Vanilla Brain

One shared, persistent memory for all of your AI agents. What one agent learns — your work, your own knowledge, the shape of a recurring task like particular paperwork or particular invoices — every agent remembers, across sessions and across machines, so your AI does not make the same mistakes over again.

How it works

Your agents connect to Vanilla Brain over MCP (the Model Context Protocol). Each agent calls brain_wake at the start of a session, pulls relevant context with brain_recall, and saves durable facts with brain_remember. Everything is one account — connect Claude Code on one machine and Codex on another, and they share the same living memory.

Set up an agent

Create an approved preview account and sign in.
Open the Agents tab and click Download setup file. It is a small JSON file with a one-time code that expires in 15 minutes.
Give the file to your AI (attach it, or paste its contents) and say “set this up.” The AI redeems the one-time code for its own least-privilege connector and capture-only hook credential, installs a short memory protocol, and verifies the connection — no credential is ever printed.
Ask the AI to confirm which calls succeeded. From then on it remembers, across sessions and machines.

Each machine gets one connector entry; its capture-only credential is managed with it. The memory itself is shared automatically across everything you connect. Revoking the connector from the Agents tab revokes both.

Create your account