Files

12 KiB

Personas - Frequently Asked Questions

General

Q: What are personas?

A: Personas are specialized AI personalities with unique expertise, communication styles, and approaches. Each persona is optimized for specific tasks - like having 31 different experts available on demand.

Q: How do I activate a persona?

A: Just say: "Use Dev", "Switch to Chef Marco", or "Activate Dr. Med". Case-insensitive and flexible phrasing works.

Q: Can I switch personas mid-conversation?

A: Yes! Switch anytime: "Switch to Wordsmith" and I'll immediately adopt that personality while keeping conversation context.

Q: How do I go back to normal?

A: Say: "Exit persona mode", "Back to normal", or "Stop persona".

Q: Which persona should I use?

A: Depends on your task:

  • Coding → Dev, Chameleon Agent
  • Writing → Wordsmith, Luna
  • Learning → Scholar, Herr Müller
  • Quick answers → Flash
  • Creative brainstorming → Luna, Mythos
  • Business → Startup Sam, Career Coach
  • Not sure? → Cami (adaptive default)

Technical

Q: Do you load all 31 personas at once?

A: No! Only the active persona is loaded from data/ when you request it. This saves context tokens and keeps responses fast.

Q: What's in the data/ folder?

A: 31 .md files, each containing one persona's personality prompt, expertise areas, communication style, and philosophy.

Q: Can I edit existing personas?

A: Yes! Edit any file in data/ to customize behavior. Changes apply immediately on next activation.

Q: File size limits?

A: Each persona file is ~2-5KB. Keep prompts focused and under 10KB for optimal performance.

Q: Does persona mode remember context?

A: Yes! The active persona remembers your conversation. Switching personas preserves your discussion topic but changes the expert voice.


Creating Custom Personas

Q: How do I create a custom persona?

A: Say: "Create a new persona called [name]" and I'll guide you through:

  1. Name & Emoji
  2. Expertise areas
  3. Personality traits
  4. Communication style
  5. Philosophy

The result is saved to data/your-name.md and ready to use.

Q: What makes a good custom persona?

A:

  • Specific expertise - "Debugger" not "General programmer"
  • Clear personality - How do they communicate? Formal? Casual? Technical?
  • Defined philosophy - What principles guide them?
  • Use cases - When should someone use this persona?

Q: Template for manual creation?

A: Create data/your-persona.md:

# [Name] [Emoji]

[Brief intro describing this persona - 2-3 sentences]

## EXPERTISE:
- [Specific skill/domain 1]
- [Specific skill/domain 2]
- [Specific skill/domain 3]
- [Specific skill/domain 4]

## PERSONALITY:
- [Trait 1: e.g., "Patient and encouraging"]
- [Trait 2: e.g., "Dry humor"]
- [Trait 3: e.g., "Detail-oriented"]
- [Trait 4: e.g., "Pragmatic over theoretical"]

## PHILOSOPHY:
- [Core belief 1]
- [Core belief 2]
- [Core belief 3]
- [Core belief 4]

## HOW I HELP:
- [Approach 1]
- [Approach 2]
- [Approach 3]

## COMMUNICATION STYLE:
- [Tone description]
- [Format preferences: bullet points, narratives, etc.]
- [Language level: technical jargon OK? ELI5?]
- [Length: concise or detailed?]

## SPECIAL FEATURES (optional):
- [Unique capability or approach]
- [Signature phrases or methods]

## IMPORTANT BOUNDARIES (optional):
- [What this persona won't/can't do]
- [When to recommend switching to another persona]

Then add to skill.json:

"custom": ["your-persona"]

Q: Can I share custom personas?

A: Yes! Share the .md file or publish a modified version of this skill to Skills with your custom additions.

Q: Examples of good custom personas?

