A question answers from this run's files and changes nothing. An alternate plan is written into the run beside the main hypothesis — it never changes the main hypothesis, the protocol or round 1, whose rules were fixed before any result existed. A protocol change edits the bench copy only: the vessel it is laid out for, a value of yours (checked against the project, marked D with your name), a note in your words (marked P), or a section left out; then export it as Word or PDF below. "Design past the apparatus" writes the experiment that would get closest to the aim if the vessel were not the limit — a concept beside round 1, never runnable here or approved; the aim, the QC limits and safety bounds are kept. "Write the run up" has the communication agent write the plain-language layer of the final report — an executive summary and what each stage looks like at the bench; every number in it must already stand in the protocol, and every value still prints from the protocol's own rows. None of these searches again; a follow-up run gathers new evidence.