I developed an AI SRE Investigator at Pirates of the Coral-bean hackathon.
The idea was simple: when something goes wrong in production, engineers are wasting time context switching between tools to find out what happened. GitHub has recent changes, Sentry logs errors, Slack is context but it all lives out there.
All of that, I wanted to tie together.
What I built
I created a wrapper around Github, Sentry, and Slack using Coral which allows you to run SQL-like queries across these tools.
Its backend, written in Go, communicates with Coral. Frontend is a dashboard that shows results simply in one place.
How it works:
Instead of going through manually all the systems, the app queries:
a. Recent pull requests from GitHub
b. Active errors from Sentry
c. Operational signals from Slack
It stitches everything together into one view so you can see what has changed and what triggered the problem (at least at a high level).
Why I built it
So I wanted to see how incident investigation could feel like a more rapid and a less fragmented thing. Detecting the error is not the most difficult part of real systems; the hardest part is connecting the dots with changes, errors, and communication.
This project is a baby step along its course.
Closing thoughts
It was an enjoyable build, a good primitive experience around systems design and integrations, and working with real operational data.
Top comments (0)