atomicstrata/atomicmemory

Portable semantic memory for AI agents: core engine, TypeScript SDK, framework adapters, MCP server, CLI, and host plugins.

What it solves

AtomicMemory provides durable, long-term memory for AI agents and applications, allowing them to maintain context across different sessions. It solves the "black box" problem of AI memory by being inspectable and correction-aware, meaning it can update, clarify, or delete memories as facts change, rather than just appending new information.

How it works

It operates as a memory backend that can be deployed in three ways: as a managed Hosted Cloud service, an open-source Core locally via Docker, or integrated via a TypeScript SDK. It uses a protocol that allows it to connect to various agent hosts (like Cursor or Claude Code) through an MCP server, and provides adapters for popular AI frameworks such as LangChain and Vercel AI.

Who it’s for

Developers building AI agents and LLM-powered applications who need a persistent, model-agnostic memory layer that they can audit and manage.

Highlights

  • Correction-Aware: Can supersede or delete memories instead of treating memory as append-only.
  • Inspectable: The open-source Core allows users to audit the mutation and retrieval paths.
  • Portable: Works across CLI, MCP server, SDK, and various framework adapters.
  • Model-Flexible: Keeps extraction, embeddings, and retrieval behind explicit provider boundaries.
  • Deployment Flexibility: Supports both managed cloud and local self-hosting.

Related

  • Project
    TencentCloud/TencentDB-Agent-Memory

    A centralized memory server and management hub that allows AI agents to share and reuse distilled experience, skills, and project knowledge across sessions and frameworks.

  • Project
    rohitg00/agentmemory

    A persistent memory server for AI coding agents that automatically captures and retrieves project context across sessions using hybrid search.

  • Project
    MemoriLabs/Memori

    A persistent, structured memory layer for AI agents that captures conversations and and agent execution traces to eliminate session amnesia and reduce token costs.

  • Project
    rtk-ai/icm

    ICM (Infinite Context Memory) is a Rust‑based, zero‑dependency binary that provides a shared, searchable SQLite memory store for AI agents. It supports episodic memories with importance‑based decay, permanent knowledge‑graph memoirs, feedback recording, and verbatim session transcripts. Integrated via MCP hooks or a CLI, it works with many popular assistants (Claude, Gemini, Codex, Copilot, etc.) and offers optional ONNX‑Runtime semantic search. Install via Homebrew, a one‑liner script, or Cargo; configure globally or per‑project, then store/recall context across tools without re‑explaining. Experimental (pre‑1.0) but usable today.