A municipal purple-team checklist must run before any AI-assisted malware-sample prompt in the city range. Which gate set best matches ethical lab practice?

Before any AI-assisted malware-sample prompt runs in a municipal purple-team range, ethical lab practice requires a complete pre-prompt gate set: documented purpose, approved rules of engagement, confirmed isolation of the lab environment, named human review of outputs, and a plan for post-test destruction of generated artifacts. Purpose ties the sample to a legitimate civic exercise or detection-development goal rather than curiosity. ROE bounds what techniques, targets, and persistence are allowed. Isolation keeps generated payloads off production court, school, or city networks. Human review catches unsafe, over-broad, or policy-violating content before it is used. Destruction prevents orphaned samples from lingering on shared disks. Skipping written purpose because an engineer ran similar prompts last quarter treats prior informal practice as authorization and breaks auditability. Requiring only network isolation while treating ROE, review, and destruction as optional after-hours notes leaves the most important accountability controls unenforced. Prompting first and documenting purpose and ROE only if leadership later asks for an audit package inverts the control sequence and creates after-the-fact paper that cannot prevent misuse. Exam caveat: GOAA expects municipal defenders to sequence AI malware-generation work behind governance gates, not treat generative convenience as a substitute for purple-team authorization. Operational check: refuse the prompt until the checklist is signed, the VLAN or air-gapped range is verified, reviewers are named, and wipe procedures are scheduled; log the gate completion with the exercise ticket.