Managing context on the Claude Developer Platform • Hacker News Board

Re: Managing context on the Claude Developer Platform

#91

Post by swader999 » Tue, Oct 07, 2025, 1:41 PM UTC

daxfohl wrote: post #73

swader999 wrote: post #53

Earlier quoted context omitted.

I found that having sub agents just for running and writing unit tests got me over 90% of my context woes

Seems like that could be a job local LLMs do fairly well soon; not a ton of reasoning, just a basic ability to understand functions and write fairly boilerplate code, but it involves a ton of tokens, especially if you have lots of verbose output from a test run. So doing it locally could end up being a huge cost savings as well.

Maybe but you still need to pass some context to the sub agent to describe the system under test.

Re: Managing context on the Claude Developer Platform

#92

Post by bdangubic » Tue, Oct 07, 2025, 4:37 PM UTC

manwe150 wrote: post #85

bdangubic wrote: post #70

Earlier quoted context omitted.

the trouble with compact is that no one really knows how it works and what it does. hence, for me at least, there is just no way I would ever allow my context to get there. you should seriously reconsider ever using compact (I mean this literally) - the quality of CC at that point is order of magnitute significantly worse that you are doing yourself significant disservice

When you hit ^O after compact runs (or anytime), it tells you exactly what compact did, so it isn’t that mysterious

if you actually hit the compact (you should never be there no matter what but for the sake of argument) more often than not you'll see CC going off the rails immediately after compacting is done. it even doesn't know what it did let alone you :)

Re: Managing context on the Claude Developer Platform

#93

Post by selinkocalar » Wed, Oct 08, 2025, 12:12 AM UTC

Context management is one of those things that seems simple until you actually build with it. We've run into issues where Claude loses important context halfway through complex tasks. Love the idea of being able to mark parts of the conversation as 'resolved' or 'outdated' would be huge.