MagicBell MCP. Manage multi-channel alerts from chat.

MagicBell manages all your alerts—the whole shebang, across every channel you use. You connect your AI client directly, and it gives you total control over broadcasting messages using natural language commands. This isn't some dashboard you gotta mess with; your agent runs these tools right from your code editor.

You want to know what happened last week? Use the list_broadcasts tool. It pulls up a paginated list of every single broadcast that’s gone out for your project. You get titles and unique IDs, giving you a full audit trail of all your communication efforts. That's how you track everything.

Need to dig into one specific message? Just grab the UUID and run get_broadcast. This tool fetches the complete status and all the metadata for that single broadcast. It tells you if it was 'processed,' what the full content was, and exactly how many people received it across all channels.

You're getting the deep dive here.

When it’s time to send a message—and you gotta send one—you call create_broadcast. This tool handles sending a brand new multi-channel alert instantly. You specify the title, the body text, and any necessary filters, and MagicBell triggers the alerts across email, SMS, or push channels automatically. It doesn't matter if you want it to hit all three; the server manages that routing for ya.

When a broadcast goes out using create_broadcast, the system is smart about reporting back. The server reports which specific channels were actually used—like Email or Web Push—and gives you a count of how many recipients got hit on each one. This means you know exactly what happened with every single message sent.

Think about it: You don't have to jump through hoops in some web portal just to send an alert or check if Uncle Joe actually got that push notification. Your AI client handles the whole cycle. First, you use list_broadcasts to get a quick overview of all previous sends; you see the titles and unique IDs pop up right away.

If you spot a UUID for something you need details on, you hit get_broadcast. That gives you the full picture: the content, the status confirmation, and the exact recipient count. It’s immediate verification. You're checking the metadata so you know it wasn't just sent; you know it was processed.

And when you need to send a message, that's where create_broadcast shines. You feed it the title and body text—the stuff people gotta read—and tell it what filters to use. It then fires off that alert across email, SMS, or push channels simultaneously. The system takes your single command and executes complex multi-channel routing.

You get confirmation on which specific channel was used for each recipient group, and how many folks got the message. It's airtight.

This whole setup means you keep all your communication control right inside your development environment. You don't gotta break flow switching between tools. Your agent calls list_broadcasts to see history; it calls get_broadcast when it needs confirmation on a specific alert ID; and it uses create_broadcast the second an alert is necessary.

It’s seamless, baby.