agent-memory-system

AI memory system combining vector search with temporal knowledge graph. Built-in cognitive engine for agents. Supports memory decay, contradiction detection, and MCP integration.

🧠 The memory that dreams — self-evolving memory for DeepSeek Harness: your AI remembers across sessions, consolidates in its sleep, forgets what matters less, and grows smarter over time. Fully offline & private.

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.

Comparison of memory systems for agents

Agent-driven proactive memory CLI for AI agents — autonomously recall, maintain, and evolve persistent, source-grounded knowledge across sessions.

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.

The system of record and control plane for AI coding agents. Keeps their memory correct, current, and safe to act on as the code moves, captures what code can't say on its own, and gates what agents are allowed to do. Retrieves by path, semantic search, and relationship (code-graph).

Agentic memory infra for AI agents

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

DE-centralized - MARKdown - for US: Persistent memory for AI agents, shared knowledge for teams, and a second brain for you. Built on the Mark Protocol.

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

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

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

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.

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

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

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

Real-time distributed agent memory.

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