Most writings about the spec-driven development I see start with a product requirements document that is assumed to be valid. But I doubt that's the case. If so, you would've written about it, and probably would've involved agents in the research that goes into it. My gut feeling tells me there's much more emphasis on implementing the feature than on questioning if it's relevant, feasible, and based on valid assumptions.

Yes, it's called "questioning attitude", one of the traits of a healthy nuclear safety culture (and a good thing to apply in other fields!)

https://www.nrc.gov/docs/ML1433/ML14338A739.pdf

Most of my energy is refining a prd these days.

Then how come that process is not agentic and not well-described?

Personally it's well-defined and agentic - just not circulated.

/understand - agents interrogate the problem /huddle - Thinking panel turns it into a PRD - attacks the premise, PRDs regularly die here /tm - claude-task-master breaks the survivor into a dependency graph

Nobody writes this half up because "agent talked me out of building it" demos worse than "agent built it".

Sorry I have to ask. How senior are you? The notion that I‘d allow an agent to talk me out of something seems weird. 99% of cases, it’s the other way around. Architecture is just not where they shine.

What’s your process? My experience matches yours, but then again I usually just give a few lines to codex. I imagine if I tried harder to give detailed specs as input, the agent would have a lot more room to spot flaws and kill the plan.

This has been my attempt at wrangling the new A.I. assisted development that seems to be overtaking the software engineering profession. I jumped head first into LLM development after observing the trends from the last year and it appears this process might be a viable path forward.

There's a selection bias here that nobody's mentioned. The people who had bad experiences with this approach probably aren't commenting.

I’ve had similar feelings how can I trust this if I no longer write the code directly.

I wrote an /assess tool. I designed it to be token light but assesses on everything I could do to regain trust and help AI to improve my code base not by add features but by adding discipline.