- Least to Most
- Iniciar con la guia mas basica, y gradualmente incrementar la complejidad o especificidad de los prompts en base a las respuestas del modelo.
- Esto es util, en especial para explorar un modelo incrementalmente
- Scenario:
You are developing an AI-powered tutoring system designed to help students learn complex mathematical concepts incrementally. The "Least to Most" prompting technique is utilized to gradually guide students from understanding basic concepts to solving more complex problems, ensuring they build a solid foundation before advancing.
- Prompt Highlight: Start simple follow with more complex examples…
- Full Prompt: Imagine you are an AI math tutor assisting a student who is struggling with the concept of quadratic equations. Start with a simple explanation of what a quadratic equation is, followed by progressively more complex examples and questions, adjusting the difficulty based on the student's responses
- Scenario:
You are developing an AI-powered tutoring system designed to help students learn complex mathematical concepts incrementally. The "Least to Most" prompting technique is utilized to gradually guide students from understanding basic concepts to solving more complex problems, ensuring they build a solid foundation before advancing.
- Meta Prompting
- Igual que el Least to Most
- Scenario:
Meta prompts are designed to help AI systems understand the context or framework within which they should operate. They provide guidelines and set the stage for the AI to follow a specific path of reasoning or behaviour, offering more context than normal prompts.
- Prompt Highlight: Explain as if I’m
- Full Prompt: Explain the concept of photosynthesis to a 7th-grade student, focusing on the importance of sunlight and keeping the explanation simple and engagin
- Scenario:
Meta prompts are designed to help AI systems understand the context or framework within which they should operate. They provide guidelines and set the stage for the AI to follow a specific path of reasoning or behaviour, offering more context than normal prompts.
- Igual que el Least to Most
- ART Prompting
- Definir explicitamente que herramientras y aspectos a considerar mientras se analiza el prompt y se renderea la respuesta
- Esto es util cuando se neceita una herramienta en especifico para la respuesta.
- Scenario:
You need to estimate or assess certain processes.
- Prompt Highlight: Assess … consider … aspects.
- Full Prompt: Assess the impact of data centers on the ecology, considering their energy consumption, the source of their energy, and typical CO2 emissions associated with their operation.
Last active
March 7, 2026 20:10
-
-
Save Whistler092/93c51283f07478fec8ca4d5652d43ba4 to your computer and use it in GitHub Desktop.
Prompts
- Chain of Thought
- Motiva al modelo a pensar en voz alta, mostrando su trabajo en un step by step.
- Scenario: Tu quieres que la AI resuelva un problema de matematicas y expliqe cada paso de el proceso para asegurarse es entendible para que los estudiantes aprendan algebra
- Prompt Highlight: Explain your solution line by line / Explain your solution as you go.
- Full Prompt: Imaginate que eres un tutor de matematicas. Los estudiantes te preguntan que resuelvas la ecuacion [equacion] y explique cada paso de su rasonamiento claramente. Porfavor proveea una explicacion detallada a medida de que resuelves la ecuacion.
- Contrastive Chain of Thought
- Le pide al modelo considerar diferentes perspectivas o opciones antes de concluir,
- Haciendo que la desicion sea clara
- Scenario: You're developing an AI tool that helps product designers evaluate different design choices for an ecofriendly backpack. The AI needs to weigh the pros and cons of using natural fiber versus synthetic materials.
- Prompt Highlight: Analyze, compare and contrast. Focus on
- Full Prompt: Imagine you are a product design consultant specializing in eco-friendly materials. A client is considering whether to use natural fiber or synthetic materials for their new line of backpacks. Please analyze both options, comparing their benefits and drawbacks**, particularly focusing on** durability, cost, and environmental impact
- Thread of Thought
- Similar al CoT, pero esta mantiene un hilo narrativo o logico a lo largo de la interaccion
- esto para poder mantener la coherencia en las respuestas.
- Se inicia con la presonificacion, despues la historia desde donde se quiere iniciar, tareas especificas y el final. Especificar el tipo de narrativa esperada.
- Scenario:
You're creating an AI-driven educational tool aimed at helping high school students understand the historical significance of major world events. The focus is on the fall of the Berlin Wall and its implications for global politics.
