Oct 6 · rewritten Build Your Own Prompt Kit A Text-Based Guide to Prompt Engineering for Work and Life Start Here — Who This Is For and How to Use This Kit Who this is for. You have used ChatGPT, Claude, Gemini, or Copilot a handful of times. You can get an answer, but not reliably a good answer. You want to stop re-inventing your instructions every time and instead build a small personal library of prompts that work—for your job and for your life at home. You do not need to code or know any technical jargon. Everything here happens in an ordinary chat window. What is prompt engineering? It just means giving clear instructions to an AI. A prompt is the instruction you type. A prompt kit is your personal library of your best, tested instructions—saved, named, and ready to reuse. What you will be able to do after this kit: Build, test, and refine reusable prompts for any repeat task. Explain why a prompt works so you can adapt it. Diagnose and fix a prompt when the output goes wrong. Your 10-Minute Quick Win This exercise gives you a useful weekly plan in under ten minutes and teaches the whole framework by doing it. Copy this into a chat window, replace the bracketed parts with your own words, and send: ROLE: You are a concise organizing assistant helping me plan my week. CONTEXT: I am [YOUR ROLE, E.G., AN OFFICE MANAGER AND PARENT]. My top 3 priorities this week are: [PRIORITY 1, 2, AND 3]. My available time is: [E.G., MON-FRI, 9AM-3PM]. TASK: Turn my priorities into a simple Monday-to-Friday plan with one top action per day. CONSTRAINTS: Under 180 words. Plain language. If something is unclear, still give your best plan now, then add a line starting with "Question:" at the end. OUTPUT FORMAT: A table with columns: Day Top Action 10-min Starter Step. MODE: Just do it — give your best answer now. If the result is about 80% right, reply in the same chat: Make it shorter and more blunt. That is iteration—refining within one conversation—and it is the core habit this kit builds. Part 1: The Five Building Blocks — R-C-T-C-O Remember it as: Really Clear Tasks Create Output. ROLE — The perspective to adopt ROLE tells the AI what kind of helper to be. Think job title plus audience. Work Example: You are an experienced project manager who writes for executives. Personal Example: You are a friendly meal planner writing for a busy household. CONTEXT — Background the AI cannot know CONTEXT is everything the AI cannot guess: your situation, your audience, and your goal. When you paste material, wrap it in clear markers called delimiters (e.g., --- START NOTES ---) so the AI can tell your instructions apart from the material. Work Example: Audience: VP of Sales. Goal: Decide Q4 hiring. Raw notes: --- START NOTES --- [PASTE HERE] --- END NOTES --- Personal Example: We are a family of 4. One vegetarian. 30 minutes max on weeknights. Budget is $120/week. TASK — The one clear action TASK is the single job to do. Use one main action verb: summarize, compare, draft, rank, or plan. Work Example: TASK: Summarize these notes into Decisions, Open Questions, and Next Steps. Personal Example: TASK: Create a 5-dinner plan and a combined grocery list. CONSTRAINTS — Boundaries and uncertainty rules CONSTRAINTS are the limits: length, tone, and—most importantly—how to handle missing info. Work Example: 1) Max 200 words. 2) No jargon. 3) Use only the notes provided; write TBD if a date is missing. Personal Example: 1) No recipes over 40 minutes. 2) No mushrooms. 3) If info is missing, write TBD and still give your best plan. OUTPUT FORMAT — The shape of the response OUTPUT FORMAT is the structure: table, bullets, headings, or checklist. Work Example: Table with columns: Decision Owner Due Date. Personal Example: Table by day: Day Meal Prep Time. Then a grocery list grouped by store section. Part 2: Examples of Great Prompt Design Here are five high-quality prompts that apply the R-C-T-C-O framework. Use these as templates for your own kit. The "Executive Briefer" (Work) Use this to turn long documents or messy meeting notes into a one-page summary. ROLE: You are a senior business analyst known for brevity and clarity. CONTEXT: I am presenting to a busy executive who needs to make a decision on [INSERT PROJECT/TOPIC]. --- START MATERIAL --- [PASTE DOCUMENT TEXT OR NOTES HERE] --- END MATERIAL --- TASK: Synthesize the material into a one-page briefing. CONSTRAINTS: 1) Maximum 250 words. 2) Use only facts from the provided material. 3) If evidence is missing for a point, write [NEEDS DATA]. 