Let the AI install it for you.
You don't need to know what a terminal is. The trick: you paste one prompt into Claude or ChatGPT, and it runs the setup with you — one small step at a time, checking each step worked before moving on. You just copy, paste, and answer when it asks.
Before you start
Open the email we sent you. It has three things you'll paste in later — keep that tab open:
A line that starts with msk_. It's yours alone — your own memory. Don't share it.
The address of our gateway, e.g. https://vimemory.xyz. This is where your memory lives.
The open code. You don't read it to use it — your AI assistant does that for you.
Grab your key first — it takes one email and a confirm click. Get your key →
Pick your helper
Either one works. This is the AI that will do the install with you by reading the setup instructions and giving you exact things to copy. Pick whichever you already have open.
Claude
Claude Desktop or claude.ai. Best if you're going to use ViMemory inside Claude Desktop anyway — it can see its own config.
ChatGPT (OpenAI)
chatgpt.com. Works just as well — it'll hand you the same copy-paste steps for your computer.
Paste this prompt
Copy the whole block below into a fresh Claude or ChatGPT chat. Replace the red bits with the key and URL from your email. Then send it — and follow along. The AI will go one step at a time and wait for you.
I'm not a developer. Help me install the "ViMemory" memory tool into Claude Desktop, one small step at a time. Wait for me to confirm each step worked before giving me the next one. Explain every command in one plain sentence before I run it, and tell me exactly where to type it (Terminal on Mac, or Command Prompt on Windows). Here are my details: - My computer is: Mac OR Windows - My ViMemory secret key: msk_paste_your_key_here - My ViMemory endpoint URL: https://vimemory.xyz The tool is a small "MCP server" published here: https://github.com/arya-wtf/vimemory-localmcp What I want at the end: 1. The code from that GitHub repo downloaded onto my computer. 2. Python installed if I don't already have it. 3. The "vimemory" connector added to my Claude Desktop config file, using my key and URL above. 4. Claude Desktop restarted so the four memory tools (save_memory, search_memory, list_memory, delete_memory) show up. Start by checking what I already have installed. Go.
Using ChatGPT instead? Same prompt — just change the first line to say "Help me install it and tell me exactly what to type on my computer," since ChatGPT can't see your files directly. It'll still give you every copy-paste step.
What it'll have you do
So nothing feels like a surprise — here's the shape of what the assistant guides you through. You don't memorise this; it's just a map.
One command copies the small program from GitHub onto your machine. The AI gives you the exact line to paste.
If Python isn't there, the AI walks you through getting it, then installs the connector's requirements. One or two commands.
The AI opens your Claude Desktop config and pastes in a small block with your key and URL. This is what links you to your own memory.
Quit it fully and reopen. You'll see vimemory in the tools (🔌) list. Done.
Try it — just talk
Once it's installed, you don't run anything. You just tell Claude in plain words. Try these:
The whole point: hit a limit in Claude, switch to another model pointed at the same key, and your memory is already there. Ask it to recall — it picks up where you stopped.
Your key holds a small number of memories so you can feel it without setup. When it fills, delete one to make room — or, when you're ready for unlimited memory for your whole team, run your own server (next section). That wall is on purpose.
Want your own server?
The trial runs on our gateway. When you want your own — unlimited memory, your data on your box, keys for your whole team — the full server is open too. Same trick: let the AI drive.
I want to run the "ViMemory" server on my own VPS (a small Ubuntu cloud server). I'm not a developer — walk me through it one step at a time and wait for me to confirm each step. Explain each command in one plain sentence first.
The server code and a step-by-step VPS guide are here:
https://github.com/arya-wtf/vimemory
Help me: rent or use an Ubuntu server, install Docker, clone that repo, start it with docker compose, point my domain at it, and create my first user key with the manage.py adduser command. At the end I should be able to open https://my-domain/health and see {"ok":true}.
Start by asking me whether I already have a server and a domain. Go.
That's the same open code we run. Nothing hidden — read it here.
If something's off
Paste any error straight back to your AI assistant — that's the fastest fix. Common ones:
msk_ key and URL are both filled in./health on the end — it should say {"ok":true}. If not, reply to your welcome email and we'll check our gateway.You're in. When the trial fills, run your own.
That's the whole tool — installed without writing a line of code. When your trial key fills up, or you want unlimited memory for your whole team, the full server is open and yours to run. The self-host prompt above gets you there the same way: let the AI drive.
Back to ViMemory →