Overview of Iterable's MCP Server

This article provides an overview of MCP servers and information that will help you use Iterable's MCP Server. You can find additional details in Setting up Iterable's MCP Server.

# In this article

# What is MCP?

Model Context Protocol (MCP) is an open standard that helps AI assistants (like Cursor or Claude Desktop) connect safely to external tools and their APIs. Think of MCP as a "bridge" that lets your AI app talk to platforms like Iterable.

You control the data and actions that are available to the AI agent by configuring the MCP server, and then the server enforces the rules that you set.

While Iterable's MCP Server is intended for technical users who understand how to configure and test connectors safely, you don’t need to know how to code or use APIs to use an MCP server. Once you've configured it, you can ask questions or perform actions in plain English, and the MCP server translates them for you into safe API calls.

# What's special about Iterable's MCP Server?

Iterable's MCP Server enables your AI assistant (LLM) to perform specific actions or queries directly in Iterable. The LLM can access nearly all of Iterable's existing API endpoints, so whatever you can do with our API, you can do by talking to your AI assistant, once Iterable's MCP Server is enabled.

# Things you should consider

Keep these things in mind when using Iterable's MCP Server.

• Make sure that you're using AI tools that are approved by your organization and that you're accessing Iterable data in a way that aligns with your organization's policies. It's up to you to monitor for risks that are associated with the integrity and security of your Iterable environment.

• API endpoints called by Iterable's MCP Server use the same rate limits as all other API consumers. When using endpoints that are rate-limited either per project or per organization, usage is aggregated with other API consumers. This means that using your MCP server may compete for API resources. When API requests exceed rate limits, the MCP server receives a 429 response and does not automatically retry.

• Project data related to the queries you make doesn't persist on Iterable's MCP Server; responses are available only for your current AI session.

# Best practices

In addition to general best practices for using AI assistants, keep these in mind when using Iterable's MCP Server:

• Iterable's MCP Server provides read capabilities, by default. You can enable write capabilities by enabling the advanced feature flag — for information, see the readme file. We recommend that you test write operations in a sandbox environment before trying them in production.

• Before sending any generated campaign or other content, send proofs to yourself first.

• Your API key permissions govern access. Use the least-privileged key you can. For more information, refer to API Keys.

# What's supported

• View the available Iterable MCP tools.

• Experiments are currently supported only for email.

• Tool behavior (along with supported permissions) mirrors the underlying Iterable API.

# Example uses and helpful tips

With Iterable's MCP Server, you can ask your AI assistant to look up a user, list campaigns, create a template, or send email. It's up to the LLM to figure out which tools to use to address your question. For example, you might ask your AI assistant to:

• "Summarize active campaigns right now."
• "Show email performance (opens/clicks) for the last seven days."
• "Show list memberships for user@example.com."
• "Create a new campaign from template template_id and schedule it to send to a sandbox tomorrow at 9:00 a.m."
• "List product catalogs and show 10 items from demo-catalog."
• "Enable debug logs and retrieve journeys filtered by state Running."

General prompting best practices apply here, as they do with any AI assistant.