One memory. Every agent.
Stop re-explaining yourself to AI.
Tell one AI once. Claude, ChatGPT and Cursor all work from the same memory, and you can see everything they write.
Set up in 2 minutes, nothing to install.
Free · no credit card · 20 notes
- Claude is writing in Ventures
- ChatGPT is reading Work
- Cursor is adding to Learning
Works with Claude · ChatGPT · Cursor · any MCP client
Hosted in the EU · plain Markdown
The same question, with and without a shared memory.
Without WhyVault
When can we do the workshop with the school?
ChatGPT: Happy to help. When are you free, and is there anything else I should know about the school?
You open a new chat and explain the move, the school and the deadline again.
With WhyVault
When can we do the workshop with the school?
ChatGPT: From 9 March. You're moving on 7 March, so I'd stay clear of the week before.
You asked once, in Claude, last week.
The built-in memory in each tool only helps that tool. WhyVault is the one all of them share.
Three steps. No installation.
- 1
Connect
Copy your WhyVault address into Claude, ChatGPT or Cursor and sign in once. You choose whether each one can only read or also write.
- 2
Work
Use your tools the way you already do. When something is decided, the agent writes it into your vault.
- 3
Recall
Next time, in any tool, the agent looks it up. You don't repeat yourself.
Where it saves you the explaining.
Switching from ChatGPT to Cursor
Your project, your conventions and your last decisions are already there. No re-briefing in the new tool.
Preparing for a client
Ask any agent for the history: what was agreed, what's open, what you promised. It reads it from your notes.
Planning your Monday
“What's on my plate this week?” gets an answer that includes the launch, the move and the school workshop, not just today's chat.
On your phone
Tell the Claude or ChatGPT app something while you're out. It's in your vault when you're back at your desk.
Life admin
Taxes, appointments, the move. Write it down once and any agent can find it later.
Every change is a version you can go back to.
Whether you type or an agent writes, WhyVault saves a new version. You see who changed what, and when.
History
Client: school- Claude changed “Dates”2 min ago
Changes
## Dates
Dates by arrangement, March too.
No dates until 9 March.
- ChatGPT linked Move in March1 hr ago
- You created the noteyesterday
- Agents are marked.
- Anything an agent wrote shows its name.
- Restore this version.
- Going back creates a new version. Nothing is lost.
- Undo a whole session.
- Reset everything one connection changed since a point in time.
- No deleting unless you allow it.
- By default no agent can delete a note. If you allow it, deleted notes go to the trash.
Your notes stay yours.
- Hosted in Frankfurt.
- Database, sign-in and files run in a Frankfurt data centre. Operated by Ohrlaf GmbH.
- Plain Markdown.
- Folders, frontmatter and [[wikilinks]]. Export a zip any time, even on the free plan.
- Opens in Obsidian.
- The export is a ready-to-open Obsidian vault. If WhyVault ever stops, your notes don't.
An agent's provider (for example Anthropic or OpenAI) sees what that agent reads. Set a connection to read-only if you're unsure.
A simple structure to start from.
The template gives you five areas: Ventures, Work, Private, Learning and System. Rename them, or start with an empty vault. See the template on GitHub · How the structure works
Pricing
Start free. Switch to Pro any time in your profile.
Free
€0
- Up to 20 of your own notes (the template doesn't count)
- 1 vault, with or without the template
- All agents, full access
- Export at any time
Pro
Recommended€99/ year
equals €8.25 / month
- Unlimited notes
- Full history
- More storage for attachments
- Export at any time
All prices incl. VAT. Outside the eurozone, checkout shows the amount in your local currency.
FAQ
- Doesn't ChatGPT already remember me?
- Yes, inside ChatGPT. Claude and Cursor can't see it, and it's not something you organise or export as notes. WhyVault is one memory that all your tools share, kept as notes you can read and edit.
- What does an agent actually write into my vault?
- Short notes and additions to existing ones: decisions, context, to-dos. It appends text or replaces single sections. Everything shows up in the history with the agent's name.
- Can an agent delete or overwrite my notes?
- Not by default. Deleting only works if you allow it for that connection, and deleted notes go to the trash. Any version can be restored.
- Which tools work with it?
- Claude, ChatGPT and Cursor, plus any other agent that supports MCP. You connect them with one address and a sign-in. Nothing to install.
- Who can read my notes?
- You, and the agents you connect. An agent's provider also sees what that agent retrieves, so set a connection to read-only if you want. Your data is hosted in Frankfurt.
- What happens if I stop using WhyVault?
- Export your vault as plain Markdown and open it in Obsidian or any editor. Reading and exporting stay possible without a paid plan.
- What does it cost?
- Free for up to 20 of your own notes. Pro is €9.99 a month or €99 a year, with unlimited notes.
- Do I need to know Markdown or MCP?
- No. You use the app like a notes tool and connect your agents by copying one address.
Guides
Go deeper.
How Claude remembers, how to add a custom connector, how to connect Obsidian and how to build an LLM wiki.
Claude memory
Built-in memory, Projects or a shared vault over MCP: an honest comparison, and when each one is enough.
ReadClaude connectors
What connectors in Claude are, how to add a custom one with an MCP URL, and which ones are actually worth it.
ReadObsidian & Claude
How to connect your Obsidian vault to Claude, ChatGPT and Cursor via MCP, locally or in the cloud.
ReadLLM Wiki
Karpathy’s idea explained: an AI maintains your knowledge as a Markdown wiki. With a template to copy and a hosted setup, no scripts.
Read
Tell your AI once.
Create your vault in two minutes and connect your first agent. Free to start, no credit card.