explaingit

smjai/open-memory-protocol

Analysis updated 2026-05-18

★ 66TypeScriptAudience · developerComplexity · 3/5LicenseSetup · easy

TLDR

An open protocol and reference server for portable AI memory: any AI tool that implements OMP can share memories, preferences, and conversation context with any other OMP-compatible tool via HTTP.

Mindmap

mindmap
  root((Open Memory Protocol))
    What It Solves
      AI memory silos
      Cross-tool context loss
      Preference repetition
    Components
      Spec document
      Reference server
      TypeScript and Python SDKs
      MCP adapter for Claude
    Memory Types
      Episodic events
      Semantic facts
      Procedural how-to
    Deployment
      Hosted demo server
      Self-hosted via Docker
      SQLite or PostgreSQL
    Cross-Tool Features
      OMP Bridge extension
      Handoff briefs
      Semantic search
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

What do people build with it?

USE CASE 1

Connect Claude Desktop via MCP to an OMP server so Claude automatically saves and recalls memories across all sessions

USE CASE 2

Build a cross-tool memory layer where preferences set in ChatGPT are instantly available when switching to Claude or Cursor

USE CASE 3

Use the OMP Bridge browser extension to save a ChatGPT conversation and inject a handoff summary into a new Claude session

USE CASE 4

Implement OMP memory in a custom AI agent using the TypeScript or Python SDK and the REST API

What is it built with?

TypeScriptNode.jsSQLitePostgreSQLDockerMCP

How does it compare?

smjai/open-memory-protocolallstarswc/allstarsevanbacon/grok-voice-demo
Stars666060
LanguageTypeScriptTypeScriptTypeScript
Setup difficultyeasyhardmoderate
Complexity3/54/53/5
Audiencedevelopergeneralvibe coder

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · easy Time to first run · 5min

For self-hosting, requires Node.js 22 or Docker, Claude Desktop integration requires editing the MCP config file.

Apache 2.0: use freely for any purpose, including commercial use, keep the copyright and license notices when redistributing.

In plain English

Open Memory Protocol (OMP) is a specification and reference implementation for sharing memory between AI tools. The core problem it addresses is that every AI assistant (Claude, ChatGPT, Cursor, Copilot, and others) stores its knowledge of you in its own isolated database, so switching from one tool to another means starting over from scratch. OMP proposes a common format and HTTP API so that any AI tool that supports the protocol can read and write memories to a central server you control. The project has four parts: a specification document describing the memory object schema and REST API, a reference server (self-hostable in one command via npm or Docker), TypeScript and Python SDKs for building OMP-compatible tools, and adapters for specific AI environments including Claude via MCP and OpenAI. A publicly hosted demo server is also available so you can try the protocol without installing anything. Memories are structured objects with a content string, a type (episodic for events, semantic for facts and preferences, procedural for how-to knowledge), a source, tags, timestamps, and an optional vector embedding for semantic search. Any tool can write a memory via a simple HTTP POST, and any other OMP-compatible tool can query it. A browser extension called OMP Bridge automates cross-tool handoffs: it monitors an active chat session, saves the conversation every two minutes, and when you open a different AI tool it offers to inject a plain-English summary so the new AI can pick up where the other left off. Connecting Claude Desktop requires adding the OMP MCP server to Claude's config file and optionally setting a system prompt that instructs Claude to automatically save and recall memories without being prompted. The protocol stores data in SQLite or PostgreSQL with vector extensions. It is released under the Apache 2.0 license.

Copy-paste prompts

Prompt 1
Set up the OMP MCP server for Claude Desktop on macOS. Show me the claude_desktop_config.json entry and how to run the server locally with npx.
Prompt 2
Write an OMP-compatible system prompt for Claude that automatically saves user preferences and recalls relevant memories at the start of each conversation.
Prompt 3
I want to implement OMP in a custom AI agent. Show me how to write a semantic memory via the REST API and then search for it by query string.
Prompt 4
How does OMP distinguish between episodic, semantic, and procedural memory types? Give me an example JSON memory object for each type.
Prompt 5
How do I export all my memories from the hosted OMP demo server and import them into a self-hosted instance? Which API endpoints handle export and import?

Frequently asked questions

What is open-memory-protocol?

An open protocol and reference server for portable AI memory: any AI tool that implements OMP can share memories, preferences, and conversation context with any other OMP-compatible tool via HTTP.

What language is open-memory-protocol written in?

Mainly TypeScript. The stack also includes TypeScript, Node.js, SQLite.

What license does open-memory-protocol use?

Apache 2.0: use freely for any purpose, including commercial use, keep the copyright and license notices when redistributing.

How hard is open-memory-protocol to set up?

Setup difficulty is rated easy, with roughly 5min to a first successful run.

Who is open-memory-protocol for?

Mainly developer.

Open on GitHub → Explain another repo

This repo across BitVibe Labs

Scan in gitsafehub Deploy in gitdeployhub smjai on gitmyhub

Verify against the repo before relying on details.

Alternatives & similar repos