Connect Claude and other AI assistants

GalleryCamp runs a Model Context Protocol server. MCP is the open standard AI assistants use to work inside other software, so an assistant connected to GalleryCamp is not guessing from what you paste into the chat. It reads your actual records and, if you allow it, changes them.

That makes it a capable extra pair of hands. Ask Claude to find every available work by an artist under a price, log the inquiry that just came in by phone, build a viewing room from a brief, draft the invoice once a collector says yes, or tell you what you owe each artist this month. It does the lookups and the data entry; you decide what goes out.

The server's address is:

https://gallerycamp.com/mcp

You add that address to your assistant, then sign in to GalleryCamp and approve it. There's no key to copy. Settings → AI Assistants in the dashboard has the same steps with copy buttons, and lists every assistant connected to the gallery.

Only a gallery's owners and admins can connect an assistant to it.

In this section

This page — connecting Claude, ChatGPT, Claude Code, Cursor and other tools.
What you can do with an AI assistant — everyday gallery work to hand over, with things to ask.
Every tool an assistant can use — the complete list, and which ones need Read & write.
Access, safety and troubleshooting — what an assistant can reach, what it can't, and what to do when something goes wrong.

Choose what it can do

When you approve the connection you choose Read only or Read & write.

Read only can see the gallery's records and reports, and change nothing. The tools that make changes are not offered to it at all.
Read & write can also create and change records — artworks, artists, collectors, inquiries, deals, orders, viewing rooms, events and blog posts.

Neither can manage your team, billing or gallery settings, and neither can delete inventory: works are archived, never deleted. Whatever it changes is credited to whoever approved the connection, and shows in change history as theirs via the app.

A few tools reach people outside the gallery — sending an offer, an invoice or a viewing-room invitation by email, or publishing your website. Assistants are told to check with you before using them that way, but treat a Read & write connection with the same care as a colleague's login. Access, safety and troubleshooting covers this in full.

If you're not sure, start with Read only. It's the quickest way to see what an assistant is good at, and connecting again with Read & write replaces the first connection.

Claude

Works on every Claude plan, on the web, desktop and mobile apps. The Free plan allows one custom connector.

Open Customize → Connectors and click Add custom connector.
Name it GalleryCamp and enter the address above. Leave the sign-in settings as they are; no client ID or secret is needed.
Click Add, then Connect. Claude sends you to GalleryCamp.
Choose the gallery and the access level, then Allow.

On a Team or Enterprise plan, an Owner adds it once under Organization settings → Connectors (Add → Custom), and each member then clicks Connect on it.

ChatGPT

In ChatGPT on the web, open Plugins, click + and choose Add custom MCP server.
Name it GalleryCamp, enter the address above under Connection, and choose OAuth for authentication.
Confirm the warning and select Create as a plugin.
Sign in to GalleryCamp when asked, choose the gallery and the access level, then Allow.

Some workspaces still show this as Settings → Apps → Create with developer mode turned on. Custom MCP servers depend on your ChatGPT plan, and on Business and Enterprise an admin may need to allow them first.

Claude Code

claude mcp add --transport http gallerycamp https://gallerycamp.com/mcp

Then run /mcp in Claude Code, choose gallerycamp and select Authenticate. Your browser opens GalleryCamp to approve it.

Cursor, VS Code and other tools

Most tools that support MCP connect with the address alone and open a browser to sign in the first time it's used. In Cursor, add this to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "gallerycamp": { "url": "https://gallerycamp.com/mcp" }
  }
}

Your first conversation

Once it's connected, open a new chat and ask:

Which gallery are you connected to, and what can you do there?

The assistant reads the gallery's name, whether the connection is Read only or Read & write, your plan, and the headline numbers: inventory by status and value, open inquiries, recent orders and upcoming events. That confirms it's working and that it's pointed at the right gallery.

Then give it something real. A few good first asks:

What's come in that nobody has answered yet?
What do we have by Mara Voss that's still available, and where is each piece?
How much did we sell in the last 90 days, and which artists led?

Talk to it as you would brief a colleague who knows GalleryCamp but not your gallery. Use names, not ids: it looks up artworks, artists and collectors for you. What you can do with an AI assistant has many more examples.

See and disconnect assistants

Every connected assistant is listed under Settings → AI Assistants, named after the app and the person who approved it. Disconnect one and it stops working immediately. Connecting the same app again replaces its earlier connection rather than adding a second one.

A connection lasts as long as the person who approved it stays an owner or admin of the gallery. Removing them from the team, suspending them or making them a member disconnects every assistant they connected.

Demo galleries can't be connected: they hold sample data.

For scripts and tools that can't open a browser, create a key under Settings → Access Keys — choose Read only unless it needs to make changes — and send it to the address above in an x-api-key header. A key has no person behind it, so it can't write notes or blog posts.