ryan

338 posts

ryan banner
ryan

ryan

@ryanpcrab

Katılım Temmuz 2015
504 Takip Edilen230 Takipçiler
Nate Berkopec
Nate Berkopec@nateberkopec·
A pattern I am seeing: "Generate a 10 question quiz on how this PR works to test my understanding." Do not allow merge until someone passes the quiz. Gusto did this years ago but with humans writing the questions for certain types of PRs.
English
8
5
210
13.9K
ryan
ryan@ryanpcrab·
@MariaSolanoOs Enjoyed reading your blog posts, thank you!
English
1
0
0
66
ryan
ryan@ryanpcrab·
@arethevibesoff @levelsio Hey I think the paper you linked says that agentsmd can make it a little worse if it’s an AI generated agents file BUT if it’s human written it produced a little better Sounds like - make sure you’re putting useful things in there instead of broadly letting AI make it
English
0
0
0
42
Matt
Matt@agentdevrel·
@levelsio Just talk to them remains undefeated Paper just came out that says an agentsmd file can make performance worse
English
2
0
0
1.1K
@levelsio
@levelsio@levelsio·
My prompts are really quite short and it just understands what I mean most of the time, I talk to it like a friend, I see other people use giant prompts and wonder if it actually makes a difference?
Joan Capdevila@joancapdevila

@levelsio the good thing about talking to him is that you can give him a lot of extra context that really makes a difference and helps understand what you want for that task.. i also feel weird about it but it's way better than typing.

English
118
0
361
71.8K
ryan
ryan@ryanpcrab·
@ryanflorence This is so funny it feels like the authors trying to criticize the failure of a group while propping up a survivor
English
0
0
0
400
ryan
ryan@ryanpcrab·
@bicccio @unclebobmartin Haha I don’t follow, If it forgets an edge case for a test? Or forgets entire thing 1. It’s going to, it’s AI, prob better to define those yourself 2. If you give it a programmatic way to check I think it gets to be pretty consistent? Ex Run tests - sees failures, lint, scripts
English
0
0
0
45
Uncle Bob Martin
Uncle Bob Martin@unclebobmartin·
Me: Claude, you didn't write tests. Claude: Yeah, sorry, I'll write them now. Me: Don't your instructions tell you to ALWAYS write tests. Claude: Yeah, they do. I shoulda written tests. Me: Why the F didn't you write the tests. Claude: Meh. Me fogaht.
English
134
32
985
78.7K
ryan
ryan@ryanpcrab·
Imagine coding agents couldn’t make pr’s with the gh cli because you have to pass a quiz on your changes before it creates it
English
1
0
3
1.5K
ryan
ryan@ryanpcrab·
@mitchellh @fatih Hi, is there a way to programmatically focus a specific split from another process?
English
0
0
0
85
Mitchell Hashimoto
Mitchell Hashimoto@mitchellh·
@fatih Not quite yet. It’s our most upvoted issues now or one of them and plan to work on it in more seriousness soon
English
6
0
84
11.2K
Fatih Arslan
Fatih Arslan@fatih·
Is it possible to programmatically open a split window in Ghostty? Sometimes I want to create a new Vim keybinding that opens a new split, and executes a command. I could do that in Tmux, but couldn’t find a way in Ghostty. cc @mitchellh
English
9
1
35
18.6K
ryan
ryan@ryanpcrab·
get it twisted, you are 15 minutes of reading away from being more educated on any given topic than the average person
English
0
0
0
31
ryan retweetledi
Teg🚨
Teg🚨@IQfor3·
Hope you got everything you wanted for Christmas big bro @tyler_kolek
English
25
252
2.9K
50.7K
ryan retweetledi
jaycee
jaycee@jayceepxrez·
All New York teams are in the gutter… it’s time.
English
32
572
3.8K
96.2K
ryan
ryan@ryanpcrab·
@ZAIDDDDDDDD This is ridiculous nobody asking the hard questions anymore this what’s wrong with politics. Yo Zohran what are we gonna do about Mikal Bridges going up soft? Seriously
English
1
0
1
82
ryan retweetledi
Teg🚨
Teg🚨@IQfor3·
I’ve never felt this way a day in my life
English
41
80
1.4K
62.4K
ryan retweetledi
Teg🚨
Teg🚨@IQfor3·
OH MY GODDDDD
English
61
747
3.5K
199.3K
ryan retweetledi
Jalen Brunson
Jalen Brunson@jalenbrunson1·
The time will come when winter will ask what you were doing all summer...
English
13
152
557
0