Stage: All · Language: English · Format: Guide with prompt templates
A “prompt” is simply the instruction you give an AI tool. Clear prompts save time: you get a usable draft instead of five rounds of “no, not like that”. This guide gives you twelve habits, a skeleton you can reuse, and techniques with classroom examples.
12 prompting habits
- Lead with the instruction. Start with a clear verb: write, list, summarise, classify, translate, explain.
- Be specific. Name the class, subject, topic, time, style and limits.
- Replace vague words with numbers. “Five questions”, “45 minutes”, “two examples per stanza”.
- Say what to do, not only what to avoid. “Use simple sentences” works better than “don’t be complicated”.
- Control the output shape. Ask for a table, bullet list, headings or a numbered sequence.
- Show, don’t hint. Add one or two short examples of what you want.
- Separate the parts. Use simple dividers such as ### between instruction, background and your text.
- Break big tasks into steps. Unit plan first, then each lesson, then the worksheet.
- Start simple, then improve. Add context only when it improves the result.
- Ask it to check itself. “Before you finish, check that every question matches the learning outcome.”
- Mind the length. Include only details that help the task.
- Keep a prompt library. Save prompts that work and reuse them.
A prompt skeleton you can reuse
Task: [what to produce — start with a verb]
Context: [class, subject, board, students’ level, local context]
Requirements:
- [must-have 1]
- [must-have 2]
Format: [table / bullet points / headings]
Tone and audience: [e.g. simple English for Class 6]
Limits: [word count, time, what to leave out]
Quality check: [e.g. “mark anything you are unsure about”]
Techniques with classroom examples
Zero-shot: just a clear instruction
Weak: “Give activities for a geometry lesson.” Better:
Produce three activities for a Class 5 lesson on the area of rectangles. For each give: title, time in minutes, materials, steps, a support idea, and one check-for-understanding question. Present as a table.
Few-shot: show an example of what you want
Write three learning objectives in this pattern.
Example: “Students will identify two examples of figurative language in each stanza and explain their effect in two sentences.”
Now write objectives for: simile, metaphor and tone.
Step by step: ask for reasoning, then a clear final answer
List up to three numbered steps to differentiate a Class 4 fractions lesson. Each step: one sentence plus a check for understanding. End with “Final plan:” and a two-sentence summary.
Template (meta) prompting: give the structure to fill
Follow this structure only: Unit title; Learning outcomes; Key question; Assessments (diagnostic, formative, summative); Lesson sequence table [Lesson | Objective | Key activity | Check for understanding | Homework]; Materials.
Fill it for Class 6, topic: ecosystems.
Several options, then choose
Write 5 possible key questions for a Class 8 unit on energy. Score each 1–5 on: open-ended, needs evidence, connects to daily life. Show scores, then pick the best and explain why in one sentence.
Generate facts first, then answer
For factual topics, ask the AI to first list the key facts it is relying on — then check those facts yourself before using the answer.
Before you use any AI output
- Check facts against your textbook or an official source.
- Match it to your syllabus and learning outcomes.
- Rewrite examples so they reflect your students’ lives.
- Never paste students’ names, marks or personal details into a prompt.
Further reading: Prompt Engineering Guide · OpenAI prompting guide · Anthropic prompt engineering overview