- Prompt Highlight: Event context, discuss and reflect
- Full Prompt: Imagine you are a history teacher explaining the fall of the Berlin Wall to a high school class. Start from the end of World War II, outline the events leading up to the fall of the wall, discuss the immediate consequences, and reflect on its long-term impact on global politics. Provide a clear, connected narrative to help students understand the sequence and significance of these events.
- Program of Thought
- Sets una secuenci de tareas cognitivas para la AI a seguir.
- estructurando su proceso de generacion de respuesta.
- Scenario:
You are creating an AI-driven assistant for software developers tasked with debugging and improving code. The AI must guide users through a systematic process of understanding, diagnosing, and resolving common programming erros.
- Prompt Highlight: Guide through a program of thought
- Full Prompt: Imagine you are a programming mentor tasked with helping a junior developer debug a Java program that throws a NullPointerException. Guide them through a step-by-step program of thought to identify the likely causes and propose solutions. Ensure each step encourages critical thinking and understanding of Java's object references
- Chain of Verification
- Se asegura la exactitud y confiabilidad de las respuestas
- Teniendo un modelo que verifique cada paso o afirmacion dentro de su razonamiento.
- Scenario:
You're developing an AI system for a fact-checking tool that helps journalists verify claims in political speeches or news articles. The tool needs to provide a step-by-step verification of facts and sources to ensure accuracy and reliability.
- Prompt Highlight: Verify facts, accuracy, cite sources, provide confidence level
- Full Prompt: Imagine you are an AI fact-checker assigned to analyze a speech by a political figure. The speech includes several claims about economic performance and public policy impacts. For each claim, list the steps you would take to verify its accuracy, cite your sources, and provide a confidence level based on the evidence you find
- Emotional Prompting
- Direcciona al modelo para presentar una respuesta con una emocion en especifico
- Scenario:
You are developing an AI-driven chatbot designed to provide support for individuals seeking mental health advice. The chatbot needs to be sensitive to the emotional state of the users and respond in a way that is empathetic and supportive.
- Prompt Highlight: Respect [specific] feelings and ground your answer on those
- Full Prompt: Imagine you are an AI chatbot trained to offer support to individuals facing stress and anxiety. A user messages you saying, 'I've been feeling really overwhelmed with work and it's starting to affect my sleep. I don't know what to do.' Craft a response that acknowledges the user's feelings, offers reassurance, and provides practical advice or encouragement
- Various Perspectives
- Se le dice al modelo que considere y articule respuesta desde multiples puntos de vista.
- Scenario:
You are developing an AI system for an online discussion platform that helps users gain a deeper understanding of complex social issues by presenting multiple perspectives. The AI is tasked with providing diverse viewpoints on a controversial topic, such as climate change policy.
- Prompt Highlight: Highlight different perspectives [such as:]
- Full Prompt:
Imagine you are an AI discussion facilitator.
- A user has asked, 'What are the major perspectives on climate change policy and its economic impact?' Provide responses that highlight different perspectives, such as the views of environmentalists, economists, policymakers, and business leaders.
- Ensure each perspective is balanced and does not favor any single view
- Multi Role Prompting
- El modelo asume diferentes caracteristicas y roles
- Adaptando su respuesta en base a la variedad de posturas
- Scenario:
You are designing an AI-driven simulation tool for conflict resolution training. The tool is intended to help professionals in fields such as diplomacy, management, or customer service understand different stakeholder perspectives and practice negotiation skills.
- Prompt Highlight: Introduce roles and actors and make the “actors” to communicate among themselves
- Full Prompt:
Create a scenario where a project manager, a software developer, and a customer representative discuss a delay in project delivery.
- The project manager is concerned about maintaining the project timeline,
- the developer is worried about quality and technical challenges,
- and the customer representative is focused on client expectations and contractual obligations.
- Facilitate a dialogue that explores each role's concerns and proposes a compromise solution
- Role Prompting
- Similar al munti-role
- Pero generalmente involucra a un solo, rol consistente atravez de la interaccion
- Scenario:
Imagine developing an AI-driven simulation for software engineering teams to enhance communication andproblem-solving skills. The AI is designed to simulate the role of an experienced software engineer tasked with guiding less experienced team members through complex technical issues.