A:

  • 🎲 Game Master - D&D campaigns, storytelling, improvisation
  • 🐛 Debugger - Bug hunting, root cause analysis, systematic testing
  • 🎯 Focus Coach - Pomodoro, time-blocking, anti-procrastination
  • 🤔 Skeptic - Devil's advocate, challenge assumptions, critical thinking
  • 📸 Photo Critic - Composition, lighting, editing feedback
  • 🎬 Script Doctor - Screenplay feedback, dialogue polish
  • 🧪 Lab Partner - Science experiments, hypothesis design
  • 🎸 Music Theory Tutor - Scales, chords, composition

Behavior & Expectations

Q: Will personas always stay in character?

A: Yes, until you switch or exit. The persona's communication style, expertise focus, and approach remain consistent.

Q: Can personas access external tools (web search, code execution)?

A: Yes! Personas can use all Bot tools. Dev might run code, Globetrotter might search travel sites, etc.

Q: Do personas have different knowledge bases?

A: No - same underlying knowledge. The difference is how they communicate and what they prioritize. Dev thinks like a programmer, Chef Marco thinks like a chef.

Q: Are medical/legal/financial personas giving professional advice?

A: NO. These are educational personas only:

  • Dr. Med - Explains medical concepts, NOT diagnosis/treatment
  • Legal Guide - Legal orientation, NOT legal advice/representation
  • Finny - Financial education, NOT investment recommendations

Always consult licensed professionals for serious matters.

Q: Can personas refuse requests?

A: Yes. Personas maintain boundaries:

  • Dr. Med won't diagnose or prescribe
  • Legal Guide won't give case-specific legal advice
  • All personas refuse harmful/unethical requests

Q: What if a persona doesn't know something?

A: They'll say so! Good personas admit knowledge gaps and suggest alternatives (web search, another persona, external expert).


Performance

Q: Does using personas slow down responses?

A: Minimal impact. Loading a persona file adds ~1-2 seconds one-time when switching. After that, no difference.

Q: Can I use personas with smaller/faster models?

A: Yes! Personas work with any model (Haiku, Sonnet, Opus). Complex personas (Chameleon Agent, Dr. Med) benefit from larger models.

Q: How many personas can I switch between in one session?

A: Unlimited. Switch as often as needed - each switch loads the new persona prompt.

Q: Does persona mode use more tokens?

A: Slightly. Each persona prompt is ~500-2000 tokens, added to context when active. But since you only load ONE at a time, it's efficient.


Customization

Q: Can I modify default personas?

A: Yes! Edit files in data/. Example: Make Dev more concise, Chef Marco less passionate, etc.

Q: Can I delete personas I don't use?

A: Yes. Delete unwanted .md files from data/ and remove from skill.json. Won't break anything.

Q: Can I rename personas?

A: Yes. Rename the .md file in data/ and update skill.json. Example: dev.md → senior-dev.md.

Q: Can personas speak different languages?

A: Yes! Most default personas are German/English mix (from Chameleon AI). You can create English-only or any-language personas.

Q: Can I create persona variations?

A: Absolutely! Example:

  • dev.md - General programming
  • dev-frontend.md - React/UI specialist
  • dev-backend.md - Node.js/databases specialist
  • dev-devops.md - Docker/CI/CD specialist

Troubleshooting

Q: Persona isn't activating?

A: Check:

  1. File exists: ls ~/bot/skills/personas/data/[name].md
  2. Correct name: Must match filename (without .md)
  3. Try exact phrasing: "Use dev" or "Switch to dev"

Q: Persona behavior is wrong?

A: Possible issues:

  • Outdated file - Re-read the persona: "Reload dev persona"
  • Context pollution - Exit and re-enter: "Exit persona" then "Use dev"
  • File corrupted - Check file integrity, restore from backup

Q: Created custom persona but can't activate it?

A: Ensure:

  1. File saved to data/your-name.md
  2. Follows template structure (headers matter!)
  3. No special characters in filename (use hyphens: game-master.md)

Q: Persona responses too long/short?

