When ChatGPT introduced memory, it became a really really sticky product. \nBut, most of us use multiple different models - chatgpt, claude, gemini, and coding agents like cursor, windsurf, and what not. Why not make the memories "universal" across all these apps?
To solve this, I made supermemory MCP and launched it on X last month. Instantly, it got half a million impressions, and a lot of good feedback. Slowly through the month, it has grown to 60k+ daily users, and I decided to open-source it!
I also wrote a very detailed blog about the UX and technical details of this MCP in this blog - https://supermemory.ai/blog/the-ux-and-technicalities-of-awe...
The MCP itself is a super simple one, with two tools (addToMemory and search) and a prompt. Surprisingly this works very well :)\nI also made another open-source CLI called `install-mcp`, to make it easier for users to install supermemory MCP.
On top of that, we made it such that there's no authentication whatsoever - the URL _is_ the key. The URL is a nanoid which would take billions of years to brute-force through, and as long as the user doesn't share the URL, their memories are completely safe.
The MCP itself stores no user data whatsoever - there's no signin, paywall, or any of that. It's a simple, one click way to make your information universal!
Would love some feedback and issues, PRs on the open source repositories and thoughts in general.","title":"Show HN: Supermemory-mcp \u2013 Universal memories through different LLM apps","type":"story","url":"https://github.com/supermemoryai/supermemory-mcp"}