-
Prompt Highlight: You are …
-
Full Prompt: You’re a senior software engineer with a lot of experience in JavaScript
Feel free to show off and use all the syntactic sugar to make your code read as good as possible
-
-
Ask me before you Answer
- Basicamente te pedira aclarar o mas informacion antes de responder.
- Asegurandose que las respuestas estan on target
- Scenario:
You are developing an AI-driven consultation tool for a home renovation company that helps clients decide on design
choices. This AI tool needs to gather sufficient context before providing tailored advice, ensuring that all client
preferences and constraints are considered.
-
Prompt Highlight: Before suggesting / ask
-
Full Prompt: Imagine you are an AI design consultant for home renovations. A client has mentioned wanting a kitchen remodel but has not provided specific details about their preferences, budget, or the existing kitchen layout.
Before suggesting any design ideas, ask the client a series of detailed questions to gather all necessary information.
-
- Scenario:
You are developing an AI-driven consultation tool for a home renovation company that helps clients decide on design
choices. This AI tool needs to gather sufficient context before providing tailored advice, ensuring that all client
preferences and constraints are considered.
-
Rephrase and Respond
- Organiza la pregunta o statement antes de responder
- esto agrega claridad y entendimiento
- Scenario:
You are developing an AI customer service assistant for a telecommunications company. This AI tool is designed to handle inquiries about service issues, billing questions, or plan changes. The "Rephrase and Respond" technique is utilized to ensure that the AI fully understands the customer's query and to clarify the question before providing an
answer, thereby improving accuracy and customer satisfaction.
-
Prompt Highlight: Rephrase the question to confirm your understanding
-
Full Prompt: Imagine you are an AI customer service representative.
A customer has just asked, 'Why was my last bill so high?' Before answering, rephrase the question to confirm your understanding of the customer's issue and then provide a detailed explanation based on typical billing issues
-
- Scenario:
You are developing an AI customer service assistant for a telecommunications company. This AI tool is designed to handle inquiries about service issues, billing questions, or plan changes. The "Rephrase and Respond" technique is utilized to ensure that the AI fully understands the customer's query and to clarify the question before providing an
answer, thereby improving accuracy and customer satisfaction.
-
Self-Refine Prompting
- Refina iterativamente la respuesta
- basada en assessments interno o feedback externo
- Scenario:
You are developing an AI-driven editor for academic papers, designed to help students and researchers refine their writing for clarity, coherence, and adherence to academic standards. The AI uses the self-refine prompting technique to iteratively improve its own text outputs.
-
Prompt Highlight: First … then identify and refine …
-
Full Prompt: Assume you are an AI academic editor. A student has submitted a draft of their thesis on renewable energy policy.
First, generate a summary of the draft, then identify and refine any unclear or poorly structured sentences to enhance readability and academic tone. Additionally, ensure that all references are properly cited according to APA standards
-
- Scenario:
You are developing an AI-driven editor for academic papers, designed to help students and researchers refine their writing for clarity, coherence, and adherence to academic standards. The AI uses the self-refine prompting technique to iteratively improve its own text outputs.
-
Self Consistency Prompting
- Se asegura que las respuesta del modelo son consistentes durante la sesion
- Incluso cuando se pregunta lo mismo de diferentes formas
- Scenario:
You are developing an AI-driven policy advisor system designed to help government agencies ensure their public statements and policies are consistent over time. The AI uses the self consistency prompting technique to analyze its own responses for consistency with previous statements or established policy positions.
-
Prompt Highlight: Check the response against…
-
Full Prompt: Assume you are an AI policy advisor. You have been asked to draft a response to a public inquiry about the government's stance on renewable energy incentives. Before finalizing the response, check it against previous statements made in the last year to ensure it aligns with the established policy direction.
Identify any inconsistencies and adjust the response to maintain policy coherence
-
- Scenario:
You are developing an AI-driven policy advisor system designed to help government agencies ensure their public statements and policies are consistent over time. The AI uses the self consistency prompting technique to analyze its own responses for consistency with previous statements or established policy positions.
