Ben Holmes
👤 SpeakerAppearances Over Time
Podcast Appearances
So if you want to like run agents inside of GitHub using the Oz runner in order to handle that, you can certainly do so.
But the boundaries of like where you can trigger these things, I mean, it's pretty soft.
As Zach was mentioning, you can trigger them using the REST API or the SDK.
like an application code that would trigger it, you could certainly do that.
Really, it's really just wherever you want to deploy it inside of your systems.
And another thing you mentioned was automations.
I should mention that there's a built-in scheduler with this as well.
So if you're used to piggybacking on GitHub for like setting up a cron job in order to have an agent run something, you no longer have to do that through a GitHub action.
You could just go to the web portal or your warp terminal, wherever you want to set it up.
And you could say, I want this agent to run every morning at 10 a.m.
And I want it to run inside of this environment that has our client and our server side code bases cloned into that environment.
And then every time it wakes up at 10 a.m., I want it to, for example, go through and look for stale feature flags.
So maybe you want to set that up on a weekly cron or a daily cron where it goes through all the code that's been committed across those repositories, analyze it and look for any stale feature flags that have existed in the code base for a month or two, which is very common across all applications.
and maybe come up with a plan on what feature flags to remove if you just wanted to do the research step.
Or you could have it go all the way and say, I'm going to remove the top five staled feature flags and open a pull request for your team to review.
And you can make it as autonomous as you want.
So you could have it open a PR, link to the planning document that it wrote, open it and ping and assign reviewers if you have specific reviewers that you want to be looking at that action.
However deep you want to go, you certainly can.