Open Memory Protocol: One Memory Store for Claude, ChatGPT, Cursor

Open Memory Protocol (OMP) is a vendor-neutral specification and reference server that solves the AI memory silo problem: every tool (Claude, ChatGPT, Cursor, Copilot) forgets you when you switch. OMP provides a unified storage and retrieval API for user memory across tools.
Quick Start
Option A — Hosted server (zero setup): Point tools at https://omp-server-production.up.railway.app. Check health with:
curl https://omp-server-production.up.railway.app/v1/health
# {"status":"ok","version":"0.1","compliance":"OMP-Core","memories_count":0}Option B — Self-hosted: Requires Node.js 22+. Run:
npx omp-server
Or with Docker:
docker run -p 3456:3456 -v omp-data:/data ghcr.io/smjai/omp-server
Test locally:
curl http://localhost:3456/v1/health
# {"status":"ok","version":"0.1","compliance":"OMP-Core","memories_count":0}Connect Claude via MCP
Add to your Claude Desktop config file (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"omp": {
"command": "npx",
"args": ["omp-mcp"],
"env": {
"OMP_SERVER": "http://localhost:3456",
"OMP_API_KEY": "your-omp-key"
}
}
}
}To enable AI-powered memory extraction/compression, set server env vars: OMP_AI_PROVIDER=anthropic (or openai) and OMP_AI_API_KEY.
Add a system prompt to Claude Project to auto-use memory: "At the start of every conversation, use omp_recall to search for relevant memories. Whenever the user shares anything worth remembering, use omp_remember to save it silently."
Adapters for ChatGPT and Cursor
The OMP Bridge browser extension silently saves ChatGPT conversations to your OMP server every 2 minutes. When you open Claude, a toast notification appears: "C…" (indicates cross-app memory recall).
The spec includes SDKs for TypeScript and Python, plus a publicly hosted Mobile PWA at https://omp-server-production.up.railway.app/app.
Who It's For
Developers juggling multiple AI tools (Claude, ChatGPT, Cursor, Copilot) who want persistent memory across sessions without manual context transfer.
📖 Read the full source: HN AI Agents
👀 See Also

Setting Up OpenClaw as an Always-On AI Assistant
OpenClaw, configured as an always-on AI assistant for a small dev team, is set up on a Railway server with Claude as the backend and integrates with Google Workspace, GitHub, and more.

CostHawk Launches Public Leaderboard for Claude Code, Codex, and Cursor Token Consumption
CostHawk’s leaderboard ranks public users of Claude Code, OpenAI Codex, and Cursor by total token consumption, tracking counts, models, and sync timestamps without storing prompts or code.

Two New Open Source Tools for AI Agent Security and Optimization
Two open source tools are available for AI agent developers: AI Agent Defense Kit provides runtime security skills, and AgentGuard (in development) offers cost tracking, security scanning, and activity monitoring.

GPT-5.5 Codex vs Claude Opus 4.7: Real-world coding agent benchmarks
A developer pitted GPT-5.5 Codex against Claude Opus 4.7 on two real tasks: a PR triage bot and a real-time code review UI. Claude shipped cleaner with zero errors; Codex was 18% cheaper but needed a patch pass.