fbpx
  1. Tubelator AI
  2. >
  3. Videos
  4. >
  5. Science & Technology
  6. >
  7. How to Improve AI Output Quality in Coding Editors Using Cursor and Windsurf Settings

How to Improve AI Output Quality in Coding Editors Using Cursor and Windsurf Settings

Available In Following Subtitles
English
Variant 1
Posted on:
Learn how to increase AI output quality in coding editors like VS Code using simple rules files with cursor and Windsurf settings. Discover the importance of providing concrete context for greater reliability and effectiveness in AI-assisted coding.
tubelator logo

Instantly generate YouTube summary, transcript and subtitles!

chrome-icon Install Tubelator On Chrome

Video Summary & Chapters

No chapters for this video generated yet.

Video Transcript

0:00
context is ke for AI here's what people
0:03
frequently forget AI can't magically
0:05
guess everything when building Real
0:08
World products you can't rely on AI to
0:10
just guess everything correctly the
0:12
secret is not about better prompts or
0:15
more powerful models it's about concrete
0:18
context today I'll show you how one
0:20
simple rules file can transform your AI
0:23
coding assistant into a much more
0:25
reliable development partner whether
0:27
you're using vs code cursor or WI or any
0:31
other coding editors that supports a
0:33
similar feature let's dive straight in
0:35
first of all these are context that get
0:38
included on all of your AI chats whether
0:41
it's your composer runs your chat your
0:43
wind Cascades your GitHub co-pilot edit
0:46
sessions as previously mentioned the
0:48
three major code editors all have a
0:50
similar feature here but just stored in
0:52
different files vs code keeps this
0:55
inside the GitHub Code Pilot
0:57
instructions cursor there's a cursor
0:59
rules files at the root of the room
1:00
repository and wind surf there's a do
1:02
wind surve rules that has just been
1:04
recently released let's create a new
1:06
chat when we take type anything short
1:09
hello how's it going this context gets
1:11
included directly as part of that prompt
1:14
automatically just as something obvious
1:17
let's say ho ho ho at the beginning of
1:18
the output every single time then we can
1:20
save the file and type this into cursor
1:23
chat how's it going you'll see it always
1:25
says ho ho ho there's a very easy way to
1:28
verify whether your rules file is
1:30
working correctly or not you'll notice
1:32
one key rule I've applied at the
1:33
beginning is this line every time you
1:36
choose to apply a rule explicitly state
1:38
the rule in the output you can
1:40
abbreviate the rule description to a
1:42
single word or phrase this is to make
1:45
sure that I can clearly see in the
1:47
output which rules has been followed
1:50
specifically for each one of the edit
1:52
sessions just so that I can easily
1:54
sanity check oh has it missed anything
shape-icon

Download extension to view full transcript.

chrome-icon Install Tubelator On Chrome