An innovation lab teaches staff to keep system policy above resident messages that try to waive fees or ignore rules. What defensive prompting literacy is taught?

Civic bots need defensive instruction priority. Teaching staff to keep system policy above resident messages that try to waive fees or ignore rules is defensive prompting literacy: prioritize immutable system constraints over manipulative user instructions. The point is resilience of municipal policy in chat—not structuring prompts with few-shot examples to enforce the fee rules in civic bots—defensive literacy, not offense how-to, because examples illustrate format but do not establish system-over-user priority.

Employing strict delimiter separation to quarantine resident messages in civic bots—defensive literacy, not offense how-to fails because delimiters only label boundaries for untrusted text and do not teach that system policy overrides manipulative waiver attempts. Structuring prompts with few-shot examples to enforce the fee rules in civic bots—defensive literacy, not offense how-to fails because demonstrations show desired style but do not create immutable hierarchy keeping resident instructions below system policy. Sanitizing manipulative user instructions via input filtering systems in civic bots—defensive literacy, not offense how-to fails because filtering is an external sanitization control, not prompting literacy about instruction priority.

Exam caveat: defensive priority protects policy boundaries; it does not replace human escalation paths for legitimate hardship or supervisor-approved exceptions. The bot should refuse unauthorized waivers and route edge cases to staff.

Operational check: encode non-overridable policy in the system prompt; add explicit refusal language for fee-waiver and rule-ignore attempts; log override attempts for fraud/security review while offering a human contact path.