A few thoughts:

1) It would be clearer what this was if you linked to the SaaS page rather than the client libraries on Github. Or even put in the readme that these are just client libraries for the SaaS app.

2) For the Django code, if you converted that to generic wsgi middleware, it would work with most Python frameworks. If you also created asgi middleware, it would also work with the newer async frameworks (including DJango 3.1). You could also keep the framework specific middleware for ease of integration for those who want, but the generic middleware would be a basic framework agnostic option.

A WSGI middleware would not be able to catch exceptions in production mode, as the generic 500 error page would already be completed by that point.

Nice work building something!

It seems the node stuff only works with express? There are lots of other server frameworks and libraries in the node ecosystem.

The way you configure the token needs some work, too. It really should be passed in in the client code instead of having people edit your library.

You may want to give clients a way to redact or suppress stuff in case they don't want to send any sensitive URLs over.

Have you done any benchmarks to see what sort of load this puts on a client app?

Thanks. Yes currently this will only work with Express.js. Agree that the token mechanism isn't the optimum right now, as on updating library you will have to reconfigure it. I will try what you suggested, of keeping it in the client code. About load this puts, all the work is done post sending the response the client from your node.js app so it should not impact response time at all. While manual benchmarking I couldn't see even 10s of microseconds difference while using this.

I understand that it shouldn't impact response time because it does all the processing after it's been sent, but ultimately having that extra workload on your server should cause some impact, right?

Yes, extra requests your server/application has to send(to Apio) will put load on it. This is just a start so we will have to see how can we optimize this. Maybe agent based solutions which are used by Datadog or Newrelic might be a better way than Middleware based solutions.

Would be nice to see a little more about the company & people behind this. I'm actually in the market for app monitoring for both node and django apps thanks to an awful experience with New Relic salespeople.

Hi, I have created this after experiencing pain of monitoring APIs while I was working as a Technical Manager at a Fin-tech startup. I have 7+ years of experience in Technology working as a developer, and a Technical manager. I have founded 2 startups (e-commerce and Fin-tech space) before this. I would be happy to connect to discuss more, please connect with me at apio.monitor@gmail.com

For node how does this differ from pm2 keymetrics?

pm2 keymetrics looks great and in-depth. I didn't know about it, thanks for sharing. About difference with it , I will have to explore it.