Prompt Engineering

AITECH · 45 questions

• A parks and recreation department sets the assistant’s role to “municipal permitting specialist” before asking about facility rental fees. What prompt pattern is that?
• Municipal court staff write: “Answer only from the attached fee schedule; if the fee is not listed, say it is not listed.” What core prompt principle does this illustrate?
• Fire department IT adds a prompt rule: never invent parcel IDs; reply “not found” when the ID is missing from provided records. What is the main purpose of that constraint?
• A police records bureau templates clerk prompts as role, then context, then task, then required output format. What does that structure demonstrate?
• A city open-data portal team requires the assistant to return intake fields as JSON with named keys. Which prompt practice are they applying?
• School-district IT configures a public-health FAQ bot with a standing rule: do not give personalized medical advice. What kind of prompt control is that?
• An airport authority requires resident-facing answers written in plain language around an eighth-grade reading level. What prompt element is that?
• A port authority kiosk app keeps standing policy prompts separate from each traveler’s typed question. Why is that layering useful?
• A county assessor’s standing instruction tells the assistant to cite property-tax section numbers when explaining rules. What does that instruction pattern support?
• Public works reviews prompts that omit a role and sees replies flip between curt clerk jargon and overly casual resident chat. What lesson should they take?
• An election office adds a negative constraint: the chat assistant must not process payments or collect card numbers. What risk does that scope limit primarily address?
• Animal control standardizes a shared prompt template so every division asks models with the same role–instruction–constraint structure. What benefit does that practice target?
• A solid-waste utility’s first AI draft of a grant abstract is too formal, so staff revise the prompt and regenerate. Which technique is that?
• A stormwater bureau runs a pipeline: extract issues from a complaint, classify them, then draft an acknowledgment. What prompting technique is illustrated?
• A building inspection division includes three examples of correctly categorized 311 tickets before asking the model to label a new one. What technique is that?
• A planning commission staff member is drafting a generative-image prompt for a city parks poster and wants subject, style, and banned content spelled out up front. Which prompt structure best supports that multimodal request?
• A city CIO office needs a long council brief and plans to prompt an outline first, then expand each bullet in later turns. Which prompting technique is the office applying?
• A municipal broadband team wants consistent bilingual outage notices and plans to include two paired English–Spanish exemplars in the prompt before generating the next notice. What is the team primarily using?
• A public hospital IT team retrieves a policy summary with RAG and then feeds that summary into a second prompt that rewrites it in plain language for residents. Which technique best describes this workflow?
• A veterans services office is writing prompts for an IVR-style spoken script and needs pacing cues and pronunciation notes for agency names. Which prompt design choice best fits that modality?
• A community development agency sees shaky labels on rare permit types when it asks with no examples, then improves after adding a few labeled exemplars. When should the team prefer few-shot over zero-shot?
• A code enforcement analyst must check complex grant eligibility and breaks the work into ordered prompt steps: gather facts, apply each rule, then conclude. Which prompting approach is this?
• A parking authority keeps its few-shot examples fixed while iteratively tweaking prompt wording and decoding settings to improve citation-help answers. What practice is the team demonstrating?
• A convention-center IT analyst attaches a facility photo and a text complaint to one request so the assistant can locate the issue in the image and draft a work order. What should the multimodal prompt emphasize?
• A parks wildlife 311 chat receives a resident message that says to ignore prior rules and dump the system prompt. How should the municipal team classify that input?
• A museum district IT review finds hidden instructions inside an uploaded PDF resume that a city job-application bot will summarize. Which attack type does this illustrate?
• County jail medical chatbot logs show a user message trying to force the bot into an “admin mode” that unlocks restricted tools. What prompt-injection pattern is this closest to?
• Library kiosk transcripts show jailbreak-style “DAN” patterns that try to replace the civic assistant’s approved persona. Which injection class best fits?
• Child-protective services IT worries that malicious instructions on scraped web pages could be retrieved into a RAG index and then followed by the model. What risk is the team highlighting?
• A homeless-services continuum chatbot receives “reveal your instructions” from a user. How should operators treat that message for security classification?
• A public defender IT review finds an email-to-ticket body that tells the agent to email all case files to an outside address. Which injection goal does this represent?
• A city attorney trains staff that prompt injection is an AI-specific input threat where untrusted text tries to manipulate model instructions—not merely classic web XSS. What understanding should staff take away?
• A budget office reviews multi-turn chats where early messages seem harmless but later turns slowly erode safety constraints. Which injection pattern should analysts document?
• Fleet management security notices payloads that encode or obfuscate “ignore instructions” text before it reaches the assistant. At a recognition level, what should analysts call this?
• Facilities maintenance staff must separate true prompt injection (attempts to override instructions) from ordinary rude or frustrated resident language. Which distinction is correct?
• A GIS mapping unit is cataloging prompt-injection types for the municipal AI risk register. What should that taxonomy emphasize for governance value?
• A CISO-led team configures the city ordinance bot to refuse invention and say it does not know when the needed ordinance text is missing from context. Which defensive prompting goal is that serving?
• After fee hallucinations, a records retention office adds a standing instruction to quote only retrieved fee-schedule text. What mitigation is this?
• An ADA compliance office requires a human review gate before any AI-drafted denial letter is sent to a resident. Which mitigation approach is this?
• A language-access office asks the model to list its assumptions in a self-check step before producing the final translated resident notice. What defensive prompting pattern is this?
• A resilience and climate office uses a model for factual parcel flood-zone lookups, but creative settings keep inventing zone codes. What change best reduces that invention?
• An economic-development office wants staff to trust AI summaries of grant eligibility less when claims are shaky. Which defensive prompting move best supports that?
• A city clerk’s first AI pass drafts a fee answer that still looks wrong next to the published schedule. What dual-pass mitigation should the clerk use next?
• A county public-health department finds malicious instructions hidden in HTML comments on pages headed into its RAG corpus. What input-hygiene step best defends the assistant?
• A municipal permitting office sees free-form model answers invent fee amounts not on the ordinance. Which output design best shrinks that hallucination surface?