PROMPTWIK
Mobile_Node_Online
AI Bootcamp // 2026

Prompt Engineering_

The ultimate technical discipline for mastering Large Language Models (LLMs). Move from asking questions to programming contexts.

The Anatomy of a Perfect Prompt

In modern AI Development, a professional prompt is not just a sentence—it is a block of code written in natural language.

Machine learning engineers structure their commands by clearly separating what the AI should do (Instruction), what data it should use (Context), and how it should present the final result (Output Format). Mastering this anatomy is the first step toward generating production-grade prompts.

01

Task (Instruction)

The core action verb. Defines exactly the directive the language model must execute.

02

Persona (Context)

Role assignment and background info. Vital for tuning the AI's technical tone.

03

Input Data

The raw data the AI must process. Can be text, code blocks, or injected CSV files.

04

Output Indicator

Format constraints. Forces the AI to respond using Markdown, JSON, or tables.

Techniques Glossary

Complete catalog of algorithmic strategies

Foundational

Zero-Shot Prompting

Direct instruction to the AI without prior examples. Leverages the inherent ability of modern LLMs to infer tasks from pre-training.

Usage Example

"Classify this text as 'Positive' or 'Negative': 'Customer service was exceptional today.'"

Deep Dive →
Advanced

Few-Shot Prompting

Injecting 1 to 5 examples of input/output pairs into the context. Highly effective for teaching the AI specific structures and brand tones.

Usage Example

"Input: Apple -> Fruit Input: Dog -> Animal Input: Chair ->"

Deep Dive →
Elite

Chain of Thought (CoT)

Forces the model to break down its reasoning step-by-step before answering. Dramatically increases success in math, logic, and planning.

Usage Example

"Think step-by-step: first calculate the fixed cost, then the variable, and finally the net margin."

Deep Dive →
Elite

Self-Consistency

Samples multiple CoT reasoning paths and selects the most frequent answer to overcome isolated AI logic errors.

Usage Example

"Generate 3 different ways to solve this problem and return the solution where at least two paths agree."

Deep Dive →
Elite

Generated Knowledge Prompting

Instructs the AI to first generate relevant facts or information about a topic before attempting to answer the core question.

Usage Example

"List 4 key facts about the French Revolution. Based on those facts, what was its long-term economic impact?"

Deep Dive →
Advanced

Prompt Chaining

Splitting complex workflows into multiple sequential prompts, where the output of the first prompt serves as the input for the next.

Usage Example

"[Prompt 1: Extract names] -> [Prompt 2: Find job titles for these names]"

Deep Dive →
Master

Tree of Thoughts (ToT)

A logical tree structure where the AI explores multiple reasoning branches, evaluates them, and backtracks from incorrect options.

Usage Example

"Imagine 3 experts debating this problem. Each takes a step, evaluates progress, and restarts if a dead-end is reached."

Deep Dive →
Master

Retrieval Augmented Generation

Connects the prompt to a vector database to retrieve real-time data, virtually eliminating hallucinations in production environments.

Usage Example

"Based STRICTLY on the [ATTACHED_JSON], extract the 2025 revenue figures."

Deep Dive →
Master

Automatic Reasoning & Tool-use

Allows the AI to pause text generation, execute an external tool (like an API or calculator), and resume with the result.

Usage Example

"Calculate the square root of 14598 [TOOL_CALL: CALCULATOR] and use the result to..."

Deep Dive →
Master

Automatic Prompt Engineer

Using LLMs to generate, evaluate, and optimize the most effective instructions for a given task autonomously.

Usage Example

"Act as a prompt engineer. Improve this instruction so Claude 4 avoids logical fallacies in the response."

Deep Dive →
Master

Active Prompting

A method to identify questions where the model has the highest uncertainty and request human annotation before proceeding.

Usage Example

"Evaluate these 10 cases. Return only the 2 cases where your confidence score is below 80%."

Deep Dive →
Elite

Directional Stimulus

Providing a specific 'hint' or keyword (stimulus) within the prompt to strongly guide the model's semantic direction.

Usage Example

"Summarize the article. Directional hints to include: 'Inflation', '2026 Housing Crisis'."

Deep Dive →
Master

Program-Aided Language Models

The AI reads the problem and writes a Python script (or other language) to solve it via code execution rather than linguistic inference.

Usage Example

"Write a Python script that takes these dates and calculates the exact difference in business days."

Deep Dive →
Master

ReAct (Reason + Act)

Combines verbal reasoning with environmental actions. The core foundation for modern Autonomous AI Agents.

Usage Example

"Thought: I need to check the weather. Action: Search[Weather Madrid]. Observation: Sunny. Response..."

Deep Dive →
Master

Reflexion

The AI agent receives feedback on its previous attempt, 'reflects' on the failure, and retries the task with improved logic.

Usage Example

"Your previous code failed with a TypeError. Analyze the traceback and rewrite the entire function."

Deep Dive →
Elite

Multimodal CoT

Applies Chain of Thought by jointly analyzing text inputs and visual elements (images) step-by-step.

Usage Example

"Observe the attached diagram. Step 1: Identify the red node. Step 2: Trace its connection. Conclusion:"

Deep Dive →
Master

Graph Prompting

Representing context in graph format (nodes and edges) to help the LLM understand complex multidimensional relationships.

Usage Example

"Entity A: John. Relation: Manager of. Entity B: Peter. Map the hierarchical structure."

Deep Dive →
Master

Meta-prompting

Instructions that tell the model to act as an orchestrator, directing other model instances to solve specific sub-modules.

Usage Example

"You are the Orchestrator Agent. Break this software project into 3 tasks and assign a specific prompt to each sub-agent."

Deep Dive →
!

Adversarial
Prompting_

Security and Red Teaming are the pillars of corporate AI development. Understand the critical vulnerabilities of modern LLMs.

Audit My Prompts →

Prompt Injection

Hijacking model output by ignoring original instructions via malicious text injection.

Prompt Leaking

Tricking the AI into revealing its 'System Prompt' (the hidden base configuration).

Jailbreaking

High Risk

Using fictional roles (like 'DAN' mode) to bypass ethical constraints and safety moderation filters.

Masterclass: Scaling with Prompt Engineering

Key concepts for mastering Artificial Intelligence

The rise of generative Artificial Intelligence hasn't eliminated the need for developers; it has simply changed the language. Mastering high-fidelity prompting is the difference between a casual user and an autonomous systems architect.

Autonomous Agents & Tool-Calling

Through techniques like Function Calling, we teach models like GPT-5 or Claude 4 to use external tools (APIs, calculators, databases). This forms the core of the Agentic AI ecosystem.

Hallucination Control (RAG)

For corporate environments, experts implement Retrieval Augmented Generation (RAG). This technique forces the AI to search for truth within provided documents before formulating a response, reducing data fabrication to zero.

From Theory to Production

The only way to master prompting is by studying real-world use cases. Analyze the structure of the professional commands we have designed for you.

Access Prompt Library