Sandy_liusy @onlyyixxs
After Ali’s testing, he exceeded my expectations! These are the four things that I got amazing and four things I didn’t do:
1. Without imposing rules, there are task implementation plans and automatic administration. This is very suitable for vibing coding
2. Clearly tell you Todo’s list and which files have been edited, which is suitable for small module programming
3. The instructions were followed in well, unexpectedly followed well!
4. Chat/agent mode, MCP, context management, all of them have been endorsed! Great, the trial version can be used!
Some of the shortcomings:
1. The interaction needs to be improved. For example, you can’t “increase the chat” when you choose the text in the terminal area. It is endorsed by both the tree and the cursor. I hope I will add this feature as soon as possible. This is the function in the code editor.
2. It seems that you cannot re -comply with a particular request. If you are not satisfied with a particular process, you can only “revoke the changes”.
3. Unable to choose a model (of course, sometimes not giving consumers a choice is a manifestation of confidence, provided the model is so strong that the cost of choosing consumers can be saved.)
4. Credits end very quickly. I have already spent 319/2000 after asking 5 questions. If I use it for work, I’m afraid it will end in 2 days. (Looking forward to improving credit consumption)