Content
@
0 reply
0 recast
2 reactions
Manan
@manan
Trying out devin.ai. On a scale of 1 to 10 - how do you think Devin did on the following? This was my prompt "Hey @Devin, Add descriptions, examples or standard components for every property in hub-rest-api open API definition file in neynarxyz/OAS repo. Use your best judgment to find out what each of property means by searching through https://docs.farcaster.xyz/reference/" and here's the PR it cooked up. https://github.com/neynarxyz/OAS/pull/272/files
3 replies
11 recasts
70 reactions
jp 🎩
@jpfraneto.eth
how do you think it did on that scale?
1 reply
0 recast
0 reaction
Manan
@manan
still reviewing, but trending towards a 7 at the moment. it messed up in a few places where it removed links and added his own interpretations like this one
1 reply
0 recast
3 reactions
jp 🎩
@jpfraneto.eth
that's pretty good. one thing that is a challenge right now is how to decide which ai tool to use. like i have cursor, o1 and claude open, and im like: ok. which one do i prompt? which one do i use? this last week ive prompted the three of them to see the difference in results, but would love to aggregate somehow
0 reply
0 recast
2 reactions