Connect your AI client

This guide connects your AI client to Spreeflo's hosted MCP server. It takes two things: a Spreeflo API key, and one entry in your client's MCP configuration.

Step 1: Create an API key

In the Spreeflo app, navigate to Settings > API Keys and click + Add API Key. Give it a name you'll recognise later — for example the name of the client you're connecting — and copy the key.

This is the same key used for the Spreeflo REST API. There is no separate MCP credential and no OAuth flow to complete.

Warning: This key has write access to your workspace. Treat it like a password — give each client its own key, keep it out of screenshots and shared documents, and revoke it if it's ever exposed.

Step 2: Add the server to your client

The server endpoint is:

https://mcp.spreeflo.com/mcp

Claude Code

Run this in your terminal:

claude mcp add --transport http spreeflo https://mcp.spreeflo.com/mcp \
  --header "Authorization: Bearer <YOUR_API_KEY>"

Cursor and other mcp.json clients

Add a spreeflo entry to your client's MCP configuration file:

{
  "mcpServers": {
    "spreeflo": {
      "url": "https://mcp.spreeflo.com/mcp",
      "headers": { "Authorization": "Bearer <YOUR_API_KEY>" }
    }
  }
}

Claude, Cowork, and other connector clients

Some clients add a remote MCP server through a form that asks only for a URL — claude.ai, Claude Cowork, and ChatGPT custom connectors are the common case. Put the key in the URL instead of a header:

https://mcp.spreeflo.com/mcp/<YOUR_API_KEY>

Same key, same workspace, same tools — the only difference is where the credential travels. Where a client can send a header, prefer the header form above.

Claude's custom connector dialog has a Request headers section where you can store an Authorization value of Bearer <key> — that's the better option if you have it, since the key never enters a URL. It's currently in beta and rolling out gradually, so use the URL form above if you don't see the section. Note that Claude Desktop's claude_desktop_config.json can't send headers to a remote server, so add it as a connector there rather than editing that file.

Step 3: Verify the connection

Ask your client to list the Spreeflo tools it can see, or try a simple read such as "list my Spreeflo audiences".

A working connection returns your actual workspace data — your audience names, your campaigns. If your client shows a tool list, you should see Spreeflo tools such as list_campaigns, get_segment_preview, and create_email_template. If you get an authentication error instead, see Troubleshooting below.

What the connection can reach

Check your client's confirmation settings

Before you start working, check how your client handles tools that modify data. Spreeflo annotates every tool so a well-behaved client can tell reads from writes, but Spreeflo will never prompt you — the decision to ask you before running a write belongs entirely to your client.

Warning: Writes apply immediately to your live workspace. There is no staging step, no pending list, and no approval queue on this surface. If you want to review changes before they land, use Spreeflo's in-app AI instead.

The riskiest actions are removed rather than left to your client's settings: deletes, starting or resuming a campaign or journey, and publishing a form are refused by the server.

Troubleshooting

Revoking access

To cut a connection off, delete its API key in Settings > API Keys. Once revoked, a key can't be reactivated — generate a new one and update the client's configuration.

Need Immediate Help? Contact our support team at support@spreeflo.com or check our other guides for more information.