in Rust.
dk — CLI for Dakera self-hosted AI agent memory. Manage memories, namespaces, vectors, knowledge graphs, and sessions.
Stop teaching your coding agent the same lesson twice.
Python SDK for Dakera AI agent memory — self-hosted, 88.2% LoCoMo. Vectors, hybrid search, knowledge graphs, sessions.
Local-first self-learning knowledge-base harness for AI agents. Retrieve rules, avoid regressions, and promote repeated lessons into stable reusable knowledge.
Unified long-term memory and knowledge platform for AI agents with MCP, governance, and graph-oriented knowledge evolution.
Open-source AI memory engine for agents: multimodal memory, temporal knowledge graphs, graph RAG, and evidence-backed long-term context.
AccInt - local-first MCP Work Model for coding agents that learns from real outcomes.
Embedded memory and storage database for AI agents and local-first apps — NodeDB in-process, offline-first, syncs to the cloud. Supports graph, vector, array, timeseries.
Self-hosted, privacy-tiered memory for your AI agents.
A cache-optimized AI memory and context management framework for long-running LLM agents, featuring tiered context storage, token budgeting, memory compaction, sub-agent delegation, and structured web intelligence.
Persistent memory + security layer for Claude Code (MCP plugin). Agents remember your project across sessions (~87% fewer context tokens), every tool call lands in a tamper-evident HMAC audit chain, and skills are AST-scanned + HMAC-verified before they run. 100% local: PostgreSQL + Ollama, no cloud, MIT.
面向长期陪伴与自我理解的隐私优先 AI 记忆平台,融合 RAG、人生图谱、多模型路由与 MCP 能力网关。
Lunar Vault is a self-contained memory infrastructure for Codex. Features zero-dependency ONNX embeddings, advanced reranking, and secure one-click installation without leaking personal data
Memory for AI agents that learns what actually helped: lessons injected at the moment of action, credited by outcome, challenged by counter-evidence. Local-first, multi-agent, test-gated.
Agentic gateway providing persistent LLM memory via a markdown wiki. Open-source core of Kortexio.
Reliable memory and context infrastructure for AI coding agents: source-backed facts, review-gated learning, MCP/SDK/UI, and replaceable Qdrant/Graphiti retrieval.
A shared memory and collaboration tool for teams. Syncs context between Claude Code, Codex, Gemini, and any AI agent.
Local-first AI agent memory, your agent's brain is an Obsidian vault you can open, read, and edit
The self-correcting memory layer for AI agents. Zero-dependency Python memory and MCP server. Supersede, revert, or forget a value deterministically. Verifiable erasure, witness-backed tamper-evident receipts, EU AI Act ready.
AI agent primitives on Upstash Redis: memory, RAG, search, chat history, caching, and rate limiting. With Vercel AI SDK and Eve
File-based markdown notebook CLI for AI agents — the leather notebook in your jacket pocket.
Modality-routed retrieval for long-term multimodal agentic memory. Code for arXiv:2608.01543.
Light-weight, query-aware compression for high-signal LLM prompts.
A tiny, universal memory engine for humans and AI agents — CLI, HTTP API, MCP server, and shared memory hub in one small Go binary
A lightweight reranker for long-term agent memory. Cross-encoder-class quality at a fraction of the latency.
Give your agent a memory it can prove. One Python file: local SQLite vault, hybrid FTS5 + vector retrieval, Cloudflare-aware fetching, and a research loop whose [V]/[E]/[H] citations are machine-verified by CLI, not prompt-enforced.
Grounded Memory Runtime — a runtime tool that links the Agent’s memory and facts
Local-first, auditable memory for AI agents, with governed retrieval, delegated context delivery, and verifiable execution evidence. Works with OpenClaw, MCP, Pydantic AI, and LangChain/LangGraph. No external API required for local memory. Install with pip: pip install atmem
Unified fleet-wide agent memory system for DeepSeek Harness — shared Obsidian Vault for dsh, Codex, Claude Code & Hermes with dependency-free Python FTS5 core.
Cross-session memory plugin for DSH — auto-extraction, semantic recall, scheduled maintenance, LLM-powered consolidation
Local, persistent memory for AI agents (MCP server). Plain JSON on your disk; fact-level search with an optional local semantic layer; corrections replace old facts; stale-value warnings; secrets kept out; a browser page to read, edit and merge it. One command installs it in Claude Code, Claude Desktop, Codex and Cursor.
Own your AI memory, don't rent it. A native Claude Code skill AND a standalone, zero-dependency MCP server over one local SQLite file. git clone and go — full-text knowledge graph, plain-markdown, you own the data.
Persistent long-term memory for Claude Code agents — SQLite + FTS5 + sqlite-vec + RRF hybrid recall over MCP, with Chinese tokenization. Local-first, graceful degradation.
GenPark AI Agent Skill - Continual learning experience replay buffer with streaming reservoir sampling and priority weighting to balance past vs new agent trajectories.