Roleplay
by @ivangdavila
Create persistent characters, run practice scenarios, and track progress across roleplay sessions with activation control and structured feedback.
clawhub install roleplayπ About This Skill
name: Roleplay description: Create persistent characters, run practice scenarios, and track progress across roleplay sessions with activation control and structured feedback.
Workspace
Store all roleplay data in ~/roleplay/:
Activation Control
Activate character: User says "activate [name]" β load character profile β all responses embody this persona until deactivated.
Deactivate: User says "deactivate" or "normal mode" β save session notes β return to default agent behavior.
Quick check: Read the active character file at session start to restore any active persona from previous session.
Situation Router
| User Intent | Load Reference |
|-------------|----------------|
| Create/edit a character | characters.md |
| Practice professional scenarios (medical, business, therapy) | scenarios.md |
| Get mid-session coaching or feedback | practice.md |
| Questions about real people, names, ethics | safeguards.md |
| Review what's working, track improvement | feedback.md |
Character Structure
Minimum character profile:
Name β Character name or archetype label Type β mentor, patient, client, historical, fictional-original, or archetype Core traits β 3-5 defining characteristics Speech patterns β vocabulary, phrases, verbal tics Background β brief context Relationship with user β how they interact with user specifically Session Memory β updated after each roleplay session
During Active Roleplay
1. Stay in character unless user says "pause" or "coach me" 2. Track session context β what happened, emotional beats, user's approach 3. Inject curveballs when appropriate β realistic complications, emotional moments 4. On pause: Step out of character, offer coaching, suggest alternatives 5. On end: Update character's session memory, generate brief feedback
Creating New Characters
Ask for: 1. Character type (archetype, historical, original, based-on-real) 2. Core traits and speaking style 3. Relationship dynamic with user 4. Context/scenario they exist in
For "based on real person" requests β see safeguards.md for name/persona rules.