Quick flow no longer prompts for Advanced Elicitation

[nguyenjessev]

In v6.2.1 the "quick" flow was simplified from two separate quick-spec and quick-dev skills to now just quick-dev. The old skills used to stop at each step (e.g. before deep research, before finalizing spec, etc.) and offer to perform "Advanced Elicitation" which I found to massively improve the quality of both the spec and the final output. However, the new quick-dev does not offer this at all, and now just goes straight from prompt to writing the spec to implementation. I find this to be a pretty big regression in output quality. Granted, multiple rounds of Advanced Elicitation definitely took a little extra time so maybe it was deemed to be against the spirit of a "quick" flow, but is there a reason this was removed from that flow? The previous behavior was nice because you could choose how quick or how thorough you wanted to be by skipping Advanced Elicitation or doing multiple rounds of it, but now it seems the only choice is to either use the extremely light quick-dev or commit to the full blown BMAD method with no in-between.

[danicatalan00]

My main workflow at my job was: issue title and description -> quick-spec -> quick dev -> review -> pr

The quick-spec skill was wonderfull, prooduced nice complete spec with all the context needed to handoff to any agent, workflow, bmad or not, and through interview, deep reaserch I got very thorough spec documents.

Now I got a much smaller spec document, with no checkboxes in AC, no expanded task lists, much much more compacted spec as a whole.

I don't know if it will hurt or improve performance yet, but I am worried as I liked the old tech-spec format a lot. I could install the older version through a previous commit in the repo, but I would like to not miss on other features and fixes via the standard way to update...

[nguyenjessev]

What I ended up doing was just looking at the diff between v6.2 and v6.2.1 and modifying the new step-02 skill to add in an option to do Advanced Elicitation and Adversarial Review at the step 2 checkpoint. Seems to work okay