Scope Is an Input
A workshop brief should not be the place where scope slowly emerges while execution has already begun. It is a materialized work order: a versioned statement of what the requester authorized, what the executor may change, what must remain untouched, and what evidence will close the task.
Calling scope an input changes the design. The engine validates that it exists before take. The brief records it before tools run. A plan gate may refine the method inside those boundaries, but it cannot silently widen the target. If the premises change, the work needs a visible amendment or a new delegation rather than a clever interpretation.
This boundary also keeps conversation from becoming hidden state. A useful discussion can happen anywhere, but the executable task cannot depend on somebody remembering an unstamped sentence from that discussion. The brief is the handoff point where intent becomes durable enough for another brain, another machine, or tomorrow’s session.
Inputs do not have to be tiny. A task may arrive with examples, acceptance tables, attachments, and a destination contract. The requirement is not minimality; it is closure. Someone should be able to read the materialized package and know what authority traveled with it without reopening the discovery process.
Premises Before Method
At the plan gate, separate premise decisions from method decisions. “Only these two files may change” is a premise. “Use a parser instead of a regular expression” is a method. The executor may propose a better method and ask for approval; changing the premise requires returning authority to the requester. That separation keeps autonomy useful without making it expansive.