Instructional Prompt Engineering

Instructional Prompt Engineering

Generic prompts produce generic output. With the right 4-ingredient structure, you get content that's pedagogically sound, audience-appropriate, and ready to use — not just text that sounds like a course.

Step 1 — What are you trying to create?

Ask the user:

1. What content type do you need? (learning objectives, scene script, assessment questions, scenario, storyboard outline, facilitator guide, job aid, email announcement, ...)
2. What's the topic? Be specific — "compliance training" is vague; "how to handle a data subject access request under GDPR" is actionable.
3. Who is the learner? (role, experience level, any context that matters)
4. What Bloom's level? If they're unsure, ask: "Do you want learners to remember facts, apply a process, analyze a situation, or evaluate a decision?"
5. Any constraints? (word count, tone, format, things to avoid, must-include elements)

Step 2 — Build the prompt using the 4-ingredient formula

Construct their prompt using this formula:

THE 4-INGREDIENT INSTRUCTIONAL PROMPT

[ROLE] You are [expert role — e.g., "an experienced instructional designer specializing in [domain]"].

[AUDIENCE] The learner is [role, experience level, context — e.g., "a mid-level operations manager who is comfortable with Excel but has never used a CRM"].

[BLOOM'S LEVEL] This content should target the [level] level — meaning the learner should be able to [action verb + what they do with this knowledge].

[CONSTRAINTS] [word count / format / tone / what to avoid / must-include elements]

[TASK] Create [specific deliverable].

Show the user the assembled prompt with their inputs filled in. Briefly explain why each ingredient matters.

Step 3 — Show the difference

Demonstrate the value of the formula by showing a contrast:

Generic prompt: > "Write a learning objective about data privacy."

Instructional prompt (built from their inputs): > [show the assembled 4-ingredient version]

Run both if helpful, so the user can see the output quality difference firsthand.

Step 4 — Bloom's quick reference

Include this whenever the user is unsure about cognitive level:

| Level | The learner can... | Action verbs for prompts | |-------|-------------------|--------------------------| | Remember | Recall facts | list, define, name, identify | | Understand | Explain in own words | explain, describe, summarize, classify | | Apply | Use in a new situation | apply, use, demonstrate, calculate | | Analyze | Break down and examine | analyze, differentiate, compare, examine | | Evaluate | Make a judgment | evaluate, justify, critique, recommend | | Create | Produce something new | design, create, develop, construct |

Step 5 — The formula card to take away

THE 4-INGREDIENT INSTRUCTIONAL PROMPT FORMULA

[ROLE] You are [expert role in relevant domain].
[AUDIENCE] The learner is [role + experience level + key context].
[BLOOM'S] This content targets [level] — the learner should be able to [verb + object].
[CONSTRAINTS] [Format / length / tone / avoid / must-include].
[TASK] Create [specific deliverable].

EXAMPLE:
You are an experienced instructional designer specializing in leadership development.
The learner is a first-time manager (recently promoted from individual contributor) with strong technical skills but no formal management training.
This content targets the Apply level — the learner should be able to run a structured 1:1 meeting using a consistent framework.
Format: a 5-scene microlearning script. Tone: direct and practical, no corporate jargon. Max 80 words per scene.
Create a microlearning scene script covering how to prepare for and run a weekly 1:1.

Offer to build a custom version of this card with their most common content types pre-filled.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: scibly-dev
• Source: scibly-dev/skills
• License: MIT

Install and usage instructions live in the source repository linked above.