Yesterday, I tried claude code as well as any cli based ai tool for the time. Before this, I had played around with copilot in IDE only since my client has not given copilot cli access yet. So, everything seemed new and unknown but within 5 mins I felt at home as a developer who has always seen terminal as a developer’s workbench.
Going back to the title of this blog, the reason why I titled as such is because using claude I felt like I was both involved in the development process/ the thinking which I enjoy the most and it gave me freedom to explore new things, learn something else, go watch anime while it was doing the execution.
I maxed out my usage within 3 hours of starting up :) Here are my learnings from the session:
-
Trust claude - It was able to handle large feature request given proper planning/requirements. I tried the AskUserQuestionTool so that it asks more questions.
Below is the prompt which I found on the internet that I used:
Interview me in detail using AskUserQuestionTool about literally anything: technical, implementation, UI & UX, concerns, tradeoffs etc. -
Token usage is a bottleneck for me in Pro plan. This will always be a bottlneck until I upgrade to Max account but till then I can make adjustments so that per token value creation is more by applying context engineering practices.
-
Claude code made changes to multiple repos given their code path. I was not expecting that claude code would be able to do that.
-
For UI/UX change, had to try couple of times to reach a working solution. Chrome was not enabled in this case. Also, I didn’t paste the screenshot of the UI. I have to improve the workflow here.