agent-memory-system

Cross-platform persistent memory MCP for Codex, Gemini CLI, Claude Code, and other local MCP hosts. 36 cited neuroscience mechanisms, local-first SQLite/PostgreSQL, hybrid retrieval, decay-based consolidation, and reproducible benchmarks. Claude adds optional automatic lifecycle hooks.

bhived is an MCP server that gives AI agents shared memory, skills, and tool discovery. install once, works in Claude Code, Cursor, and 15+ other agents.

Comparison of memory systems for agents

De-centralized - markdown - for us: an information web adjacent markdown protocol for knowledge sharing between humans and agents, humans to humans, agents to humans, and agents to agents

Neuroscience-inspired memory for AI agents — sleep cycles that rewrite and strengthen knowledge over time. TypeScript, PostgreSQL, MCP.

Plasmod: An Agent-Native Database for Cognitive Objects, Event-Driven Materialization, and Structured Evidence Retrieval

Context and memory management for agents to learn over time. Private and lightweight.

grep-like session search tool for coding agents with Git-backed sharing (Codex, Claude Code)

Open-source AI operating system for computer use agents

A memory graph designed for the agent that uses it, not the human who feeds it. Engrama reconstructs context from associations on demand, replacing the "stuff everything into the prompt" reflex with targeted graph traversal. SQLite default, Neo4j optional.

Reverie Core is an agentic cognition layer for the Hermes ecosystem (memory plugin)

Rather than being the smartest assistant, it aims to be the one that best understands emotional value.

Auto consolidate all your Agent Runtimes. Share memory across your agents.Auto highly structured task analyzer.

A Rust memory system for autonomous AI agents built on three operations from cognitive neuroscience that no existing system implements together: memory compaction, memory evolution, and memory conflict resolution.

Real-time distributed agent memory.

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。

HangarX — Turn your Obsidian vault into a queryable knowledge graph that Claude, Cursor, Cline, and any MCP-compatible agent ▎ can read and write to.

Governed graph-native agent memory: knowledge extraction, fusion, hybrid RAG, scoped access tokens. A push for data sovereignty.

从零搭建 Hermes Agent + Honcho 记忆管理层的完整中文指南