Ready-to-use prompt library for ChatGPT, Claude, and Gemini. Professional writing, meetings, data analysis, code, and brainstorming.
Mastering these techniques is worth more than memorizing 100 prompts.
Add "Think step by step before answering" or "Show your reasoning"
"Calculate the campaign ROI. Think step by step and show each calculation."Define the role before the task: "You are a [specialist]. Analyze..."
"You are an experienced CFO. Evaluate this investment proposal with healthy skepticism."Show 2–3 input→output examples before your real task
Input: "item arrived broken" → Category: Logistics | Input: "I want to cancel" → Category: Cancellation | Input: [your text]Specify exactly what you want: structure, length, tone, what NOT to include
"Respond in JSON with fields: title, summary (max 50 chars), tags (array). Do not include explanations outside the JSON."Explicitly tell the model what not to do
"Write the email. Do NOT use: 'Hope this finds you well', passive voice, or more than 3 paragraphs."Emails, reports, proposals, and corporate communications
When you need to give negative feedback or decline something
Write a professional, assertive email to [person/role] communicating that [difficult situation: proposal rejection / delay / negative feedback]. The tone should be direct but respectful, without excessive apologies. Include: (1) objective opening, (2) main message without hedging, (3) clear next steps. Maximum 150 words.
Reports, contracts, or articles that need to be distilled for leadership
Read the text below and create an executive summary of no more than 200 words for a senior leadership audience without technical expertise. Structure it as: (1) Context in 1 sentence, (2) 3 main points in bullets, (3) Recommendation or next step. Prioritize business implications over technical details. [Paste the document here]
Proposal for a client with known objections
Create a commercial proposal for [client/segment] offering [product/service] at [price]. The client has the following objections: [list the objections]. The proposal should: address each objection with data or argument, highlight 3 measurable tangible benefits, include legitimate urgency (not artificial), and end with a clear CTA. Tone: confident and consultative, not pushy.
You have the ideas but the text is raw
Rewrite the text below preserving 100% of the information and my voice, but improving: clarity, paragraph structure, grammar, and flow. Do not add information that is not in the original. Do not make the text more formal than the original. If you find ambiguities, ask before inventing. [Your draft here]
Agendas, meeting notes, follow-ups, and time management
After a meeting, turn raw notes into formal minutes
Turn the notes below into professional meeting minutes with: Date and attendees, Points discussed (concise bullets), Decisions made (bold), Next steps with owner and deadline (table), and Open items. Be objective and factual — do not invent missing information, mark it as [TO CONFIRM]. [Paste your notes here]
An important meeting that needs a clear outcome
Create an agenda for a [duration] meeting on [topic/objective]. The meeting should end with [expected decision/deliverable]. Attendees: [list attendees and their roles]. Structure with: start time for each item, person responsible for leading, and what needs to be prepared in advance. Reserve the last 10% of the time for next steps.
After a meeting, ensure commitments are kept
Write a post-meeting follow-up email on [topic] with attendees [names]. The commitments made were: [list them]. The tone should be friendly but create accountability without being overbearing. Include: 3-line summary, action table (what/who/when), and an offer of support. The email subject line should be specific, not generic.
Interpreting results, research, and synthesizing information
You have a table or CSV and need insights, not raw numbers
Analyze the data below and tell me: (1) The 3 main patterns or trends, (2) Anomalies or outliers that deserve attention, (3) A hypothesis to explain the most relevant pattern, (4) A question these data do not answer but would be important to investigate. Avoid repeating the raw numbers — focus on interpretation. [Paste your table or data here]
You need a fast overview of a market or topic
Provide an analysis of [market/sector] with: estimated market size, main players and their positions, trends for the next 2 years, relevant barriers to entry, and an unexplored niche opportunity. Be specific and data-grounded. Cite limitations of what you know.
You need to defend a position or stress-test an idea
My position: [your position here]. First, present the 3 strongest arguments IN FAVOR of this position with evidence. Then, present the 3 strongest counterarguments from an intelligent critic. Finally, tell me which counterargument is hardest to rebut and how I could respond to it. Do not take sides — be an efficient devil's advocate.
You have read several articles or reports and need a coherent synthesis
Synthesize the texts below identifying: (1) What all sources agree on, (2) Where there is disagreement and why, (3) What is missing — perspectives not covered by any source, (4) Your most important conclusion considering the full set. If there are contradictions, point them out explicitly. Do not cite sources by number — integrate the ideas. [Paste the texts here]
Debugging, documentation, review, and code generation
Code with unexpected behavior
My code has the following issue: [describe the bug — what you expected vs what happens]. Context: [language, framework, version]. Relevant code: ``` [paste the code] ``` Before suggesting a fix, list your 3 most likely hypotheses for the root cause, with a probability for each. Then fix based on the most likely hypothesis and explain why the solution works.
Before merging or deploying
Review the code below focusing on: (1) Bugs or edge cases that could cause production issues, (2) Performance problems at high volume, (3) Security issues (injection, data exposure, authentication), (4) Readability and maintainability. Do NOT suggest aesthetic refactors — only what impacts production, performance, or security. ``` [paste the code] ```
Code with no documentation or outdated documentation
Generate technical documentation for the function/module below. Include: purpose in 1–2 sentences, parameters with type and description, return value, usage examples (happy path and edge cases), and known limitations. Write for a developer who does not know the code but knows the language. ``` [paste the code] ```
Idea generation, naming, concepts, and problem solving
When "any idea" generates useless suggestions
Generate 10 ideas for [objective]. Mandatory constraints: (1) [constraint 1, e.g. budget < $5,000], (2) [constraint 2, e.g. team of 2 people], (3) [constraint 3, e.g. implementable in 30 days]. For each idea: name, 1-line description, why it respects the constraints. Do not filter by originality — include both obvious and non-obvious ideas. Then identify which has the most potential and why.
A product, project, company, or feature that needs a name
Create 15 name options for [what it is: product/service/project] that [what it does] for [target audience]. Criteria: [e.g. memorable, max 2 syllables, .com domain available, no confusion with [competitor]]. Organize into 3 groups: descriptive names, evocative names, and invented names. Indicate which you would choose and why.
A complex problem that needs a fresh perspective
My problem: [describe the problem clearly]. Solve it using analogies from 3 different domains: (1) How would a software engineer solve it?, (2) How would a chef solve it?, (3) How would a military general solve it? For each analogy: what is the applicable principle and how does it translate to my problem. At the end, synthesize the most useful insight from all three.
Generic prompts produce generic results. The secret is adapting templates with real context. For each prompt in this library, the most efficient cycle is:
For critical tasks (important communications, analyses that will be presented), always review the output as you would the work of a competent intern: are the facts correct? Is the tone right? Did the model invent or over-generalize anything? AI is an accelerator, not a substitute for your judgment.