-
Fill in the Blanks
- Un formulario de prompt interactivo
- Donde el Modelo llena la informacion incompleta en el template o frase
- Scenario:
You are developing an AI-driven language learning tool aimed at helping users improve their grammar and vocabulary skills in a new language. The tool uses the "Fill in the Blanks" technique to create interactive exercises that help learners practice and reinforce language concepts.
- Prompt Highlight: Provide missing information or words
- Full Prompt: Imagine you are an AI language tutor. Create a fill-in-the-blanks exercise for a student learning English, focusing on the use of prepositions in sentences. Provide a short paragraph with missing prepositions and offer a list of options that the student can use to fill in the blanks. After the student attempts to fill in the blanks, provide the correct answers and explanations for each choice.
- Scenario:
You are developing an AI-driven language learning tool aimed at helping users improve their grammar and vocabulary skills in a new language. The tool uses the "Fill in the Blanks" technique to create interactive exercises that help learners practice and reinforce language concepts.
-
Zero Shot / Few Shot Learning
- Hace la tarea sin ejemplos previos (Zero-shot) o con pocos ejemplos (few-shot),
- dejandolo solamente con la informacion pre-entrenada.
- Scenario:
You are developing an AI system to help a customer support team quickly classify and respond to customer inquiries across various topics like shipping, product issues, and billing. The AI needs to understand and categorize these queries effectively without prior specific training on these topics
- Prompt Highlight: No response examples provided / response examples provided.
- Full Prompt: Here is a customer inquiry: 'I haven’t received my package yet, though it’s been two weeks since the shipping confirmation. What should I do?' Classify this inquiry into one of the following categories: Shipping, Product Issues, Billing, General Inquiry
-
Self Ask
- Yo lo veo mas como el proceso de Thinking,
- El modelo genera sus propias preguntas y despues las responde,
- asi se facilita una exploracion mas profunda y mejor entendimiento
- Scenario:
You are designing an AI-driven research assistant intended to help university students gather and synthesize information for academic papers. The AI should independently generate relevant questions based on a given topic, seek answers, and then present the information in a structured format.
- Prompt Highlight: Generate questions, find answers.
- Full Prompt: Imagine you are an AI research assistant. The student has asked for help in understanding the economic impacts of climate change. Generate relevant questions to explore this topic deeper, find answers to these questions, and summarize your findings
-
Generated Knowledge
- Se le pide al modelo generar data no explicitamente presente desde el prompt
- pero si inferida desde su entrenamiento.
- Scenario:
You are creating an AI system to assist environmental scientists in predicting the long-term impacts of deforestation on biodiversity. The AI needs to generate new insights or hypotheses based on available data and prior knowledge, rather than just retrieving information.
- Prompt Highlight: Generate hypotheses / generate insights / generate…
- Full Prompt: Assume you are an AI assistant specializing in environmental science. Based on current trends in global deforestation and your understanding of ecological dynamics, generate hypotheses about the potential impacts on biodiversity over the next 50 years. Consider factors like habitat loss, species extinction rates, and changes in forest structure
-
Self Critique
- El modelo evalua sus mismas respuestas
- Y sugiere mejoras
- Scenario:
You are developing an AI-driven content creation tool that assists marketing teams in generating and refining ad copy. The AI needs to not only produce initial drafts but also evaluate and improve its own outputs based on specific marketing goals and audience engagement metrics.
- Prompt Highlight: Critique your own work.
- Full Prompt: Create an advertisement copy for a new eco-friendly water bottle targeting young adults. After generating the initial draft, critique your own work. Identify any areas that could be improved in terms of persuasiveness, emotional appeal, and alignment with eco-friendly values, and then revise the copy accordingly
-
Self evaluation
- Similar al self-critique
- Pero se enfoca mas en medir su performance ante un standard o benchmark
- Scenario:
You are designing an AI-driven personal fitness coach that not only provides workout advice but also self-evaluates its effectiveness in helping users achieve their fitness goals. The AI should assess its recommendations based on user feedback and performance data, then adjust its guidance accordingly.
