agents guess.
this one asks.
Hand an agent “add saved payments” and it invents scope, picks a data model you didn’t want, and skips the decisions that were yours to make. The Planning agent interviews you first — then writes tasks small enough to actually build.
$ ptln plan$ brew install partyline-sh/tap/partylineThe working doc is empty — it fills in as the agent proposes edits.
how it works
say what you want
A sentence or two. No template, no ticket format. Web or ptln plan — same agent, same plan.
answer the questions
It reads your project first, then asks only the decisions it can't make for you — scope, data model, UX — one at a time, each with a recommendation.
approve the plan
A tree of bounded tasks, each with acceptance criteria. Edit it, reorder it, promote what you approve.
what you're wondering
how is this not just prompting?
A prompt makes the agent guess. An interview pins the decisions first and records them — so every task downstream inherits them.
can the team join?
Yes. PM, designer, engineers in the same session, pressure-testing the plan on a shared doc before anyone writes code.
do I have to build it?
No. The session just produces the plan. Keep it as a backlog, run one task, or send the whole feature to your agents.
the decisions stay yours.
$ brew install partyline-sh/tap/partylinethe front half of the pipeline — see what happens after you approve →