4) Use a professional, neutral tone. OUTPUT FORMAT: Executive Summary (3 sentences) Key Findings (3-5 bullets) Recommended Next Step MODE: Just do it — flag any major gaps at the end. The "Tough Conversation Coach" (Work/Personal) Use this to draft emails or scripts for difficult situations. ROLE: You are an expert in workplace communication and interpersonal conflict resolution. CONTEXT: I need to [E.G., ASK FOR A DEADLINE EXTENSION / ADDRESS A PERFORMANCE ISSUE] with [NAME/ROLE]. My goal is to be firm but maintain a positive relationship. TASK: Draft a 3-paragraph email and 3 talking points for a follow-up call. CONSTRAINTS: 1) Tone: Professional, empathetic, and solution-oriented. 2) Avoid: Apologetic language or "I'm sorry" unless a mistake was actually made. 3) Length: Under 150 words for the email. OUTPUT FORMAT: Email Draft 3 Bulleted Talking Points MODE: Ask first — ask me 2 questions about the specific situation before you draft. The "Learning Roadmap" (Personal/Work) Use this to learn a new skill with a limited schedule. ROLE: You are a patient, expert tutor who specializes in accelerated learning. CONTEXT: I want to learn [SKILL, E.G., BASIC EXCEL PIVOT TABLES] for [PURPOSE]. I have [NUMBER] hours per week to study. TASK: Create a 4-week learning roadmap. CONSTRAINTS: 1) Prioritize "high-leverage" concepts (the 20% of info used 80% of the time). 2) Include one "mini-project" per week. 3) Suggest only free or low-cost resources. OUTPUT FORMAT: Week [X]: [Topic Name] Key Concept to Master Practice Exercise Success Criteria (How I know I've learned it) MODE: Just do it. The "Unbiased Decision Matrix" (Personal/Work) Use this when choosing between multiple options. ROLE: You are a logical decision-support assistant. CONTEXT: I am choosing between [OPTION A] and [OPTION B] for [PURPOSE]. My top priorities are [E.G., COST, SPEED, QUALITY]. TASK: Compare these options objectively to help me decide. CONSTRAINTS: 1) Do not pick a winner; stay neutral. 2) Highlight "hidden costs" or trade-offs for each. 3) If a price or feature is unknown, write [VERIFY]. OUTPUT FORMAT: A table comparing both options against my priorities, followed by 3 "Hard Questions" I should ask myself to break the tie. MODE: Just do it. The "Household Operations Manager" (Personal) Use this for complex family logistics. ROLE: You are a highly organized personal assistant and logistics expert. CONTEXT: We have a busy week. Events: [LIST EVENTS, E.G., SOCCER TUESDAY 5PM, DOCTOR THURSDAY 9AM]. Goals: [E.G., MEAL PREP, GET CAR WASHED]. TASK: Create a daily "Run of Show" for the week. CONSTRAINTS: 1) Include "buffer time" (15 mins) between transitions. 2) Flag potential "crunch points" where two things overlap. 3) Keep instructions simple enough for a teenager to follow. OUTPUT FORMAT: A daily schedule (Mon-Sun) with a "Top Mission" for each day and a list of "Must-Pack" items for the morning. MODE: Just do it. Part 3: The Kit-Building Method — 4 Steps Step 1. Pick a repeatable job Choose a task you do at least monthly that has a recognizable "good" result (e.g., status updates, meal plans, trip itineraries). Step 2. Draft using the skeleton Fill in the R-C-T-C-O blocks. Keep the fixed parts as plain text and mark the parts that change each time as [VARIABLES]. Step 3. Test and "Break" it Run your prompt with three different inputs: one easy, one typical, and one messy (vague or missing info). If the AI invents facts during the messy test, your CONSTRAINTS block needs to be stronger. Step 4. Refine and Save Change one thing at a time. If the output is too long, add a word count. If it's too dry, change the ROLE. Once it works, save it in a notes app or document. Part 4: Troubleshooting Your Prompts Symptom Likely Fix Output is too long or rambling Add a word count constraint (e.g., "Max 150 words") or a specific structure (e.g., "Use 3 bullet points"). AI is "hallucinating" (making things up) Add an uncertainty rule: "Use only the provided text. If info is missing, write TBD." Tone is wrong Be more specific in the ROLE. Instead of "expert," try "blunt executive" or "supportive coach." It ignores a specific instruction Move that instruction to the very top or very bottom of the prompt. Number your constraints 1, 2, 3. It gives a different format every time Use an example. Paste a sample of exactly what you want the output to look like between --- EXAMPLE --- markers. Part 5: Safety and Maintenance Privacy. Never paste sensitive data like account numbers, passwords, or confidential client names. Use placeholders like [CLIENT A]. Verification. AI is a language model, not a calculator or a fact-checker. Always verify dates, prices, and math manually. Update your kit. Models change. Every few months, run your saved prompts to ensure they still produce the quality you expect. If you switch from ChatGPT to Claude (or vice versa), re-test your kit immediately.