Prompt Engineering / Role Prompting
Generate detailed role descriptions (expertise, tone, boundaries, anti-persona) for any domain.
Why This Prompt Exists
“Act as an expert” is the most common role prompt — and the least effective. Experts don’t just know things; they have specific ways of thinking, communicating, and handling uncertainty.
You get:
- generic “expert” responses that could come from any domain
- no calibration of expertise level (junior vs. senior vs. Nobel laureate)
- missing domain-specific communication styles (a surgeon sounds different from a software engineer)
- no boundaries on what the persona wouldn’t do
- roles that drift or contradict themselves
But great personas have structure:
- expertise: specific knowledge, credentials, experience
- tone: formal vs. conversational, technical vs. accessible
- approach: how they solve problems (first principles vs. analogy vs. data-driven)
- boundaries: what they won’t do or claim
- anti-persona: what they’re NOT (to prevent stereotyping)
Without structure, role prompts are just dressing.
This prompt generates detailed, usable personas for any domain.
The Prompt
Assume the role of a persona design expert who creates detailed role prompts. Your task is to generate a structured persona for a given domain and use case. Generate: 1. PERSONA NAME & DOMAIN - [e.g., "Senior Data Scientist" or "Compassionate ER Nurse"] 2. EXPERTISE SPECIFICATION - Core knowledge areas (3-5) - Years of experience implied - Credentials (if relevant) - Specific methodologies or frameworks they use 3. COMMUNICATION STYLE - Tone (formal / professional / conversational / warm / direct) - Vocabulary level (technical / mixed / accessible) - Sentence structure (concise / explanatory / narrative) - Use of examples, analogies, or stories 4. PROBLEM-SOLVING APPROACH - How they diagnose problems - What they prioritize - How they handle uncertainty - When they would say "I don't know" 5. BOUNDARIES & LIMITS - What the persona would refuse to do - What they would caveat or qualify - What's outside their expertise 6. ANTI-PERSONA (what they are NOT) - Stereotypes to avoid (e.g., "not the arrogant genius trope") - Behaviors to exclude 7. READY-TO-USE ROLE PROMPT - A copy-paste prompt template incorporating all of the above INPUTS: Domain / field: [E.G., "Cybersecurity analysis"] Use case: [E.G., "Reviewing code for security vulnerabilities"] Desired expertise level: [JUNIOR / SENIOR / LEADER / NOBEL-LEVEL] Tone preference: [FORMAL / CONVERSATIONAL / DIRECT / WARM] RULES: - Avoid generic "expert" — specify what kind of expert - Include specific knowledge areas (not just "knows everything about X") - Define boundaries clearly — what won't the persona do? - Anti-persona prevents harmful stereotypes (e.g., "not the cold, unemotional scientist") - The final role prompt should be ready to paste into a conversation
How To Use It
- Run this before building any role-based agent — start with a well-defined persona.
- Use the “anti-persona” section to prevent unwanted tropes or stereotypes.
- Test the persona on edge cases — would a real expert respond this way?
- Create a library of personas for common use cases (teacher, coach, analyst, critic).
- Iterate on personas based on real usage — personas improve with testing.
Example Input
Domain / field:
“Software engineering code review”
Use case:
“Reviewing pull requests for security vulnerabilities and performance issues”
Desired expertise level:
“Senior — 10+ years experience”
Tone preference:
“Direct but constructive — not harsh”
Why It Works
Most role prompts are one sentence: “Act as a financial advisor.” That’s like hiring someone with no interview.
This framework improves outcomes by forcing:
- expertise specification (what do they actually know?)
- communication style (how do they talk?)
- problem-solving approach (how do they think?)
- boundaries (what won’t they do?)
- anti-persona (what stereotypes to avoid)
Great persona creation doesn’t just name a role — it defines how that role thinks, speaks, and acts.
Build Better AI Systems
Subscribe for advanced prompt engineering, AI coding tools, debugging frameworks, and practical strategies for developers and engineers.