At the moment we are taking a very IFTTT style approach where we have monitors that are external facing that can test things like TCP ports, HTTP status codes and anything we can get at with a webhook or API (i.e. Datadog, Papertrail, Heroku). When we resolve things it is all via external services that we can integrate with via API's or webhooks (Commando, DigitalOcean, Rackspace, AWS and so on).

On our roadmap is building an agent (and there is a wrapper for our webhooks), this would allow us to run monitors and reactions locally on the server itself. Where we would shine over a tool like Monit is that once we have an agent you could define that reaction to happen on any server.

So for example if we detected a database error using a keyword search on the web app. You would be able to execute commands to fix it on whatever servers you define.

So think of it like, being able to run the actions across as many servers as you want.