Replies: 2 comments
-
|
I personally have my CLAUDE.md pointing to the relevant documentation for common things and some basic info about agent-os that hints it to read the necessary files as needed. I also always switch to claude's plan mode when I start on these small things and manually point to the relevant context if it can't guess based on my initial prompt. Perhaps with the support for claude skills that came with 2.1 this will be easier? I just upgraded some projects to that yesterday, will start using today 🤞 |
Beta Was this translation helpful? Give feedback.
-
|
I've had some success with just calling /execute-tasks directly and providing a detailed prompt after that. Uses up those tokens though so maybe using cursor or the like is better. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Sometimes I don't have a big project that has a spec. I just have a tiny issue that needs fixing, or maybe I'm iterating on an old implementation.
How do you usually use agent os for this? I have a lot of context there that I want, and normal claude doesn't cut it and doesn't reach for it automatically when I ask for such a change.
I wish agent OS was like a tool that claude always uses instead of requiring manual effort from me to use.
Beta Was this translation helpful? Give feedback.
All reactions