Added an internal feedback button to a side project that's basically designed for AI debugging.
When clicked, it captures:
- •My quick feedback text
- •Current user context / db state
- •Frontend state
And it formats everything for LLMs (with xml tags around the various sections for example).
So now I can literally just click “Feedback”, write something, click copy, paste into Claude Code, and finally press enter.
The agent has everything it needs to understand and fix the issue or make the improvement.
No hunting for logs, no "can you send me the state?", no missing context.
So I'm now designing features specifically to work well with AI agents, it’s part of the process. Both in side projects and Typefully.
Completely different product dev mindset from even just a few months ago.
