I am developing my custom ratings for benchmark and currently a lot of my detection mechanism is in ratings.
What would be really cool is that executor on env would have method performLint that would be executed inside the agentic loop.
It does not have to have repair logic (though would be nice).
It simply has to run npm run lint and take the report and pass it on to attempt object - this way i can get that in PER_BUILD ratings
I am developing my custom ratings for benchmark and currently a lot of my detection mechanism is in ratings.
What would be really cool is that executor on env would have method performLint that would be executed inside the agentic loop.
It does not have to have repair logic (though would be nice).
It simply has to run
npm run lintand take the report and pass it on to attempt object - this way i can get that in PER_BUILD ratings