Banning shortcuts in personal projects is savage but I respect it. My AGENTS.md would just be a list of things I'm not allowed to do anymore. No quick fixes, no temporary solutions - the model has to defend every band-aid. That's a high bar.
Distributed backend specialist. Perfectly happy playing second fiddle—it means I get to chase fun ideas, dodge meetings, and break things no one told me to touch, all without anyone questioning it. 😇
Yes, it is a very high bar. It's also expensive run cost, but it accomplishes the end result well enough. I also allow zero issues/vulnerabilities of any kind regardless of severity in the codebase. MCPs are connected with self-healing so each prompt can take > 25 minutes to execute. But I'm not hand-holding AI through changes any more and that was the goal. 🤩
Twenty‑five minutes per prompt is a heroic commitment. I used to roll my eyes at that, but honestly? Precision that high means you're never burning hours or millions of tokens cleaning up misunderstood requirements later. That trade‑off might be nuts for a quick prototype, but for zero‑vulnerability, self‑healing infrastructure? Respect. You've chosen accuracy over speed, and that's a valid, underrated path.
Distributed backend specialist. Perfectly happy playing second fiddle—it means I get to chase fun ideas, dodge meetings, and break things no one told me to touch, all without anyone questioning it. 😇
That is true and storing/updating skills as you go cuts a lot of time down over the long-term. Which I say while I'm also iterating a third UI design... 🤣 There's a balance!
For further actions, you may consider blocking this person and/or reporting abuse
We're a place where coders share, stay up-to-date and grow their careers.
Banning shortcuts in personal projects is savage but I respect it. My AGENTS.md would just be a list of things I'm not allowed to do anymore. No quick fixes, no temporary solutions - the model has to defend every band-aid. That's a high bar.
Yes, it is a very high bar. It's also expensive run cost, but it accomplishes the end result well enough. I also allow zero issues/vulnerabilities of any kind regardless of severity in the codebase. MCPs are connected with self-healing so each prompt can take > 25 minutes to execute. But I'm not hand-holding AI through changes any more and that was the goal. 🤩
Twenty‑five minutes per prompt is a heroic commitment. I used to roll my eyes at that, but honestly? Precision that high means you're never burning hours or millions of tokens cleaning up misunderstood requirements later. That trade‑off might be nuts for a quick prototype, but for zero‑vulnerability, self‑healing infrastructure? Respect. You've chosen accuracy over speed, and that's a valid, underrated path.
That is true and storing/updating skills as you go cuts a lot of time down over the long-term. Which I say while I'm also iterating a third UI design... 🤣 There's a balance!