I've just started using Claude for my project and I'm curious about its memory capabilities. Claude often prompts me that the chat is too long and suggests starting a new one, but when I do, it seems like it doesn't remember anything from the previous chat, even when in a project folder. How do you all manage to retain important information for more complicated projects when you have to switch chats? Any tips on keeping track of vital details?
2 Answers
To handle the memory issue, you can use the conversation picker by starting Claude with `claude --resume`. If you use `claude --continue`, it will resume your last conversation. You can check `claude --help` for more launch options too; some are in the docs but not super obvious. For remembering key info, consider using `# put xyz we’ve talked about in memory` to save what’s important for future chats. It's all about organizing your thoughts!
Is that feature only available via API, or can it be used in regular chats too?
If you want to retain essential details, you could either save it directly in your project files, or ask Claude to summarize the key points from your conversation and use that summary in your new chat. Just keep in mind that for software development, that might not always work seamlessly. Breaking your discussions into smaller parts could be more effective!
I'm working on a CMS module too! I had Claude create detailed documentation. It was somewhat helpful, but took a lot of effort.
For coding specifically with Claude, check out that other response—it has some great advice that's super helpful!
Thanks for the tip! When you use --continue, does Claude remember everything from the last chat?