It's the same story that it's always been, agents or not, that engineers need to be analysts and translate poorly defined criteria into something that's fit for purpose.
Something I've been doing in my own organization, but also trying to help other organizations with, is getting engineers closer to the customers now that building and the time it takes to build is no longer the resource, which is scarce.
But because we will do so anyway, people should adapt and start to write their tickets like prompts.
This has always worked like this. Let the person working on the ticket discover the implementation details, break it down into tasks, maybe delegate some after they’ve dipped their toes and know how to split it into sub-tasks better.
And of course, obligatory Shape up mention: https://basecamp.com/shapeup
The way to really figure out what needs to be done is to start doing real work.
https://basecamp.com/shapeup/3.1-chapter-10#imagined-vs-disc...
imagined tasks == Jira
discovered tasks == Dark Jira
Planned work should be organically discovered from the plan by the developers (or agents) who will be implementing it, not assigned via Jira tickets by the project manager. Shape Up recommedns using Hill Charts for per-scope (vertical slice) progress updates.
I have a favourite illustration of the way projects work, right at the top of this page: https://bureau.ru/about/fff/
(The page itself is in Russian. Basically, it’s about the “flexible scope” principle from, you guessed it, Getting Real: https://basecamp.com/gettingreal/02.4-fix-time-and-budget-fl...)
"Assign agents the biggest piece justifiable. I can summarize a product outcome or a feature in two lines. That’s what goes on the ticket. Let the agents figure out subtasks when the work is ready for review, not before. Once you break an initiative into technical issues upfront, the outcome gets lost and the focus shifts to minutiae."
This is not about the ticket being well defined, this is about the agent having the larger context of what you are trying to do
And, it goes both ways. Code and back-and-forth prompting can write clarifying comments or update acceptance criteria with specificity. Likewise, agents can add comments for the handover log and, in the morning, pull comments for context.
I've even had the agents read/write sub-tickets, follow-on tickets, sub-tasks, etc. which can new reviewed and modified by myself and teammates in the larger planning context. It's a delight!