Best AI Agent Memory Tools in 2026

Local-first memory operating system with Markdown as the source of truth.

EverOS

E

EverOS is an open-source memory runtime that stores conversations, files, and agent trajectories as readable Markdown, then syncs SQLite and LanceDB indexes for fast retrieval and self-evolving reuse across coding assistants and workflows.

Provider-neutral memory adapter for coding agents, with local SQLite and MCP.

PAXM

P

PAXM is an open-source memory adapter that carries decisions, conventions, and working context across Codex, Claude Code, OpenCode, Pi, and MCP clients. It starts with local SQLite and can route recall and writes to multiple supported memory providers.

Long-term memory layer for agents and assistants.

Mem0

M

Best when the product needs explicit remembered facts and preference updates across conversations.

Local-first personal knowledge layer across Cursor, Claude Code, and other AI tools.

Nowledge Mem

NM

Best when you work across multiple AI clients and need one durable memory graph for decisions, threads, and working context—not a memory SDK to embed in your own product.

Dialectical user modeling memory for stateful agents.

Honcho

H

A memory library for building stateful agents with dialectical user modeling and peer cards.

Filesystem-paradigm context database for agent memory.

OpenViking

O

An open-source context database that manages agent memory, resources, and skills through a filesystem paradigm.

Knowledge graph memory with cross-memory synthesis.

Hindsight

H

Agent memory that learns through knowledge graphs and cross-memory synthesis.

Holographic Reduced Representations memory with trust scoring.

Holographic

H

A compositional memory system using HRR algebra with local SQLite storage, FTS5 search, and trust scoring.

Hybrid retrieval memory with quality filtering for coding agents.

RetainDB

R

RetainDB is an open-source AI memory database for coding agents — local-first hybrid vector + BM25 retrieval with built-in noise filtering for cross-session context.

Portable memory with pre-compressed extraction and knowledge trees.

ByteRover

B

A portable memory layer for autonomous coding agents with pre-compressed extraction and knowledge tree organization.

Temporal knowledge graph memory for agents.

Zep

Z

Best when memory needs relationships, temporal context, and retrieval over evolving user or organization knowledge.

Data and retrieval framework for LLM and agent apps.

LlamaIndex

L

Best when the agent's value depends on ingestion, indexing, retrieval, and structured access to private knowledge.

Stateful graph orchestration for controllable agents.

LangGraph

L

Best when agent behavior must be represented as explicit nodes, edges, state, and recovery paths.

Stateful agents with long-term memory and context engineering.

Letta

L

Best when agents need persistent identity, multi-session memory, and explicit control over what gets stored, updated, and recalled over time.