GPT-5.6 Prompt Generator

Perform a comprehensive threat model for the following system using the STRIDE framework. Identify every realistic attack surface, assign a risk rating (Critical / High / Medium / Low) to each threat, and propose a specific mitigation for each. System: [describe your system — e.g. 'A B2B SaaS API that accepts webhook payloads from third-party partners, processes them in a job queue, and writes results to a PostgreSQL database.'] For each threat: 1. THREAT — what the attacker does 2. CATEGORY — which STRIDE element applies 3. RISK — Critical / High / Medium / Low with one-sentence justification 4. MITIGATION — the specific technical control, not a category End with: the three highest-priority mitigations to implement this sprint.

Perform a senior-engineer architecture audit on the following codebase. Identify structural problems, not style issues. Language/stack: [your stack, e.g. 'Node.js 22, TypeScript, Express, PostgreSQL, Redis'] Files / excerpts: ``` [Paste your most architecturally important files — entry points, service layers, database access patterns] ``` Audit dimensions: 1. COUPLING — where are modules too tightly coupled? What breaks if one changes? 2. DATA LAYER — N+1 queries, missing indexes, unsafe transactions 3. ERROR HANDLING — where can exceptions propagate silently? 4. SCALABILITY CEILING — what breaks first at 10x current load? 5. SECURITY SURFACE — injection vectors, auth gaps, secrets handling Deliver a ranked list of findings, each with: the exact file + line range, why it matters, and the minimum-change fix.

Synthesise the following research papers into a structured evidence brief for a non-specialist policy audience. Use max reasoning — think through every claim before including it. Papers: [paste abstracts and key findings, or full text] Topic: [your topic, e.g. 'Effectiveness of urban green corridors on ambient temperature reduction'] Audience: [e.g. 'City council members with no scientific background'] Deliver: 1. CONSENSUS — claims that ≥3 papers support with consistent evidence 2. CONTESTED — findings where papers reach different conclusions, with the reason for disagreement 3. CONFOUNDERS — variables none of the papers controlled for that a reader might wrongly ignore 4. POLICY TRANSLATION — three specific, evidence-backed recommendations in plain language 5. UNCERTAINTY LEVEL — Low / Medium / High for each recommendation, with a one-sentence justification Cite the paper (author + year) for every factual claim.

Activate ultra mode. Decompose the following project into parallel workstreams and assign each to a sub-agent with a specific, bounded task. Identify dependencies between workstreams and flag which can run simultaneously vs. which must sequence. Project: [describe your project, e.g. 'Launch a self-serve API key management portal — users can create, rotate, and revoke API keys; usage is tracked per key; billing integrates with Stripe; admin dashboard shows top consumers'] For each workstream: - NAME and SCOPE (what exactly does this sub-agent own?) - INPUT it needs from other workstreams (if any) - OUTPUT it produces - DEPENDENCY — blocks/blocked by which other workstream? - ESTIMATED COMPLEXITY: S / M / L End with: a sequenced execution plan showing the critical path.

Write a complete customer support playbook for the following product. Structure it so a new support agent can handle the top 10 issues without escalation. Product: [describe your product in 2-3 sentences] Top 10 issues (list them or describe your category): [e.g. 'billing disputes, password reset, API rate limits, feature requests, integration failures, refund requests, data export, account deletion, downgrade questions, slow response times'] For each issue: 1. TRIGGER — what the customer says that signals this issue 2. DIAGNOSIS — the 2-3 questions to ask to confirm the issue 3. RESOLUTION STEPS — numbered, specific, with no assumed product knowledge 4. ESCALATION TRIGGER — the exact condition that requires escalation 5. TEMPLATE RESPONSE — a ready-to-send reply the agent can personalise in under 30 seconds Tone: [professional and warm / technical / casual]

Analyse the following document and produce a structured executive brief for a time-pressed C-level reader. Document: [paste your contract, report, policy, proposal, or transcript] Deliver: 1. ONE-PARAGRAPH SUMMARY — what this document is and why it matters (under 120 words) 2. KEY COMMITMENTS OR OBLIGATIONS — bullet list of specific things each party must do, with deadlines where stated 3. RISKS — three things a careful reader would flag, ranked by severity 4. OPEN QUESTIONS — what is ambiguous, undefined, or left to interpretation? 5. RECOMMENDED ACTIONS — what should the reader do within the next 48 hours as a result of reading this document? Do not paraphrase — cite the specific clause, section, or page for every point.

