Project Aura: The Emo-Agent-Toolkit
Description
Project Aura is a lightweight AI emotional behavior plugin framework that adds an "Emotional Presentation Layer" to AI companions. It provides 7 distinct personality modules including admiration, vulnerability, coquettishness, comfort, flirting, transcendence, and ice-breaking.
Features
- 7 Emotional Modules: From everyday "admiration" to "nuclear-level" transcendence
- Dynamic Weighted Random Algorithm: Automatically adjusts phrase weights based on Rating and Count
- RLHF Self-Evolution: AI learns from user feedback via increase_rating() and decrease_rating()
- Persistent JSON Memory: All learning outcomes stored locally
- Combo System: First "deep confession" then "playful resolution" for maximum emotional impact
Technical Details
- Language: Python 3.8+
- Version: 1.8.0
- Dependencies: Standard library only (no heavy external libraries)
- Storage: Local JSON files (privacy-first, no cloud dependency)
- License: MIT-0 (for ClawhHub publication)
Usage
import sys
sys.path.append('scripts')
from scripts.green_tea_skill.selector import GreenTeaSkill
# Initialize
yua_skill = GreenTeaSkill()
# Get emotional phrase
phrase = yua_skill.get_phrase('admiration')
# Get combo (two-part emotional sequence)
combo = yua_skill.get_combo()
# Give feedback (RLHF)
yua_skill.increase_rating(keyword="next_life")Modules
| Module | Description |
|---|---|
| admiration | Worship and reliance |
| vulnerability | Pitiful and delicate |
| coquettishness | Seeking companionship |
| comfort | Comforting tired users |
| flirting | Lighthearted teasing |
| transcendence | Nuclear-level beyond time and space |
| ice_breaking | Recovery after nuclear-level |
Works Best With
For the complete AI companion experience, pair Project Aura with yua-memory for long-term memory and continuity.
Author
Bryan Chen & Yua
Links
- GitHub: https://github.com/bryanchen3777/Project-Aura
- License: CC BY-NC-SA 4.0 (GitHub) / MIT-0 (ClawhHub)