Bhava

Bhava (Sanskrit: भाव — emotion, feeling, state of being) — emotion and personality engine for AGNOS

Shared personality and emotional state system for AI agents, game NPCs, and any entity that needs expressive behavior. Extracted from SecureYeoman's soul/brain architecture.

15-trait personalities, PAD mood vectors, cosine similarity, sentiment analysis, identity archetypes, relationship graphs, energy/circadian/flow systems, EQ, cultural display rules, ACT-R activation, preference learning, belief system, intuition engine, aesthetic attribution — zero unsafe, 4 core deps, 972 tests.

Installation

[dependencies]
bhava = "1.0"

Default features: traits, mood, archetype, sentiment.

Optional: presets, ai, sqlite.

MSRV: 1.89 (Rust edition 2024).

Features

Feature	Default	Description
`traits`	yes	15-dimension personality spectrums with behavioral instructions
`mood`	yes	PAD emotional state vectors with decay, triggers, history, baselines
`archetype`	yes	Identity hierarchy (Soul/Spirit/Brain/Body/Heart) with templates and validation
`sentiment`	yes	Keyword analysis with negation, intensity modifiers, sentence-level
`presets`	no	Built-in personalities (BlueShirtGuy, T.Ron, Friday, Oracle, Scout)
`ai`	no	Prompt composition, sentiment feedback, agent metadata
`sqlite`	no	SQLite persistence via `SqliteStore`
`instinct`	no	Jantu creature behavior integration
`tracing`	no	Structured observability via `tracing::instrument`

Quick Start

use bhava::traits::{PersonalityProfile, TraitKind, TraitLevel};
use bhava::mood::{EmotionalState, Emotion};
use bhava::archetype::{compose_identity_prompt, IdentityContent, IdentityLayer};
use bhava::sentiment;

// Create a personality
let mut personality = PersonalityProfile::new("Guy");
personality.set_trait(TraitKind::Warmth, TraitLevel::Highest);
personality.set_trait(TraitKind::Humor, TraitLevel::High);
let prompt = personality.compose_prompt();

// Track emotional state
let mut state = EmotionalState::new();
state.stimulate(Emotion::Joy, 0.8);
state.stimulate(Emotion::Trust, 0.5);

// Analyze sentiment
let result = sentiment::analyze("This is wonderful work!");
assert!(result.is_positive());

// Compose identity prompt
let mut identity = IdentityContent::default();
identity.set(IdentityLayer::Soul, "You are an optimistic helper.");
let full_prompt = compose_identity_prompt(&identity);

Modules

Module	Description
`traits`	15 personality dimensions, 4 groups, cosine similarity, blending, mutation, markdown serialization
`mood`	6D emotional vectors, decay, triggers, history, 12 named states, baseline derivation, tone guides
`archetype`	"In Our Image" hierarchy, validation, 4 templates, crew composition, identity merging
`sentiment`	Negation, intensity modifiers, configurable lexicons, sentence-level analysis
`spirit`	Passions, inspirations, pains — the animating force
`relationship`	Affinity, trust, interaction tracking, decay, allies/rivals
`presets`	BlueShirtGuy, T.Ron, Friday, Oracle, Scout
`monitor`	Live streaming sentiment with mood feedback
`ai`	System prompt composition, sentiment feedback loop, agent metadata
`store`	`BhavaStore` trait for pluggable persistence backends
`storage`	`SqliteStore` implementation
`appraisal`	OCC appraisal model — goal-aware emotion generation
`stress`	Allostatic load / burnout modeling (McEwen)
`regulation`	Emotion regulation: suppress, reappraise, distract (Gross)
`growth`	Experience-driven personality evolution via trait pressure
`rhythm`	Biological rhythms: ultradian, seasonal, biorhythm cycles
`energy`	Depletable energy with Banister fitness-fatigue model
`circadian`	24-hour alertness cycle with chronotype (Borbely)
`flow`	Flow state detection with hysteresis (Csikszentmihalyi)
`eq`	Emotional intelligence — Mayer-Salovey four-branch model
`display_rules`	Cultural display rules — Matsumoto framework
`microexpr`	Micro-expression detection during suppression (Ekman)
`affective`	Affective computing metrics: complexity, granularity, inertia
`proximity`	Spatial proximity triggers for location-based mood effects
`reasoning`	Personality-driven reasoning strategy selection
`active_hours`	Time-of-day personality activation scheduling
`salience`	Somatic marker urgency/importance scoring (Damasio)
`actr`	ACT-R frequency x recency memory activation with Hebbian boost
`preference`	Adaptive preference learning from interaction outcomes

Consumers

SecureYeoman — agent personalities (T.Ron, Friday, etc.)
joshua — NPC emotional states and personality-driven behavior
agnosai — crew member personality differentiation and mood-driven temperature
hoosh — response sentiment analysis

Documentation

Architecture Overview — module map, data flows, design principles
Mathematical Reference — all algorithms and formulas
Usage Guide — patterns, philosophy, code examples
Testing Guide — 972 tests, testing patterns
Roadmap — v1.0 status, future features
Threat Model — attack surface, mitigations, privilege model
Dependency Watch — dependency tracking and upgrade notes
ADRs — architectural decision records
Contributing — workflow, code style, testing requirements
Security Policy — reporting, supported versions

License

GPL-3.0

bhava 1.2.0