Draw Neo4j graph models from an AI assistant

toro4n is a Model Context Protocol (MCP) server. Connect an assistant you already use, and it can sketch a graph model from a description, a document, a spreadsheet, or a database schema, lay it out in your toro4n account, and edit your models while you watch.

The address

https://toro4n.com/mcp

You need a toro4n account, because the assistant works only with models saved online, never with the models in your browser. Signing in is a code sent to your email.

Connecting

Claude (claude.ai and the desktop and mobile apps): Settings → Connectors → Add custom connector. Paste the address, add the connector, choose Connect, then turn it on in a chat with the + button.
ChatGPT: in Settings, under Apps, turn on developer mode (custom connectors depend on your plan), create an app with the address, and choose OAuth for sign-in.
Claude Code: claude mcp add --transport http toro4n https://toro4n.com/mcp
Gemini CLI: gemini mcp add --transport http toro4n https://toro4n.com/mcp
VS Code with Copilot, Cursor, and other MCP clients: add a remote MCP server over HTTP with the address.

The first time, the assistant sends you to toro4n to sign in and choose Allow. Only that click, made in toro4n by a signed-in user, gives the assistant access; toro4n never forwards you anywhere on its own.

What to ask

"Sketch a graph model of our order system from this document." "Draw the schema in this spreadsheet as a Neo4j model." "Add invoices to my web shop model." The assistant replies with a link that opens the model, laid out automatically, with the labels, types, and keys written the Neo4j way.

The six tools

list_models – your online models and the ones shared with you, with your role and folder.
get_model – a model's nodes, relationships, notes, and layers.
create_model – a new model from nodes, relationships, and notes, laid out automatically.
edit_model – a list of changes, all or nothing: add, change, and remove nodes, relationships, properties, and notes, and remove and rename layers.
list_versions and restore_version – the model's version history.

Everything the assistant sends goes through the same rules as everything else in toro4n: one node per label combination, Neo4j casing, property types from the list. A broken rule is answered with a message the assistant can act on. The assistant has the same permissions you have, and it can't delete or share whole models.

Undo and control

Before every change an assistant makes, the previous version of the model is saved in Version history, and the change is listed there as your email address "via Claude" (or whichever assistant). You see the changes as they happen. What the assistant reads is handled by the company that provides it, under its own terms; toro4n's server never contacts an AI provider. Under Account → AI assistants you see what's connected and can disconnect any of them.

Questions

Which AI assistants work with toro4n?

Any assistant that supports the Model Context Protocol: Claude (claude.ai and its apps), ChatGPT with custom connectors, Claude Code, Gemini CLI, VS Code with Copilot, Cursor, and others. The address to connect is https://toro4n.com/mcp.

Can the assistant delete my models?

No. It can list, read, create, and edit models and restore versions, with the permissions you have. It can't delete or share a model. Every change it makes can be undone in Version history.