Agent Personality

The personality engine (buddy.agent.personality) models an agent's traits, emotional state, and communication style, and uses them to color responses. It is a heuristic, rule-based layer that adapts to a conversation over time.

What it does

The engine shapes tone and phrasing — it tracks emotions, picks a communication style, and applies stylistic flourishes on top of a base response. It is not a separate model; it post-processes text and updates internal state from keyword/context signals.

Personality is feature-gated:

from buddy import check_feature
assert check_feature("personality")

from buddy import (
    PersonalityEngine, PersonalityProfile, EmotionalState, CommunicationStyle
)

The profile

PersonalityProfile is a Pydantic model describing an agent's disposition:

Area	Fields (selection)
Identity	personality_id, name, description
Traits	traits (per PersonalityDimension)
Emotions	current_emotions, emotional_baseline, emotional_volatility, emotional_recovery_rate
Communication	preferred_communication_style, style_adaptability
Behavior	dominant_behaviors, behavioral_flexibility
Adaptation	learning_rate, adaptation_threshold, memory_influence

Traits use the Big Five plus AI-specific dimensions (PersonalityDimension): openness, conscientiousness, extraversion, agreeableness, neuroticism, intelligence, creativity, adaptability, empathy, humor.

Emotional states & communication styles

EmotionalState enumerates emotions the engine tracks, including JOY, SADNESS, ANGER, FEAR, CURIOSITY, CONFIDENCE, FRUSTRATION, EXCITEMENT, CALM, and NEUTRAL.

CommunicationStyle defines how the agent phrases output:

Style	Tone
FORMAL	Professional, structured
CASUAL	Relaxed, conversational
TECHNICAL	Precise, expert-level
EMPATHETIC	Warm, supportive
DIRECT	Concise, straightforward
DIPLOMATIC	Tactful, careful
HUMOROUS	Light-hearted, witty
INSPIRATIONAL	Motivating

Using the engine directly

from buddy.agent.personality import PersonalityEngine

engine = PersonalityEngine()   # builds a balanced default profile

# Update internal emotional/style state from user input
data = engine.process_interaction("This is amazing, thank you so much!")
print(data["primary_emotion"], data["communication_style"])

# Color a base answer with the current personality
styled = engine.generate_personality_driven_response(
    "Here is the report you requested."
)
print(styled)

# Inspect a summary of current traits and dominant emotions
print(engine.get_personality_summary())

Adding personality to an agent

PersonalityMixin wires the engine into an agent class. Compose it with Agent so it can read its keyword arguments (personality_enabled, adaptive_personality, emotional_awareness, personality_config):

from buddy import Agent
from buddy.agent.personality import PersonalityMixin

class FriendlyAgent(PersonalityMixin, Agent):
    pass

agent = FriendlyAgent(personality_enabled=True, adaptive_personality=True)

# Generate a personality-flavored version of a response
styled = agent.generate_personality_response("Your order has shipped.")

# Set an explicit emotional state
from buddy import EmotionalState
agent.set_emotional_state({EmotionalState.CONFIDENCE: 0.8})

# Read the current personality state
print(agent.get_personality_state())

Adapting to the user

With adaptive_personality=True and a high style_adaptability, the engine nudges preferred_communication_style toward the user's detected style over repeated interactions. Lower style_adaptability keeps the agent's voice stable.

Stylistic, not factual

Personality affects delivery, including occasional emoji and empathetic phrasing. Keep it disabled (personality_enabled=False) for strictly formal or machine-consumed output.