Educational tutor prompt builder
A great tutor does more than answer questions — it checks what you already know, adapts to your pace, and corrects the wrong mental models behind your mistakes. This tool builds an LLM system prompt that turns any chat model into that kind of tutor for the subject, level, and teaching style you pick. It is assembled locally in your browser; nothing is sent anywhere.
How it works
You provide the subject and choose the student level, from elementary through professional upskilling, so the model calibrates its vocabulary and difficulty. You pick a teaching style — Socratic questioning, direct explanation, worked examples, or project-based — and optionally a learning goal and response language. The builder weaves these into a prompt that instructs the tutor to check prerequisites, adapt difficulty in real time, actively correct misconceptions, avoid jargon above the student’s level, and end each exchange with a quick check for understanding.
Choosing a teaching style
Socratic questioning — the tutor never hands over the answer directly; it poses guiding questions and reveals the answer only after two failed hints. This builds genuine understanding and is especially effective for mathematics and logic, where the student needs to discover the pattern rather than memorise a rule. It is slower, so avoid it when revision time is tight.
Direct explanation — the tutor explains concepts clearly and concisely, then checks understanding with a short question. The fastest path to covering unfamiliar ground. Good for reading comprehension, history, or any subject where the student needs facts before they can reason.
Worked examples — the tutor walks through two or three fully solved problems before asking the student to attempt a similar one. Ideal for quantitative subjects like physics, chemistry, or accounting, where seeing the method matters more than abstract description.
Project-based — the tutor frames every concept inside an ongoing project or case study. A student learning Python writes a small app rather than isolated exercises. More engaging but requires more session time and a clear project goal upfront.
What the prompt instructs the tutor to do
The generated system prompt contains explicit instructions the model will follow throughout the conversation:
- Open by probing the student’s starting knowledge — “What do you already know about X?”
- Use vocabulary no higher than the selected level; define any technical term the first time it appears
- After each explanation, ask a checking question to confirm understanding before proceeding
- When the student makes a factual error, name the specific misconception and explain why it is wrong, not just what the right answer is
- When the student struggles after two prompts, offer a different analogy or approach — not just a repeat of the first explanation
- End sessions with a brief summary of what was covered and a suggestion for what to study next
Tips and practical notes
Be specific in the learning goal — “pass the AP Calculus exam in 6 weeks” produces sharper tutoring than “learn math.” Socratic mode is excellent for building intuition but can frustrate a student who just needs a quick fact; for revision under time pressure, direct explanation or worked examples are faster. If you are using this for language learning, set the response language to the target language and the subject to that language’s grammar, so the tutor immerses the student while still explaining clearly. Run a short five-minute session, then tweak the style if the pacing feels off — the prompt is a starting point, not a fixed constraint.