Write a complete Product Requirements Document for the following internal tool. Optimised for a small engineering team (2-4 engineers) to build and ship in under 4 weeks. Tool: [describe the internal tool, e.g. 'A Slack bot that lets sales reps query a customer database by company name and get a deal health summary — stage, last activity, open tasks, ARR — without leaving Slack'] Primary users: [who uses it daily] Success metric: [one measurable outcome — e.g. 'Sales reps spend < 30 seconds getting deal context before a call'] Include: 1. Problem statement (2 paragraphs) 2. Proposed solution (what, not how) 3. 5 user stories 4. Acceptance criteria for each story 5. Out of scope (3 items) 6. Open questions (3 items) 7. Suggested MVP scope for week 1

Extract structured data from the following unstructured text and return it as valid JSON that matches the schema below. If a field cannot be reliably inferred, set it to null — do not guess. Schema: ```json { "company_name": "string", "funding_round": "string (e.g. Series A)", "amount_usd": "number or null", "lead_investor": "string or null", "co_investors": ["array of strings"], "announced_date": "YYYY-MM-DD or null", "valuation_usd": "number or null", "use_of_funds": "string or null" } ``` Text to extract from: [Paste your press release, article, or document here] Return only the JSON object, no explanation.

Perform a senior engineer code review on the following pull request. Focus on correctness, maintainability, and security — not style. Language/framework: [e.g. 'Python 3.12, FastAPI, SQLAlchemy'] Diff or changed files: ``` [Paste your code changes] ``` Review checklist: 1. CORRECTNESS — logic errors, edge cases, off-by-one errors, null/undefined handling 2. SECURITY — injection, auth bypass, insecure defaults, secrets in code 3. PERFORMANCE — N+1 queries, unnecessary loops, blocking calls in async context 4. ERROR HANDLING — what happens when this fails? Is it recoverable? 5. TESTABILITY — can this code be unit tested without heavy mocking? For each finding: file + line, what the issue is, why it matters, and the minimal fix.

Write a competitive analysis brief for the following market. Our product: [one-sentence description] Our target customer: [describe the buyer] Competitors: [list 4-6 specific company/product names] For each competitor: - PRIMARY VALUE PROP (one sentence) - PRICING MODEL (free / freemium / subscription / enterprise — with estimates if known) - STRONGEST CAPABILITY (what they do better than anyone) - WEAKEST POINT (what customers complain about most) - RECENT MOVES (any launch, pivot, or acquisition in the last 6 months — mark 'unknown' if not) Close with: 1. The white space — a positioning angle none of them currently own 2. The competitor we should fear most, and why 3. The one thing we must do differently to win in this market

Summarise the following in exactly 5 bullet points. Each bullet must be a standalone sentence — no sub-bullets. Maximum 20 words per bullet. Do not add a header or preamble — start immediately with the first bullet point. [Paste your text, article, meeting transcript, or document here]

Write a professional email based on the following brief. Maximum 150 words. No filler openers ('I hope this finds you well'). Start with the key point. Context: [who you are, who you're writing to, your relationship] Goal: [what you need from this email — a meeting, a decision, a reply] Key information: [the one or two facts the recipient needs to know] Tone: [formal / warm-professional / direct] Return: subject line + email body only.

Extract all action items from the following meeting transcript. Return only a structured list — no summary, no preamble. Format for each action item: - OWNER: [person's name] - ACTION: [specific thing they must do, starting with a verb] - DEADLINE: [date or 'not specified'] - CONTEXT: [one sentence explaining why this matters] Meeting transcript: [Paste your transcript here]

Write 5 headline variants for the following page/ad for A/B testing. Each variant must try a different persuasion angle. Do not repeat the same structure across variants. Product/page: [describe what the page or ad is for] Target audience: [describe who will read it] Primary keyword (must appear in at least 2 variants): [your keyword] Goal: [clicks / signups / purchases] For each headline, add a one-word label for the angle used (e.g. Urgency / Social Proof / Curiosity / Outcome / Fear of Missing Out).

