Alterable

Model Context Protocol

The Alterable MCP server

Build your email content from a conversation. Connect Claude, ChatGPT, Cursor or any MCP client to your Alterable account, ask for a countdown timer that ends Friday at midnight in Paris, and get back the embed code.

Connect a clientRead the API reference

Streamable HTTP  ·  API token or OAuth  ·  Same rules as the dashboard

alterable · connected

Make a countdown for the Black Friday sale. Ends Nov 28 at midnight, Paris time. Dark, our green, square digits. Link to /sale.

create-countdown-timer

get-embed-code

Done. Black Friday 2026 counts down to 2026-11-28 23:59:59 Europe/Paris. Paste this into your template:

<a href="https://next.alterable.com/countdown-timers/click/9f3e2b1c-4c2a-4e7b-9d1f-2a6b8c0d4e5f">
  <img src="https://next.alterable.com/countdown-timers/9f3e2b1c-4c2a-4e7b-9d1f-2a6b8c0d4e5f.png?d=%TODAY%&uid=%SUBSCRIBERID%" />
</a>

31 tools

covering every service and the library

2 ways in

API token or OAuth, depending on the client

0 new rules

it calls the same API the dashboard does

What it does

What the Alterable MCP server does

An MCP server for email marketing content: anything you can do in the Alterable app, an AI client can do for you. It reads the reference material first, so a request like "show umbrellas when it rains" becomes a dynamic image with the right targeting rules, not a guess.

Every service, as a tool

Create, update, list and delete countdown timers, dynamic images, dynamic products, local maps, scratchers and surveys. Upload images, manage your product catalog and merge tags. Fetch the embed code for anything you make.

Analytics in the same conversation

Ask how a timer performed last month, pull survey results, or list who scratched a card. Impressions, votes and plays come back as data the assistant can summarise or chart.

Reference material built in

The server ships resources the assistant reads before it acts: a services overview, the full targeting-rules grammar, allowed values and the OpenAPI spec. It does not have to guess what a weather rule looks like.

Your rules, not a side door

Every tool call goes through the same REST API as the dashboard. Validation, plan limits and team scoping apply exactly as they would if you clicked the button yourself.

Connect

Connect the Alterable MCP server to Claude, ChatGPT or Cursor

The server is at https://next.alterable.com/mcp. Desktop tools take an API token from Settings → API; hosted assistants sign in with OAuth.

Claude Code

One command in the terminal, with an API token.

claude mcp add --transport http alterable https://next.alterable.com/mcp \
  --header "Authorization: Bearer <your-token>"

Claude.ai and Claude Desktop

Add a custom connector with the URL below and no token. You sign in to Alterable and approve the connection.

https://next.alterable.com/mcp

ChatGPT

Same URL, added as a connector in ChatGPT’s settings. OAuth handles the rest.

https://next.alterable.com/mcp

Cursor, Windsurf, VS Code

Any client that speaks Streamable HTTP works. Paste the URL, and either supply a token as a Bearer header or let the OAuth flow sign you in.

{
  "mcpServers": {
    "alterable": {
      "url": "https://next.alterable.com/mcp",
      "headers": { "Authorization": "Bearer <your-token>" }
    }
  }
}

Tools and prompts

What the assistant can call

Typed tools for each service, so the assistant fills in a deadline or a prize pool rather than composing raw JSON. Three prompts package the most common jobs into a single request.

build-countdown-campaign
Give it a deadline, a timezone and a few brand hints. It creates the timer and hands back the embed snippet.
personalize-with-dynamic-image
Describe the goal ("umbrellas when it rains, sunglasses otherwise"). It reads the targeting rules, proposes variants, and builds the dynamic image with a default.
review-campaign-performance
Point it at a service and a look-back window. It pulls the numbers and writes them up.

Plus four resources the assistant reads for context: the services overview, targeting rules, allowed values, and the OpenAPI spec behind the API reference.

Find and inspect

  • list-teams
  • list-services
  • get-service
  • get-embed-code

Countdown timers

  • create-countdown-timer
  • update-countdown-timer

Dynamic images

  • create-dynamic-image
  • update-dynamic-image

Dynamic products

  • create-dynamic-product
  • update-dynamic-product

Local maps

  • create-local-map
  • update-local-map
  • list-map-locations
  • add-map-locations
  • update-map-location
  • delete-map-location

Scratchers

  • create-scratcher
  • update-scratcher
  • list-scratcher-plays

Surveys

  • create-survey
  • update-survey
  • get-survey-results
  • list-survey-votes

Library

  • upload-image
  • create-product
  • update-product
  • create-merge-tag
  • update-merge-tag

Analytics and housekeeping

  • get-impressions
  • delete-service
  • restore-service

Frequently asked questions

What is an MCP server?

The Model Context Protocol (MCP) is an open standard that lets AI assistants call tools on external services. A server describes what it can do, the assistant decides when to call it, and the results flow back into the conversation. Alterable’s server lives at https://next.alterable.com/mcp and uses the Streamable HTTP transport.

Do I need an API token?

It depends on the client. Claude Code, Cursor and similar tools take an API token from Settings → API as a Bearer header. Claude.ai custom connectors and ChatGPT use OAuth instead: add the URL with no token, and you will be asked to sign in to Alterable and approve the connection.

Can I limit what an assistant is allowed to do?

Yes, and it works the same way for both ways in. API tokens are scoped per resource and action when you create them, and an OAuth connection shows the same grid on the consent screen, so you can allow an assistant to read countdown timers but not write surveys, for example. Either can be revoked at any time from Settings → API.

Which plans include it?

The MCP server uses the same access as the REST API, which is included on the Pro and Scale plans.

Is it safe to let an assistant delete things?

Deleting a service is reversible: there is a restore-service tool and the server instructs assistants to confirm with you first. Deleting images, products and merge tags is permanent, and the server says so to the assistant before it acts.

Does it count against my API rate limit?

Yes. Tool calls are dispatched to the REST API internally, so the usual 60 requests per minute apply. A single tool call can cost two or three requests (a write followed by a re-read), which the server tells assistants so they pace bulk work.

Still have questions? Contact us →

Get started

Give your assistant an Alterable account

Create a free account, generate an API token under Settings → API, and add the server to the client you already use.

Create a free accountAPI reference

Free to start · No credit card · Works with every major ESP