mcp

This MCP server acts as a bridge between your AI client and Iterable, translating natural language requests into safe Iterable API calls. It lets AI assistants securely connect to Iterable data and operations without exposing raw API keys to the AI, and it supports multiple AI clients through the MCP protocol. With this server running, you can ask questions like "+How many campaigns did we send last week?+" or instruct it to perform actions such as creating templates, exporting data, or listing catalogs, depending on the permissions you enable during setup. The server exposes a set of tools that map to Iterable API endpoints, allowing your AI client to read data, manage resources, and (if enabled) perform writes and sends through a controlled interface. You can manage API keys, select endpoints (US/EU or custom), and switch between keys as needed, all while the MCP server automatically uses the currently active key.

To use it with your AI client, configure the client to route tool calls through the MCP server and ensure the appropriate permissions are set. You’ll typically interact with the server by issuing natural language prompts that are translated into safe, auditable API calls to Iterable. If you enable advanced capabilities, the agent can perform actions like creating campaigns, updating templates, or sending messages, so it’s important to review actions before they execute in production environments.

Prerequisites:

• Node.js 20+ (22 LTS recommended)
• An Iterable API key (stored and managed via the MCP setup workflow)

Installation steps:

1. Install and configure via NPX setup (recommended):

npx @iterable/mcp setup

2. (Optional) Use auto-update on restart:

npx @iterable/mcp setup --auto-update

3. If you need advanced permissions during setup (e.g., writes or sends):

npx @iterable/mcp setup --advanced

4. If you prefer a global install as a shorthand:

pnpm add -g @iterable/mcp
iterable-mcp setup

Note: The setup command configures the MCP server paths and prompts you to connect to your Iterable account via API keys managed by the tool.

Tips and common considerations:

• Permissions: Enabling writes and sends allows real actions against your Iterable project. Enable these only if you understand the implications and have reviewed actions prior to execution.
• Key management: Keys are stored with active key management and can be switched with the provided commands. You can have multiple keys for production, staging, and dev environments.
• Environment variables: If you configure the MCP server manually, you can supply ITERABLE_API_KEY and ITERABLE_BASE_URL in the env section. This is optional if you rely on the built-in key management, but it can be useful for project-scoped configurations.
• Claude/Cursor/Windsurf/Gemini/Antigravity integration: The MCP server provides a consistent configuration format for the various clients. You can edit their config files to point to the MCP server using the same mcp_servers structure shown in the documentation.
• Security: Treat the MCP server’s actions as potentially impactful. Always validate tool calls and review results, especially in production settings.