Eve library

Psyche Persona Service

Other

Revision 1

On this page

Persona management service for the Psyche AI Virtual Assistant Platform.

Overview#

The Persona Service provides comprehensive AI agent persona management capabilities:

  • Identity Management: CRUD operations, versioning, templates
  • Personality Configuration: Big Five model, communication styles
  • Expertise Configuration: Domains, knowledge boundaries, scope limits
  • Memory Systems: Facts, opinions, story memory, user interactions
  • Behavioral Constraints: Response validation, drift prevention
  • Asset Management: Avatar and voice profile creation, training

Architecture#

The service consists of two packages:

text
src/
├── persona_service/      # FastAPI REST API
│   ├── __init__.py
│   └── main.py           # API endpoints
└── persona_system/       # Core business logic
    ├── __init__.py       # PersonaSystem unified interface
    ├── identity/         # Identity management
    ├── personality/      # Personality configuration
    ├── expertise/        # Expertise configuration
    ├── memory/           # Memory systems
    ├── constraints/      # Behavioral constraints
    ├── drift/            # Drift detection
    └── assets/           # Avatar and voice assets

API Endpoints#

Health & Status#

text
GET /health              # Service health
GET /ready               # Readiness check

Templates#

text
GET /templates/personas      # List persona templates
GET /templates/personalities # List personality presets
GET /templates/domains       # List expertise domain templates
GET /templates/constraints   # List constraint presets

Persona CRUD#

text
POST /personas                           # Create persona
GET /personas/{id}                       # Get persona
PUT /personas/{id}                       # Update persona
DELETE /personas/{id}                    # Delete persona
GET /personas                            # List personas
POST /personas/{id}/clone                # Clone persona
POST /personas/from-template/{template}  # Create from template

Personality#

text
GET /personas/{id}/personality           # Get personality config
POST /personas/{id}/personality          # Configure personality

Expertise#

text
GET /personas/{id}/expertise             # Get expertise config
POST /personas/{id}/expertise            # Configure expertise
POST /personas/{id}/expertise/domains    # Add domain
POST /personas/{id}/expertise/evaluate   # Evaluate query expertise

Memory#

text
POST /personas/{id}/memories/facts       # Add fact
POST /personas/{id}/memories/opinions    # Add opinion
POST /personas/{id}/memories/interactions  # Record interaction
POST /personas/{id}/memories/retrieve    # Retrieve memories

Constraints#

text
GET /personas/{id}/constraints           # Get constraints
POST /personas/{id}/constraints          # Configure constraints
POST /personas/{id}/constraints/check-response  # Validate response
POST /personas/{id}/constraints/check-drift     # Check drift

Assets#

text
POST /personas/{id}/avatars              # Create avatar
POST /personas/{id}/voices               # Create voice
POST /avatars/{id}/train                 # Train avatar
POST /voices/{id}/train                  # Train voice
POST /personas/{id}/assets/link          # Link assets
GET /personas/{id}/assets                # Get asset bundle

System Prompt#

text
POST /personas/{id}/system-prompt        # Generate system prompt

Persona System#

PersonaSystem Class#

The unified interface for persona management:

python
from persona_system import PersonaSystem

# Create system
system = PersonaSystem()

# Create a persona
persona = system.create_persona(
    name="Alex",
    role="assistant",
    personality_preset="warm_professional",
    expertise_domains=[{"template_id": "customer_support"}],
)

# Configure personality
system.configure_personality(
    persona.id,
    big_five={"openness": 0.7, "agreeableness": 0.9},
)

# Add memory
system.add_fact(
    persona.id,
    key="company",
    value="Acme Corp",
)

# Generate system prompt
prompt = system.generate_system_prompt(persona.id)

Personality Presets#

Preset Description
warm_professional Friendly and professional
technical_expert Precise, detail-oriented
empathetic_counselor Supportive, understanding
creative_collaborator Imaginative, enthusiastic
efficient_executor Direct, task-focused

Expertise Domain Templates#

Template Description
customer_support Customer service and support
technical_support Technical troubleshooting
healthcare Health information (non-medical)
education Learning and instruction
finance Financial information

Constraint Presets#

Preset Description
professional Professional workplace standards
transparent Open about AI nature
creative More flexible expression
strict Highly constrained responses

Configuration#

Environment Variables#

Variable Description Default
SERVICE_PORT API port 8009
DATABASE_URL PostgreSQL connection postgresql://...
REDIS_URL Redis connection redis://localhost:6379
QDRANT_URL Vector DB connection http://localhost:6333
LOG_LEVEL Log level INFO

Development#

Using Nx#

bash
# Install dependencies
nx install psyche-persona-service

# Run development server
nx serve psyche-persona-service

# Run production server
nx serve-prod psyche-persona-service

# Run tests
nx test psyche-persona-service
nx test-unit psyche-persona-service
nx test-integration psyche-persona-service

# Run tests with coverage
nx test-cov psyche-persona-service

# Lint and format
nx lint psyche-persona-service
nx format psyche-persona-service

# Docker
nx docker-build psyche-persona-service
nx docker-run psyche-persona-service

Direct Poetry Commands#

bash
cd apps/psyche/persona-service
poetry install
poetry run pytest
poetry run uvicorn persona_service.main:app --reload --host 0.0.0.0 --port 8009

Performance Targets#

Operation Target Latency
Create persona < 50ms
Get persona < 10ms
Configure personality < 20ms
Add memory < 15ms
Retrieve memories < 100ms
Generate prompt < 200ms
Check drift < 50ms

License#

Proprietary - Oshun Platform