Reply cited one hundred markdown files at repository root.
+without careful cleaning and rules, coding agent workspaces become insanely, unusably messy within a week
90% of issues people hit with models are actually due to the shitty harnesses people are using. like you guys do realize it’s just an api endpoint you have to engineer around?
Some users dismissed the post arguing that poor harnesses cause most AI model issues as immature middle-school content and replied with personal insults.
No Digg Deeper questions have been answered for this story yet.
@menhguin wdym. i love my 100 markdown files stored at repo root
+without careful cleaning and rules, coding agent workspaces become insanely, unusably messy within a week
@menhguin Similar theories.
@MmHabitual damn this post is hitting #middleschooltwt fr
@menhguin @MmHabitual small penis jungle monkey mentally stuck in middle school 😂😂😂
@andersonbcdefg yeah i have a lot of questionably categorised files and at least timestamps help me with "wtf was i think here making 10 vaguely named files" and then my agent has to spend 10 minutes just on forensics lmao
@menhguin i already have loose md and then every so often i put the old ones in a folder called "old". timestamp is good idea
@menhguin even context alone is something wortht to actively clean, many models claim 1m token capacity but their performance degrades significantly after 200-300k tokens in context
@BerntsenN49007 @MmHabitual average sized (!)