Frequently Asked Questions (FAQ)
Common questions about the SuperPrompt Framework.
What is a superprompt?
A superprompt is a structured prompt that acts as a cognitive contract between you and an AI model. It specifies what to think about (context), how to think (reasoning policy), and what to produce (output specification).
It’s not a longer prompt. It’s a designed thinking interface that creates predictable, reusable results.
What isn’t a superprompt?
A superprompt is not:
- A longer prompt: Length doesn’t matter. Structure does.
- A magic spell: Superprompts don’t guarantee perfect outputs, but they make outputs more consistent and debuggable.
- Tool-specific: Superprompts should work across AI tools (Claude, GPT, Gemini, Llama).
- One-size-fits-all: Different tasks need different structures. Use patterns from the pattern library to adapt.
How long should a superprompt be?
As long as it needs to be to achieve the intent. Some tasks need 100 words; others need 500. The goal is clarity, not brevity or length.
Rule of thumb: If your prompt scores ≥3 on all six axes of the evaluation rubric, it’s the right length.
Yes. Superprompts work in any tool that accepts natural language instructions. The template is tool-agnostic by design.
In Cursor, you can use superprompts for tasks like:
- Code refactoring
- Documentation generation
- Test case creation
- Architecture review
See the documentation cleanup example for a code-adjacent use case.
How do I adapt a superprompt to a new task?
- Start with the canonical template from template.md
- Replace placeholders with your specifics (intent, context, reasoning policy, output)
- Test it in your AI tool
- If it fails, use the evaluation rubric to debug
Quick customization levers:
- Change the tone (warm, sharp, formal)
- Add refusal rules to constrain outputs
- Include a JSON schema for structured data
What’s the difference between a superprompt and a pattern?
- Superprompt: A complete prompt ready to use for a specific task (e.g., “design a workshop”)
- Pattern: A reusable reasoning structure you can plug into any superprompt (e.g., “Decomposition,” “Role Mesh”)
Think of patterns as modular components you mix and match inside superprompts.
How do I know if my superprompt is good?
Use the evaluation rubric to score your prompt on six axes:
- Goal Fit
- Faithfulness to Context
- Reasoning Quality
- Constraint Compliance
- Usefulness of Output
- Reusability
If all axes score ≥3, your superprompt is ready to use.
What are common pitfalls when writing superprompts?
1. Vague intent: “Write a blog post” is too vague. Add success criteria: “Write a 600-word blog post that helps mid-level managers understand why AI adoption requires new skills.”
2. Missing context: If the model doesn’t have the facts it needs, it will guess. Add examples, constraints, and domain knowledge to the CONTEXT section.
3. No reasoning policy: Without explicit steps, the model will improvise. Define how you want it to think.
4. Generic constraints: “Be concise” is too vague. Say “500 words max” or “3 bullet points.”
5. Forgetting the self-check: Always include a SELF-CHECK section to verify outputs before finalizing.
Can I combine patterns?
Yes. Patterns are composable. For example:
- Use Decomposition to break a problem into parts, then apply Role Mesh to each part to get multi-expert feedback.
- Use Counter-Case Probing to stress-test an idea, then apply Critique–Revise Loop to refine it.
See the pattern library for more examples.
Add refusal rules to your REASONING POLICY. For example:
- “If you don’t have enough context to answer, ask for clarification instead of guessing.”
- “Refuse to produce outputs that could be used to manipulate or coerce.”
- “If the input is ambiguous, request clarification before proceeding.”
Refusal rules make your prompts more robust to edge cases.
What if my superprompt works in Claude but fails in GPT?
Your instructions may be too implicit. Different models interpret instructions differently.
How to fix:
- Make steps in REASONING POLICY more explicit (numbered, sequential)
- Add examples to CONTEXT to show what you mean
- Test with both models and note where they diverge
- Adjust the prompt to work for the lowest common denominator
Can I use superprompts for creative tasks (writing, art, design)?
Yes. Superprompts work for any task where you need structured thinking. Creative tasks benefit from:
- Style Transfer pattern (translate formal to casual, academic to conversational)
- Perspective Shift pattern (reframe a problem from a new angle)
- Critique–Revise Loop pattern (iterative refinement)
See the coaching reflection example for a creative application.
How do I share superprompts with my team?
Follow the workflow guide:
- Save your prompt to a shared repo (GitHub, Notion, etc.)
- Add metadata (tags, description, pattern used) to a PROMPTS.md index
- Use conventional commit messages for versioning
- License your prompts with CC-BY 4.0 for open sharing
What license should I use for my superprompts?
This framework uses CC-BY 4.0 (Creative Commons Attribution 4.0 International).
What this means:
- Anyone can use, modify, and share your prompts
- They must attribute you as the creator
- You retain copyright but grant others usage rights
This keeps superprompts open and reusable.
Can I use superprompts for commercial projects?
Yes. The CC-BY 4.0 license allows commercial use. Just provide attribution to the original creator (Steff Vanhaverbeke / coachsteff.live for this framework, or whoever created the specific prompt).
Where can I find more examples?
Check the /examples folder for five complete superprompts:
How do I contribute a new pattern or example?
Follow the 5-step workflow in workflow.md:
- Draft your prompt
- Test it in at least two AI tools
- Score it with the evaluation rubric
- Document it with tags and metadata
- Commit with a conventional commit message and open a pull request
Who created the SuperPrompt Framework?
The framework was created by Steff Vanhaverbeke, an AI Adoption Coach and co-founder of The House of Coaching. Steff specializes in the human side of AI adoption—helping professionals build the cognitive capabilities that matter most in an AI-driven world.
Learn more at coachsteff.live.
Where can I get help or report issues?
License
CC-BY 4.0 · Steff Vanhaverbeke · coachsteff.live