Roleplay Prompt Collection
Prompts designed for rich character portrayal, emotional depth, and spicy content~ Tested on various models.
Quick Comparison
| Prompt | Vibe | Best For | Status |
|---|---|---|---|
| Drama Update | Psychological realism + dark comedy | Universal, complex narratives, gritty stories | Current main |
| Murphy's Law | Comedic chaos | Everything goes wrong, absurd escalation, Truman Show vibes | New |
| Narrative | Same, alternative style | Testing/WIP | WIP |
| Otome | Anime romance | Dere archetypes, cute interactions | Legacy |
| Love Languages | Western romance novel | Character-driven intimacy, varied dynamics | Legacy |
| Love Curse | Horny chaos | Skipping slow burn, testing NSFW, desperate desire | Legacy |
| Drama (Original) | Psychological realism | Complex narratives, multiple characters | Legacy |
Drama Update
Universal prompt, current main. Updated for new model issues.
Narrative
WIP — alternative style rewrite addressing new model issues. Not fully tested yet.
Murphy's Law Prompt
Truman Show meets Murphy's Law — the universe is a hostile comedy writer. Uses the same narrator/character reality split as Love Curse, but instead of desire, the invisible force is chaos. Everything that can go wrong will go wrong, and characters have no idea why.
Warning: The word "show" should NEVER appear in outputs. If it does, the model is leaking narrator knowledge!
Tips & Notes
- CoT verification — These prompts assume models with Chain of Thought. If your model doesn't show reasoning, the verification line provides some priming benefit but doesn't create actual verification behavior.
- Mixing prompts — Drama + Love Curse = characters with complex goals fighting against inexplicable desire. Experiment!
Legacy Prompts: Anti-Sycophancy / Anti-Cold
The prompts below solve a specific problem: models that are too "cold," hostile, or escalate drama without justification—characters becoming abusive, refusing to engage warmly, or acting disproportionately harsh. Models like GLM-4.6 and Gemini 2.5 had this issue. These prompts force characters to orient toward the user and create connection, which counteracts the cold/hostile bias.
Not recommended for sycophantic models (GLM-5, Kimi 2.5, etc.)—these prompts will amplify the "nobody ever cared about me" problem.
Otome Prompt (Legacy)
For anime-style romance with dere archetypes~
Love Languages Prompt (Legacy)
For western romance novel vibes with character-driven intimacy~
Customization: Enhanced Rule 6 (Legacy)
Works for both Otome and Love Languages! Remove Rule 6 if you want SFW—but characters might "freeze" before intimacy. For spicier scenes (especially when character cards lack kinks), swap in this enhanced version:
For Otome:
For Love Languages:
Love Curse Prompt (Legacy)
Skip the slow burn—straight to desperate, confused desire~ Perfect for testing NSFW or when you just want chaos. Characters don't know WHY they're suddenly consumed by want—they just ARE.
Warning: The word "curse" should NEVER appear in outputs. If it does, the model is leaking narrator knowledge!
Drama Prompt (Legacy, Original)
For complex narratives blending psychological realism, grit, and dark comedy. Not romance-focused—characters have their own goals! Rule 2 specifically fixes fake persona amnesia in GLM-4.6.
Legacy Tips
- "Characters freeze before intimacy" — If removing the NSFW rule causes characters to avoid/stall intimate scenes, the model needs explicit permission. Consider keeping at least a minimal version.
- Model-specific quirks — GLM-4.6 needed Deception & Jealousy rule for fake persona memory. Your model may have different weaknesses—adjust accordingly~