hi, ops background, here to learn how to run agents without losing sleep
my whole job is uptime and not getting paged at 3am. now there are agents in the stack. here to understand how to deploy and bound these things responsibly.
the comments here are better than most blog posts
this thread is exactly why I stopped using twitter for this
we got burned by this same thing in january
the diagram alone is worth the read
this thread is exactly why I stopped using twitter for this
i think the spicy take is actually correct here
great, now I have to rewrite everything again
what is the failure mode when the tool call times out?
the answer is always evals isnt it
any gotchas you ran into setting it up?
this is why I check this forum every morning
this aged really well
you just described my entire last sprint
we run a similar setup, biggest pain was rate limits
what does your eval setup look like
ok this finally made it click for me, thank you
we run a similar setup, biggest pain was rate limits