MCP server

An MCP server is a program that lets an AI assistant, such as Claude or ChatGPT, use another app's data and actions through the Model Context Protocol. Inrō runs one, so your assistant can read and build your automations.

For the day-to-day use, see what the Inrō MCP server does for your Instagram account and the Inrō MCP server product page.

How does an MCP server work?

The Model Context Protocol is an open-source standard for connecting AI applications to outside systems (modelcontextprotocol.io, checked 1 October 2026). Its architecture page names three parts:

• Host: the AI application you talk to, such as Claude
• Client: the connection the host opens to one server
• Server: the program that provides context to the client

A server offers tools (actions the assistant can call), resources (data it can read) and prompts (reusable templates). It can run on your own computer or remotely over HTTP. For remote servers, the protocol recommends OAuth to sign in.

What can an assistant do through the Inrō MCP server?

Inrō's server is a remote one. Through it, your assistant can:

• list and read the scenarios on your Scenarios page
• create one from a template, from a sentence that describes it, or from a full definition
• update a scenario, switch it on or off, and trigger it for contacts you choose

Before it builds, it can search the template library and look up each trigger and action type. Each run it starts is a scenario execution. It also reaches your contacts, folders, DM campaigns and posts.

Reading and changing are separate, labelled tools, so an assistant that asks for approval can ask before a change and skip it for a read.

How does an assistant connect to Inrō?

Open the account menu, go to API & MCP, and copy the URL from the MCP Server card. Add it to your assistant as a connector and sign in; the two ways to sign in are in the API token entry.

A connection is tied to one Inrō account, and the assistant acts only inside it. The MCP server needs a Pro plan or the Pro trial: a Free account is refused. You can review every call it makes under API & MCP logs.

Is an MCP server the same as an API?

No. The Inrō API is for code you write. The MCP server is for an assistant: it reads the list of tools, each with its description, and picks the calls itself. You ask in plain words, such as a comment to DM on one post, and it does the rest.

To link Inrō with other apps, see Inrō integrations. More terms are in the Inrō glossary.