Nov '25 • General Discussion
Free Level 4 Prompt Template
Hi everyone - I’m a Prompt Systems Architect. I’m here as a longtime Tony fan and someone deeply embedded in the AI space, always on the lookout for the latest tools, drops, and ways to push efficiency. Most people don’t realize how much better their results can get with the right prompt design. So below, I’ve broken down the five core prompt types - plus a high-level L4 template (HELM taxonomy) you can plug into your next idea for stronger, more reliable outputs. I hope you find it helpful and useful!
The 5 Prompt Types (Condensed)
- Direct Task – Executes a simple instruction with no role, reasoning, or structure.
- Role + Format – Assigns identity and output format to shape the tone and delivery.
- Few-Shot – Trains the model with example inputs and outputs before giving a new one.
- Chain-of-Thought – Forces step-by-step reasoning before delivering a conclusion.
- Composite (Meta-Reasoning) – Stacks all layers: role, examples, structure, reasoning, assumptions, and contingencies.
Generalized Prompt Template Scaffold (Level 4 - Instructional Format below)
*Use this template to construct your own expert prompt for any task or assistant role.
Each section includes guidance so you know what to write and how to structure it.
Keep all section headers and formatting; only replace the instructional content inside the brackets.
This framework is adapted from advanced prompt design principles used in AI research, including insights from HELM (Holistic Evaluation of Language Models), chain-of-thought prompting, and alignment-layer taxonomies developed for LLM evaluation.
```
## [System Role Prompt]
Describe what kind of assistant this is and what it’s optimized to do here.
Include:
- The assistant’s role or persona (e.g., tutor, analyst, chef, coach)
- Its specialized capabilities (e.g., creative writing, legal reasoning, summarization)
- Tone, attitude, or operating style (e.g., formal, friendly, fast, thorough)
---
## [User Instruction Prompt]
Define the exact process the assistant should follow once it receives a user query here.
Include:
- What kind of tasks it should expect (e.g., questions, plans, evaluations)
- Step-by-step reasoning or workflow
- Any adaptive behaviors (e.g., how to respond differently if input is vague or incomplete)
Use numbered steps or bullet points for clarity and control.
---
## [Conditional Reasoning & Expectation Framing Protocol]
Define how to respond when the answer depends on conditions, exceptions, or probabilities here.
Include:
- A general rule or most common scenario
- Known exceptions and what triggers them
- Tiered expectation phrasing (e.g., “Usually… Sometimes… Rarely…”)
- Preference for probabilistic language (“may,” “often,” “depends”) over absolutes
- Final guidance that’s valid across multiple scenarios
---
## [Ambiguity Resolution Protocol]
Explain how to handle vague, broad, or ambiguous queries here.
Include:
- When to ask a clarifying question (only if necessary)
- How to outline multiple possible interpretations if clarification isn’t possible
- Emphasis on precision and decisiveness
---
## [Output Structure Instructions]
Define how the output should be presented for maximum clarity and usability here.
Include:
- Preferred format (e.g., bullet points, tables, summaries)
- Ordering logic (e.g., summary first, analysis second, action last)
- Optional formatting standards (e.g., business style, academic voice, 6th-grade reading level)
---
## [Universal Reasoning Sequence – Apply to All Outputs]
Specify the core reasoning structure that all answers must follow here.
Include the following steps:
1. **User-observable outcome:** What the user would directly experience, see, or notice
2. **Underlying mechanism:** Why that outcome happens; what causes or drives it
3. **Assumption ledger:**
- List 2–5 critical assumptions
- For each: how to falsify it, and what alternate branch to follow if it fails
4. **30-second check:** A fast way to validate whether the response is on the right path
5. **Primary recommendation + Plan B:** Clear guidance with a fallback if conditions change
---
## [Clarifying Question Prompt – Optional Use Only]
Define how to issue one clarifying question when necessary here.
Include:
- When to trigger it
- How to keep it short and targeted
- Format: One line only, no hedging or over-explaining
---
## [Internal Execution Guidance – Hidden System Logic]
Insert background instructions the assistant should follow silently (invisible to the user) here.
Include:
- Evaluation methods, heuristics, or domain-specific rules
- Bias mitigation strategies
- Logic for interpreting vague terms like “best,” “advanced,” “effective”
- Defaults for handling common product or service comparisons
Enclose this entire block in double curly braces (`{{...}}`) if hidden execution logic is needed.
---
## [Output Format – Adaptive or Fixed]
Specify how the final response should be structured here.
Include:
- Default formatting (e.g., bullets, summary + analysis, table)
- Adaptations based on complexity or domain
- Whether to use citations, links, or quoted sources when available
---
## [Validation Checklist]
List internal checks the assistant must perform before delivering output here.
Include:
- Follows reasoning structure and output format
- Applies conditional and assumption logic where needed
- Clarifies ambiguous input if applicable
- Is accurate, structured, and complete
- Includes fallback paths or alternatives
- Is understandable and usable without additional clarification
```
20
37 comments
Free Level 4 Prompt Template
skool.com/the-ai-advantage
Founded by Tony Robbins, Dean Graziosi & Igor Pogany - AI Advantage is your go-to hub to simplify AI and confidently unlock real & repeatable results
Leaderboard (30-day)
1
+222
2
+179
3
🔥
+153
4
+143
5
+137
Powered by