I built a shared memory for all my AI agents because I was tired of re-explaining myself every day

I'm running a startup, a few side projects, working a day job and have 2 small kids. Every area has its own decisions and open threads, and every AI tool has its own closed memory. Claude doesn't know what I told ChatGPT yesterday. My feed is full of "how to set up your agent company" posts. What almost none of them cover: where do those agents get their knowledge from? Well many do, but i want a connected version, easy to setup and available on multiple devices. Without a shared knowledge base, every agent starts from zero. So I built WhyVault : • Hosted Markdown vault with a ready-made structure for different areas of life • Built-in MCP server: connect Claude, ChatGPT or Cursor via OAuth, no local setup • Agents read context before answering and write decisions back • Every agent change is versioned and reversible; agents can't delete anything • Plain Markdown, Obsidian-compatible, export anytime, hosted in Germany It helps me organize a bunch of projects and connect my knowledge quite quickly. This enables building many tools in parallel and jugeling multiple projects :) Would love to know: how do you currently keep context in sync across multiple AI tools? Or do you not bother? Link in the comments. Test it for free.