How to connect indigitall to your AI assistant with MCP

Your indigitall data already holds the answers to most of your day-to-day questions. Which journey converted best last week? How did last month's WhatsApp costs break down? Until now, getting to those answers meant opening the platform and building a view. With the indigitall MCP, you can ask your AI assistant instead, in plain language, and get the answer back in seconds.

What is the indigitall MCP?

The Model Context Protocol (MCP) is the open standard that AI assistants like Claude use to connect, securely and in a standardized way, to external data and tools. The indigitall MCP Gateway exposes your platform through that protocol. Instead of building a one-off integration, your assistant gets a single, secure connector to the statistics that live inside indigitall. Once it is connected, you describe what you want in natural language and your assistant returns the numbers: campaign performance, journey conversion, chat metrics, and messaging costs, ready to read and act on.

It is read-only, it works on your real live data, and it is included for every indigitall customer.

What you can access

The connector covers the statistics across your platform:

• Natural-language stats. Ask for campaign, journey, chat, and pricing metrics in plain language and get clear, structured answers back.
• Customer Journey. Daily conversion stats per journey, broken down by platform (iOS, Android, Web Push) and by journey state.
• Chat and Contact Center. Aggregate and per-campaign chat metrics such as sent, delivered, read, and errors, plus daily, weekly, and monthly contact-center reports.
• Meta and WhatsApp pricing. Billable messages by category (authentication, marketing, utility) with cost time-series by hour or day.
• Raw CSV and ZIP export. Non-aggregated data served as authenticated file downloads, so you can pull the underlying numbers straight into your own tools.
• Read-only by design. The connector only reads statistics. It never sends a campaign or changes anything in your account.

What can you ask it?

These are real prompts that work against the live connector:

• Show my Customer Journey conversion stats for the last 2 days
• How did each journey state convert yesterday?
• What are the most common chat error types this week?
• List my chat campaigns with delivery and read rates
• Which chat campaigns had the most errors last week?
• Break down Meta billable messages by category for last month
• Show the WhatsApp messaging cost trend by day
• Give me a project overview with the latest Meta aggregates
• Export the raw daily chat data as a CSV

How to connect indigitall to each assistant

The gateway speaks standard MCP, so it works with any MCP-capable assistant. The MCP server URL you will need is https://mcp.indigitall.com/mcp. Here is how the four most common assistants connect.

Claude (one-click or manual)

The fastest path is one click, which opens Claude with the connector pre-filled and sends you through indigitall's authorization. To set it up manually, open Settings, Connectors, choose Add custom connector, set the name to indigitall, and paste the MCP server URL. Click Add, and Claude walks you through indigitall's authorization to finish. There is no key to paste.

ChatGPT (connector)

Open Settings, Connectors. If your plan requires it for custom MCP connectors, enable Developer mode. Choose Add custom connector, set the name to indigitall, and paste the MCP server URL. Save, then start a chat with the connector enabled. ChatGPT sends you through indigitall's authorization the first time it runs.

Gemini CLI (config file)

Add the block below to ~/.config/gemini/settings.json, then restart the CLI:

{
  "mcpServers": {
    "indigitall-mcp": {
      "type": "http",
      "url": "https://xxx.mcp.indigitall.com/mcp",
      "headers": {
        "Authorization": "ServerKey <valid-server-key>",
        "X-Application-Id": "12345"
      }
    }
  }
}

Replace <valid-server-key> with your ServerKey and 12345 with your project's Application Id. Keep the word ServerKey and the space before the key. Your ServerKey stays in the config file and should never be pasted into a chat.

GitHub Copilot (config file)

Add the block below to ~/.copilot/mcp-config.json, then reload Copilot:

{
  "servers": {
    "indigitall-mcp": {
      "type": "http",
      "url": "https://xxx.mcp.indigitall.com/mcp",
      "headers": {
        "Authorization": "ServerKey <valid-server-key>",
        "X-Application-Id": "12345"
      }
    }
  }
}

Replace the same two values. As with Gemini, the key stays in the config file.

Is it secure?

Yes. The connector is read-only. It reads your statistics and nothing else. It cannot send a campaign, edit a journey, or change any setting in your account. Authorization happens on indigitall's side, so for the one-click assistants you never paste a key, and for the config-based assistants the key stays in your local file. Internal traffic between the gateway and platform services is signed with short-lived tokens, and your data stays in your platform. The connector adds a conversational layer on top of indigitall. It does not move your data anywhere new.

Frequently asked questions

What data does it read? Only your platform statistics: campaign, journey, chat, and pricing metrics from the stats service.

Can it operate on my account? No. It is read-only. It retrieves stats and reports and never changes anything.

Which assistants work? Any MCP-capable assistant. This guide covers the four most common: Claude, ChatGPT, Gemini CLI, and GitHub Copilot. Claude and ChatGPT use server-side authorization. Gemini CLI and Copilot use a config file with your ServerKey.

Who maintains it? indigitall's engineering team, as part of the MCP Gateway initiative.

Start asking your data

The indigitall MCP is included with your account. Connect your assistant, ask a question, and get the answer in seconds.