Separate evidence from machinery
Operational records can help describe realistic tasks and states. An environment still needs interfaces, state transitions, tools and checks. The historical record may not contain enough information to reproduce every action safely or faithfully.
Specify the success test
Define what completion means and which evidence supports it. Guard against shortcuts that satisfy a superficial check while failing the real task. Keep evaluation examples distinct from training material when the study requires a held-out test.
Disclose the simulation boundary
State which behaviors are simulated, omitted or simplified. A successful benchmark run does not prove safe performance in the original business. Consider rights, privacy and operational risk throughout environment construction and testing.
Practical guidance, not legal advice or a buyer specification. Examples are illustrative. Reviewed October 20, 2026.