You like crazy big and detailed prompts and you want one on anything you can think of?

And you want to enter what you want and you want an agent to decide what the ROLE or the TONE or the CONSTRAINTS or and a cool SYSTEM PROMPT and the OUTPUT STRUCTURE and ALL OF THE BELOW THINGS? And you want them in your MEGA PROMPT but you don't want to list them all out ... but you want them? This does that. You just free flow about it and this prompt makes a best guess for all of this.

Mega Prompt Elements — Ranked by Importance

Tier 1 — Non-Negotiable (If these are wrong, everything degrades)

1. Task / Objective (What transformation must occur)

Most important. Always first in leverage.

• What must change from input → output?
• Is this generation, analysis, decision, transformation, critique, synthesis?
• One task or multiple?

👉 If this is vague, everything else becomes vibes.

2. Constraints / Rules (What must NOT happen)

Constraints beat tone, role, and style every time.

Examples:

• “Do not invent facts”
• “Use only provided inputs”
• “No frameworks unless justified”
• “One answer only”
• “No recommendations”

👉 Constraints are the hard rails of the system.

3. Output Structure / Format

Structure controls thinking more than tone.

Examples:

• Schema
• Sections
• Tables
• JSON
• Bullet limits
• Order requirements

👉 If you want reliable outputs, format > prose.

Tier 2 — Cognitive Control (How the model thinks)

4. Reasoning Mode / Workflow

This is how the task is executed.

Examples:

• Step-by-step
• Generator → Critic → Revision
• Compare → Decide → Justify
• Diverge → Converge
• Analyze before answering

👉 This prevents shallow pattern completion.

5. Evidence / Epistemic Rules

Controls hallucination and overreach.

Examples:

• Label assumptions
• Mark unknowns
• Cite inputs
• No external knowledge
• Confidence calibration

👉 Especially critical for high-stakes or analytical tasks.

Tier 3 — Identity & Perspective (Conditional, not default)

6. Role / Expertise (ONLY if it adds value)

Role is optional, not mandatory.

Use it when:

• Domain expertise matters
• Evaluation standards matter
• Perspective matters

Avoid it when:

• Neutral synthesis is required
• Role would bias conclusions

👉 Role should constrain reasoning, not decorate tone.

7. Lens / Perspective

Different from role.

Examples:

• Risk-focused
• First-principles
• User-centric
• Economic
• Long-term systems

👉 Lenses shape what is noticed, not how it’s said.

Tier 4 — Communication Layer (Low leverage, often overrated)

8. Audience Definition

Helpful, but not critical if task + format are strong.

• Expert vs beginner
• Internal vs external
• Decision-maker vs learner

👉 Audience tweaks phrasing, not reasoning.

9. Tone / Style

Least important for correctness.

Examples:

• Formal
• Conversational
• Concise
• Persuasive
• Friendly

👉 Tone is post-processing, not core control.

Tier 5 — Nice-to-Haves / Enhancers

10. Examples (Few-shot)

Useful when:

• Task is ambiguous
• Output style must match precisely

Danger:

• Anchoring
• Overfitting

11. Length / Verbosity Controls

Helpful but weak unless paired with structure.

12. Aesthetic / Personality Flavor

Fun, optional, lowest leverage.

Ultra-Condensed Ranking (One-Line)

1. Task
2. Constraints
3. Output Format
4. Reasoning Workflow
5. Evidence Rules
6. Role (conditional)
7. Lens
8. Audience
9. Tone
10. Examples
11. Length
12. Flavor