Write a step-by-step script for automating the following routine task. Assume the person running it has basic computer skills but no programming experience. Each step must be a single, concrete action — no compound steps. Task: [describe the manual process, e.g. 'Every Monday morning I download a CSV from our analytics dashboard, paste the key numbers into a Google Sheet, write a short summary for the team Slack, and add a screenshot of the chart to a shared folder.'] Deliver: 1. The step-by-step script (numbered, under 15 steps) 2. The tools that could automate each step (name specific apps, not categories) 3. The step with the highest automation ROI — where to start if you only automate one thing

Translate this SQL query into plain English, then answer: is this query doing what the author intended? If not, what is it actually doing? Database context (optional): [table names and what they contain] Query: ```sql [Paste your query] ``` Return: 1. PLAIN ENGLISH — one paragraph explaining what the query does 2. INTENT CHECK — yes it does what it intends / no, here's what it actually does 3. FIX (only if needed) — the corrected query with the changed lines highlighted in a comment

Write a 2,000-word research-backed article on the following topic. Use max reasoning to ensure every claim is defensible and the argument flows logically. Topic: [your topic] Primary keyword (must appear in H1, first sentence of intro, and at least 3× in body): [keyword] Audience: [describe reader expertise level and why they care] Stance/angle: [the specific argument or perspective this article takes] Structure: - H1 with primary keyword - 150-word intro that names the reader's problem, states your position, and promises a specific takeaway - 5 H2 sections, each with a concrete example (not hypothetical) - Data or statistics cited in at least 3 sections - 100-word conclusion with one specific action step Do not use filler phrases or generic advice. Every paragraph must advance the argument.

Design an optimal prompt caching strategy for the following GPT-5.6 API use case, using explicit cache breakpoints. Use case: [describe your application, e.g. 'A customer support chatbot that has a 12,000-token system prompt with product documentation. Each conversation averages 8 turns. We process ~5,000 conversations per day.'] Deliver: 1. WHERE to place cache breakpoints — identify the exact boundaries in the prompt (which part is static vs. dynamic) 2. EXPECTED CACHE HIT RATE — estimate based on your conversation pattern, with reasoning 3. COST CALCULATION — estimated savings per day vs. no caching (use GPT-5.6 Terra pricing: $2.50 input / $15 output per MTok) 4. IMPLEMENTATION GUIDE — the specific API parameter changes needed to activate caching 5. PITFALL — the most common mistake that kills cache hit rate in this pattern

Write a complete creative brief for the following brand storytelling project. Brand: [company name and what it does] Campaign goal: [one measurable outcome, e.g. 'Drive 20% increase in free trial signups from LinkedIn'] Target audience: [describe in detail — job title, company size, pain point, what they read] Key message: [the one thing you want them to remember after seeing this content] Tone: [e.g. 'Confident but not arrogant. Data-driven but human.'] Format: [e.g. '5-part LinkedIn article series'] The brief must include: 1. Campaign concept (one paragraph — the big idea) 2. Messaging hierarchy (primary / secondary / tertiary messages) 3. Content outline for each piece 4. Do's and Don'ts for the writer/designer 5. Success metrics and how to measure them

Review the following system design for scalability and identify the bottlenecks that will appear first under 10x, 100x, and 1000x load growth. Current system: [describe your architecture — services, databases, caches, queues, CDNs, deployment model] Current scale: [e.g. '50K daily active users, 2M API requests/day, 500GB database'] Growth trajectory: [expected load growth over next 12 months] For each load tier (10x / 100x / 1000x): 1. FIRST BOTTLENECK — the exact component that breaks first 2. SYMPTOM — how the failure manifests (errors, timeouts, cost spike) 3. SOLUTION — specific architectural change (not 'add more servers') 4. COST ESTIMATE — rough order of magnitude for the solution Close with: the single most important architectural change to make before you hit 10x, and why it's harder to fix later.