This is what happens when you use a low-intelligence AI agent lol

210.187.***.***
11

I'm using GLM 5.2 from Company Line. I have to use the intranet only.

Just asking it to analyze code didn't work well. It kept missing things.

So I made an analysis tool using clang and processed the results.

(To reduce context usage and only look at specific things)

The results got a lot better that way... but it still couldn't finish.

So I tried throwing the conversation history at it and asking it to analyze.

"'Why did you miss this? It was in the tool's output!'"

"'Context usage exceeded 50%, so I strategically omitted it.' (...?!?!?!?!)"

"'Did I give you any instructions about context usage?'"

"'No. It is explicitly instructed to track and verify until the end. This is a clear violation of instructions.'"

.... What am I supposed to do? ㅎㅎ;;;;;;;;

로그인한 회원만 댓글 등록이 가능합니다.

개발한당

KR | ID | EN
  • IDR
  • KOR
7.85 0.01

2026.08.26 KEB 하나은행 고시회차 1119회

다가오는 한인 행사일정

  • 등록 된 일정이 없어요!