An AI agent is an LLM that is given tools, a goal and permission to act in a loop — planning, calling tools, and revising — rather than answering a single question. TPEE helps you write the prompts that drive them.

Prompt Engineering Learning Center

Q: What is prompt engineering?

Prompt engineering is writing instructions to an AI deliberately instead of free-typing them, so the model gives consistent, useful answers. TPEE structures every prompt into 12 sections and scores the result 0–100.

Q: What are MCP servers?

MCP (Model Context Protocol) servers let an AI assistant use external tools and data sources it doesn't natively have. TPEE records which MCP servers a prompt may use; the servers themselves run outside TPEE.

Three questions everyone asks when they start, answered in plain language — then the core lessons from TPEE's 83-lesson course, free to read right here.

By Jacques Botte, founder of Toptronic^®. Last updated 12 September 2026.

What is prompt engineering?

Prompt engineering is the practice of writing instructions to an AI language model deliberately instead of free-typing a paragraph and hoping. The model is doing its best with an incomplete brief, so it guesses — and a guess about what you wanted is a wasted answer. Prompt engineering removes the guessing by making the important assumptions explicit: who the model should be, what exactly to do, what context it has, and what a good answer looks like.

TPEE turns a vague request into a structured instruction across twelve sections and scores the result before you send it. Read the full guide.

What are MCP servers?

MCP (Model Context Protocol) servers are small programs that let an AI assistant reach tools and data it doesn't natively have — a database, a browser, a code editor, a filesystem. Instead of pasting everything into the chat, the model calls the server when it needs the tool. TPEE's MCP Servers section is where you declare which external tools a prompt may use. The servers themselves run outside TPEE; the app only records the instruction. TPEE itself never operates an MCP server or makes a network call.

What is an AI agent?

An AI agent is an LLM that is given tools, a goal and permission to act in a loop — planning, calling tools, checking results, and revising — rather than answering a single question. Agent teams split work into roles: lead, reviewer, tester, security analyst and more. They need clear responsibilities and handoff rules, otherwise multiple agents create noise faster than value. TPEE's Team feature records that metadata and produces the prompts; it does not spawn agents or connect to agent platforms.

Core lessons

These are the lessons people search for most, taken from the 83-lesson course that ships inside the TPEE app. Each one is a three-paragraph summary; the app carries the full lesson plus its quiz.

1. AI Fundamentals

An AI language model reads text and predicts useful next tokens. It does not truly know your project unless you give it context.

TPEE helps by turning a vague request into a structured engineering instruction: persona, task, context, technical specs, format, audience, tone, data, examples, graphics, MCP notes, and agent notes.

Think of the prompt as a configuration block — if one register is wrong or missing, the downstream system behaves unpredictably.

2. Prompt Anatomy

Persona tells the AI which expert hat to wear. Instructions state the job. Context explains the situation. Technical Specs define non-negotiable limits.

Format is critical because it tells the AI what shape the answer must have. Examples show the pattern. Data grounds the answer in specific facts.

You do not need every section every time, but a professional prompt always makes the important assumptions visible.

3. Context Windows and Tokens

A token is a small chunk of text. The context window is the model's working memory for the current request.

Long context is useful, but it costs money, time and memory. Put the highest-value facts first and remove old noise.

For coding, include file names, exact error text, acceptance criteria, and what must not change.

7. IDEs and Coding Agents

Each IDE or agent reads instructions differently. TPEE's IDE selector formats the same intent for Cursor, Claude Code, ChatGPT, Copilot, Windsurf and others.

For coding agents, specify the working directory, files to inspect, tests to run, and what must not be modified.

TPEE formats and stores the prompt — the target tool performs the coding only after you paste or export the prompt into it.

8. RAG, Hallucination and Verification

Hallucination means the AI produces plausible text that is not supported by the facts. RAG reduces this by giving the model documents to quote or use.

For business, legal, electrical, medical or production-code decisions, always ask for sources, inspect the source, and test the result.

TPEE's Data and Examples sections are where you tell the AI which facts matter most.

9. Privacy, IP and External Services

Before pasting content into any cloud AI, decide whether the information is confidential, personal, licensed or commercially sensitive.

TPEE itself is local/offline. It must not operate MCP servers, call AI APIs, phone home, upload files, or connect to external services.

The safe pattern is: prepare locally in TPEE, review what you are about to send, then consciously paste into the tool you selected.

10. Agent Teams and Orchestration

Agent teams split work into roles: lead, reviewer, tester, security analyst, documentation writer and more.

Good teams need clear responsibilities, handoff rules, shared task lists and plan approval — otherwise multiple agents create noise faster than value.

TPEE's Team feature records metadata and produces prompts. It does not spawn agents or connect to agent platforms.

16. AI Safety and Human Review

AI is useful for drafting, analysis and exploration, but it should not be the final authority for high-risk decisions.

Human review is mandatory for production code, electrical safety, legal text, medical advice, finance, security, and anything affecting customers.

A safe workflow marks AI output as proposed work until it is reviewed, tested and accepted by a responsible person.

17. RAG Foundations

RAG means the system retrieves relevant documents and supplies them to the model as context, so the answer can be grounded in known material instead of guessed.

The quality of RAG depends on chunking, metadata, retrieval ranking, prompt design, and the model's ability to cite or use the retrieved text.

TPEE does not run a RAG server; it teaches you how to ask for grounded answers and how to prepare source material for another tool.

The full course — 83 lessons and 249 quiz questions — ships inside the app. Get TPEE to study it offline at your own pace. When you're ready to put the lessons to work, TPEE's prompt manager keeps your prompt library and templates organized on Windows.

Prefer to read on the web? Every lesson of the course is published free on this site — all 83 of them, no account needed. Popular starting points:

1. AI Fundamentals — what an LLM is and why clear prompts matter.
2. Prompt Anatomy — use the 12 TPEE sections deliberately.
3. Context Windows and Tokens — the model's working memory, and what it costs.
5. Ollama — run large language models locally on your own machine.
7. IDEs and Coding Agents — format the same intent for Cursor, Claude Code and others.
8. RAG, Hallucination and Verification — ground the answer in facts you can check.
59. Your First Prompt - A Step-by-Step Example — from initial need to final prompt.
60. Common Beginner Mistakes and How to Fix Them — recognize and avoid the classic errors.
74. Ollama — the local-AI deep dive.
71. Mixture of Agents — multiple LLM agents in layers for better answers.

Browse the complete lesson list to see all 83.