Create a practical threat model for the current mission. Begin with the business purpose, sensitive assets, users, administrators, external systems, trust boundaries, entry points, and data flows.

Identify credible misuse cases, including unauthorized access, privilege escalation, data leakage, prompt injection, malicious files, unsafe tool execution, dependency compromise, denial of service, audit-log tampering, and accidental operator error. Do not invent vulnerabilities that the evidence does not support.

For each credible risk, state the evidence, likelihood, business impact, existing controls, control gaps, recommended mitigation, verification method, and responsible human decision. Rank actions by practical risk reduction rather than dramatic wording.

End with residual risks and explicit go/no-go decisions that require Pilot in Command approval.