Start
There is no slash command — describe the experiment:
text
Design a two-arm randomized experiment to test whether a 4-week HRV biofeedback program lowers perceived stress (PSS-10) in university students compared with an active control. Ground it in how related studies were designed and include participants, measures, protocol flow and a power analysis.
The design starts immediately — there is no survey. Anything the request leaves open (participants, the primary measure, the manipulated levels, the setting) is settled from what the reference studies typically do; those assumptions are listed in the design basis and in the first line of the chat summary, so you can redirect them in a follow-up turn. Put the constraints you care about in the request itself.
What you get
<topic>_experiment_design.md— one-line summary, research objective, hypotheses, design, participants and inclusion criteria, conditions, measures, procedure, analysis plan, power analysis, ethics and limitations.<topic>_reference_designs.md— how the closest published studies were designed, with citations.<topic>_protocol_flow.svg— recruitment → screening → randomization → intervention → follow-up.<topic>_power_curve.svg— required N across plausible effect sizes.<topic>_expected_results.svg— what the primary result would look like if the hypothesis holds.
Refine
Ask for changes in plain language: "make it a crossover", "add a 3-month follow-up", "use PSS-4 instead". The report is revised as a new version; the figures are regenerated to match.
Walkthrough: Designing the experiment.