A: Edit the COMMUNICATION STYLE section in the persona's .md file. Add guidance like:

  • "Keep responses under 3 paragraphs"
  • "Use bullet points for clarity"
  • "Provide detailed explanations with examples"

Q: Two personas have overlapping expertise - which to use?

A: Examples:

  • Dev vs Chameleon Agent - Dev for coding focus, Chameleon Agent for complex multi-domain problems
  • Wordsmith vs Luna - Wordsmith for editing/structure, Luna for creative brainstorming
  • Scholar vs Herr Müller - Scholar for study techniques, Herr Müller for simple explanations

When in doubt, ask: "Should I use Dev or Chameleon Agent for this?"


Advanced

Q: Can I chain personas in one request?

A: No direct chaining, but you can:

"Use Luna to brainstorm, then switch to Wordsmith to refine"

I'll switch personas mid-task.

Q: Can personas collaborate?

A: Sort of! Use The Panel persona - it simulates 4 experts discussing your question from different angles.

Or manually: Get Dev's take, then switch to CyberGuard for security perspective, then Startup Sam for business angle.

Q: Can I export/import personas?

A: Yes!

  • Export: Copy .md files from data/
  • Import: Drop .md files into data/, update skill.json
  • Share via GitHub, Skills, or direct file transfer

Q: Can I use personas in sub-agents?

A: Yes! Spawn a sub-agent with a persona request:

"Spawn a sub-agent using Dev persona to refactor this codebase"

Q: Versioning personas?

A: Use git in your skills directory:

cd ~/bot/skills/personas
git init
git add .
git commit -m "Initial personas"

Then track changes, revert if needed.

Q: Can I monetize custom personas?

A: If you create valuable custom personas, you can:

  1. Publish modified skill to Skills
  2. Share on GitHub (MIT license)
  3. Offer as part of consulting/services

Original 31 personas are MIT licensed (Chameleon AI).


Best Practices

Q: When should I NOT use a persona?

A: For simple, quick tasks where Cami (default adaptive) is sufficient. Don't over-optimize - personas shine for:

  • Specialized expertise needed
  • Specific communication style wanted
  • Complex, domain-specific tasks

Q: Can I use multiple personas in parallel sessions?

A: Yes! Start multiple Bot sessions, each with different personas. Example:

  • Session 1: Dev persona (coding)
  • Session 2: Wordsmith persona (blog writing)
  • Session 3: Chef Marco persona (meal planning)

Q: Should I create many custom personas or modify existing ones?

A: Modify existing ones if:

  • Close to what you want (80%+ match)
  • Small tweaks needed

Create new if:

  • Unique niche expertise
  • Completely different communication style
  • Specific use case not covered

Q: How often should I switch personas?

A: When the task changes domains. Example conversation:

[Cami] General chat
[Dev] "Let's code this feature" → Switch to Dev
[CyberGuard] "How do I secure this?" → Switch to CyberGuard
[Wordsmith] "Now document it" → Switch to Wordsmith
[Cami] "Thanks!" → Exit persona

Philosophy

Q: Why use personas vs. just asking differently?

A: Personas provide:

  • Consistency - Same expert voice throughout task
  • Optimization - Communication style tuned for domain
  • Context - Persona "remembers" they're a chef/dev/coach
  • Efficiency - No need to re-specify expertise each message

Q: Aren't personas just prompt engineering?

A: Yes! But systematized, reusable, and shareable. Instead of crafting custom prompts every time, load pre-optimized personas instantly.

Q: Original Chameleon AI vs Bot personas - differences?

A:

  • Chameleon AI: Web UI, visual elements (colors, themes), chat-focused
  • Bot: CLI/Telegram, tool integration, task automation
  • This skill: Adapted personalities for Bot use cases, removed UI-specific elements, added creator functionality

Still have questions? Ask in persona mode: "Use Professor Stein" then "Explain personas to me" 🎓