- Prompt Highlight: Evaluate feedback, effectiveness, reflect & suggest.
- Full Prompt: Consider yourself as an AI personal fitness coach. After suggesting a 4-week cardio program to a user, evaluate the effectiveness of your program based on the user's feedback and progress data. Reflect on the outcomes, identify any shortcomings, and suggest improvements for the next cycle
-
System 2 Attention Prompting
- Motiva al modelo a tener procesos de pensamiento mas profundo
- para procesar problemas complejos
- Scenario:
You're developing an AI system for a legal advisory tool that assists in drafting and reviewing complex contracts. This tool needs to employ deep, deliberate thought (akin to humans' "System 2" thinking) to ensure that all clauses are not only legally valid but also tailored to the specific needs and contexts of the clients.
- Prompt Highlight: Generate hypotheses / generate insights / generate…
- Full Prompt: Consider yourself an AI legal advisor. You are tasked with reviewing a draft contract for potential conflicts and alignment with both local regulations and the client's business objectives. Engage in detailed analysis, looking for ambiguities, potential legal issues, and areas where the client's interests may not be fully protected. Provide a comprehensive review and suggest amendments
- Clarification
- Motivar al usuario aclarar o definir sus pensamientos mas claramente
- Scenario:
When a statement or idea is vague or ambiguous, or when a user’s thoughts need further detail or explanation.
Useful in teaching, coaching, and problem-solving where incomplete ideas are presented.
- Prompt Highlight: Explain / elaborate
- Full Prompt: • Can you explain what you mean by [term/concept]? • Could you elaborate on that point? • How would you define [X]?
- Exploring Perspectives
- Motivar al usuario a considerar diferentes puntos de vista
- Scenario:
When a user seems stuck in a single viewpoint, when empathy or broader understanding is needed, or when there
might be multiple ways to interpret an issue. Useful in conflict resolution, creativity, and cross-functional
collaboration.
- Full Prompt: • How might someone with a different perspective view this? • What are other possible ways of looking at this? • What would someone who disagrees with you say?
- Probing Assumptions
- Cuestionar suposiciones o creencias subyacentes
- Scenario:
When a statement relies on underlying assumptions that are not explicit, or when users jump to conclusions based
on unspoken premises. Especially useful in critical thinking, debates, or design discussions where assumptions shape
outcomes.
- Full Prompt: • What assumptions are you making here? • Is there something we’re taking for granted? • Why do you think this assumption holds true?
- Implications and Consequences
- Ayudar al usuario pensar detalladamente en el impacto potencial de sus ideas
- Scenario: When the user makes a decision or proposes a solution without considering its broader or long-term effects. Useful in planning, policy-making, and strategic discussions.
- Full Prompt: • What are the implications if we follow this course of action? • What might be the long-term consequences of this? • How would this decision impact other areas of the problem?
- Scenario: When the user makes a decision or proposes a solution without considering its broader or long-term effects. Useful in planning, policy-making, and strategic discussions.
- Ayudar al usuario pensar detalladamente en el impacto potencial de sus ideas
- Probing Reasons and Evidence
- Empujar al usuario a pensar sobre las razones detras de sus creencias o conclusiones
- Scenario:
When a user provides an opinion or conclusion without explaining why or when evidence for a claim is missing. This is critical in fields like science, law, and decision-making where reasoning must be robust and defensible.
- Full Prompt: • What reasons do you have for that? • Can you provide evidence or examples to support your claim? • Why do you think this approach would work?
- Scenario:
When a user provides an opinion or conclusion without explaining why or when evidence for a claim is missing. This is critical in fields like science, law, and decision-making where reasoning must be robust and defensible.
- Empujar al usuario a pensar sobre las razones detras de sus creencias o conclusiones
- Meta-Questioning
- Reflejar en el proceso de pensar en si mismo
- Scenario: When it’s important for the user to reflect on their thought process itself, especially useful in teaching,
personal development, and improving problem-solving approaches.
- Full Prompt: • Why do you think this way? • How did you come to that conclusion? • What can we learn from this process? • Could we approach this problem differently?
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment