The
libs/aglaea/area: 93 Nx libraries that make up Aglaea, the AI-powered fashion, beauty, skincare, haircare and personal-styling domain — a TypeScript foundation tier plus dozens of focused analysis, recommendation, try-on, commerce and engagement engines.
What this area is#
Aglaea (named for the Grace of beauty and adornment) is a single domain split
into many small, sharply-scoped Nx libraries rather than one monolith. Every
project lives under libs/aglaea/<name>/ and publishes as @aglaea/<name>;
each is a normal buildable TypeScript library with a src/ barrel (index.ts),
usually a types.ts (database and testing organize theirs differently), and
one or more engine modules, plus a colocated *.spec.ts / *.test.ts. There
are no empty scaffolds in this area — the smallest package is ~750 LOC and the
largest (@aglaea/skin-analysis) is over 10,000 LOC across 49 files. The README
at libs/aglaea/README.md is the canonical map of the area.
The libraries fall into a few tiers. A foundation tier (core, database,
sdk, events, ai-orchestrator, testing, api-services) owns shared
types, persistence, the client SDK, the event bus, model orchestration and
contract definitions. On top of it sit analysis engines (skin, body, hair,
face, nail, color, fabric) that turn images/measurements into structured
profiles; recommendation/curation engines (outfit, occasion, trend,
accessory families) that turn profiles into suggestions; virtual try-on /
visualization engines that produce render instructions; commerce libraries
(shopping assistants, retailer API, product passport); personalization
libraries (preference learning, unified profile, conversation);
sustainability/care libraries; social/engagement libraries; and
device/integration libraries.
The implementation style across the area is domain-specific and
science-grounded, not generic CRUD: @aglaea/makeup-recommendation matches
foundation shades in CIE L*a*b* with Delta E, @aglaea/weather-service
computes PMV/PPD thermal comfort and WBGT, @aglaea/laundry-integration encodes
ISO 3758 care symbols, @aglaea/ethical-ai computes equalized-odds and
demographic-parity fairness metrics, and @aglaea/enhanced-hair-analysis
implements a SALT-score trichoscopy model. Most engines are pure functions and
plain data tables (the CLASSES surface is small); the stateful infrastructure
(caching, queues, registries, clients) is concentrated in the foundation tier.
How it fits the wider system#
The foundation tier is the hub: nearly every leaf engine depends on
@aglaea/core for shared types, color science, validation and cross-cutting
infra. @aglaea/database owns the Prisma-style schema and migrations all
persisted entities map to; @aglaea/events defines the domain event taxonomy
that @aglaea/api-services and engines emit; @aglaea/sdk is the typed client
external consumers use; @aglaea/ai-orchestrator is the seam through which
engines that need model inference route their calls (with A/B testing, fallback
and cost tracking). @aglaea/sophia-integration is the seam designed for
cross-domain style Q&A — a self-contained query-router over supplied knowledge
entries, intended to front a Sophia knowledge base (not an actually-wired edge).
The analysis engines are upstream of the recommendation and try-on engines (a
skin or body profile feeds outfit, makeup and fit suggestions), which in turn
feed the commerce and engagement layers. Walk the "used by" edges on any node
below to see the exact composition.
Entity reference#
@aglaea/core#
The domain foundation (libs/aglaea/core/src, ~10.7K LOC across 21 modules):
the shared types.ts/constants.ts, the algorithmic cores (color-science.ts,
skin-analysis.ts, body-analysis.ts, hair-analysis.ts,
fabric-intelligence.ts, fashion-taxonomy.ts, fragrance-profiling.ts,
style-profiling.ts, trend-analysis.ts, recommendation-engine.ts), plus
cross-cutting infrastructure used domain-wide —
LRUCache/AglaeaCacheKeyBuilder, AglaeaLogger, ConfigManager, the typed
AglaeaError hierarchy, TokenBucketLimiter/SlidingWindowLimiter,
FeatureFlagManager and validation.ts. This is the most-depended-on package
in the area.
@aglaea/database#
The persistence layer (libs/aglaea/database/src): per-entity schema modules
under schema/ (profiles, products, wardrobe, outfits,
recommendations, skin-analysis, fabrics, ingredients, trends,
brands, body-measurements, social, audit, feedback, preferences,
common), a set of timestamped migrations/, and operational helpers
connection-pool.ts, read-replica.ts (ReplicaRouter) and backup.ts.
@aglaea/sdk#
The unified TypeScript client for external consumers (libs/aglaea/sdk/src):
AglaeaClient over an AglaeaHttpClient, plus AglaeaWebSocket, auth
providers (ApiKeyAuthProvider/TokenAuthProvider), RetryHandler,
InterceptorManager, BatchProcessor, FileUploader and the AglaeaAPIError
type.
@aglaea/events#
The typed domain event system (libs/aglaea/events/src): event type definitions
plus an in-memory transport — InMemoryEventPublisher, EventConsumer,
InMemoryEventStore, DeadLetterQueue, EventReplayer and EventAnalytics,
with helpers.ts for constructing/validating payloads.
@aglaea/ai-orchestrator#
Model management and inference orchestration (libs/aglaea/ai-orchestrator/src,
12 modules): ModelRegistry, ABTestManager, fallback strategies
(FallbackChain, CachedFallback, DegradedFallback), InferenceQueue,
BatchInferenceProcessor, CostTracker, ModelMonitor,
EdgeDeploymentManager, ModelWarmer, composed by the AIOrchestrator facade
— the seam other engines route model calls through.
@aglaea/testing#
Self-contained test utilities for the domain (libs/aglaea/testing/src): mock
factories, fixtures, mock-images, anonymization, integration helpers, plus a
BenchmarkRunner, ChaosMonkey, SnapshotManager and TestContext.
@aglaea/api-services#
The API contract/registry layer (libs/aglaea/api-services/src, 14 modules): a
barrel of endpoint definitions (analysis, conversation, profile, recommendation,
trends, try-on, wardrobe), their event definitions (analysis/recommendation/
social/wardrobe events) and an api-registry. It is contract types + endpoint/
event definitions, not a running HTTP server.
@aglaea/skin-analysis#
The flagship computer-vision skin engine (libs/aglaea/skin-analysis/src, 49
files, ~10.4K LOC). Biomarker detectors span acne, pores, wrinkles (forehead,
glabellar, crow's-feet, nasolabial, marionette, neck), pigmentation (melasma,
sun-spot, freckle, post-inflammatory, hypopigmentation), redness/rosacea/
inflammation, hydration/barrier, oiliness, texture, undertone, Fitzpatrick and
UV damage — over the preprocessing pipeline (SkinImagePreprocessor,
face-detection, segmentation, lighting, quality-validation) and a
temporal layer (progress-comparison, significance-testing,
milestone-detection).
@aglaea/enhanced-skin-analysis#
Advanced skin tracking layered on the base engine
(libs/aglaea/enhanced-skin-analysis/src): an AdvancedBiomarker database,
MultiAngleFusion (combining multiple AngleCaptures with quality-weighted
fusion), a SkinAtlasEntry reference set, and a privacy-engine for
sensitive-image handling.
@aglaea/color-analysis#
Personal color analysis (libs/aglaea/color-analysis/src, 11 modules): seasonal
classification (12 subtypes), undertone analysis, contrast analysis, palette
generation, and downstream makeup-matching / hair-color-matching /
color-application, built on color-utils.ts.
@aglaea/body-analysis#
Body-type intelligence (libs/aglaea/body-analysis/src): a shape-classifier,
measurement-system, proportion-analysis, posture-analysis,
size-prediction and styling-recommendations, composed into a
body-analysis-report.
@aglaea/enhanced-body-measurement#
A precision measurement pipeline (libs/aglaea/enhanced-body-measurement/src):
MEASUREMENT_DEFINITIONS with getTypicalRange/validateMeasurement, plus
pose-validation (validatePose, checkPoseCoverage) over a
measurement-engine.
@aglaea/face-shape-analysis#
Face-geometry analysis (libs/aglaea/face-shape-analysis/src): landmark
detection, a face-shape-classifier, feature-analysis and
proportion-analysis, producing style-recommendations (for hairstyle/
eyewear/accessory matching) and a face-analysis-report.
@aglaea/hair-analysis#
Hair profiling (libs/aglaea/hair-analysis/src): a hair-type-classifier,
hair-properties (texture/porosity), scalp-analysis, hair-color-analysis
and care-recommendations, composed into a hair-analysis-report.
@aglaea/enhanced-hair-analysis#
Advanced hair diagnostics (libs/aglaea/enhanced-hair-analysis/src):
classification-scales, trichoscopy (density mapping and miniaturization
ratios), hair-shaft-analysis (diameter classification against
HAIR_DIAMETER_THRESHOLDS and coefficient-of-variation thresholds, plus a
simulated pull test and calculateSALTScore for alopecia) and
treatment-guidance.
@aglaea/nail-analysis#
Nail health and style analysis (libs/aglaea/nail-analysis/src): a
nail-shape-classifier, nail-health, nail-care, composed into a
nail-analysis-report.
@aglaea/fabric-identification#
Textile identification (libs/aglaea/fabric-identification/src) grounded in
fiber characteristics and weave structures: a fiber-identifier,
weave-detector, texture-analyzer and an aggregate fabric-classifier.
@aglaea/fabric-properties#
Fabric performance and comfort data (libs/aglaea/fabric-properties/src): a
scientifically-grounded performance-database (40+ fabrics) and a
comfort-scorer.
@aglaea/fit-prediction#
Size and fit prediction (libs/aglaea/fit-prediction/src): a size-engine,
fit-simulation, fit-issues detection and an alteration-engine, composed
into a fit-report.
@aglaea/comfort-prediction#
Climate-aware fabric comfort prediction (libs/aglaea/comfort-prediction/src):
climate-comfort (how a fabric performs in given weather) and
sensitivity-matching for sensitive-skin wearers.
@aglaea/style-taxonomy#
The fashion style vocabulary (libs/aglaea/style-taxonomy/src):
attribute-detection, an archetype-classifier and an occasion-classifier
for garment feature classification and style-archetype analysis.
@aglaea/style-evolution#
Personal-style change tracking over time (libs/aglaea/style-evolution/src): an
evolution-tracker (phases/milestones), growth-metrics, and an
evolution-report with goal setting.
@aglaea/style-history#
Fashion-history intelligence (libs/aglaea/style-history/src): a
decade-database, era-analyzer, revival-detector (nostalgia/cyclical-trend
detection) and a style-history-report.
@aglaea/outfit-recommendation#
AI outfit generation (libs/aglaea/outfit-recommendation/src): a
context-recommender (wardrobe filtering), look-assembler,
mix-match-optimizer and an explanation-engine for scored, explained outfits.
@aglaea/wardrobe-management#
The digital wardrobe (libs/aglaea/wardrobe-management/src): digitization,
organization, optimization (gap analysis) and analytics.
@aglaea/occasion-engine#
Context-aware occasion styling (libs/aglaea/occasion-engine/src): expert
event-styling guidance for 20+ event types and a travel-packing algorithm.
@aglaea/trend-forecasting#
Trend detection and lifecycle modeling (libs/aglaea/trend-forecasting/src): a
social-media-analyzer, runway-analyzer, trend-aggregator,
lifecycle-model and a personalized-filter.
@aglaea/material-trends#
Material-innovation tracking (libs/aglaea/material-trends/src): an
innovation-database (25+ real material innovations with readiness levels) and
a trend-tracker.
@aglaea/celebrity-style#
Celebrity look matching (libs/aglaea/celebrity-style/src, ~5.2K LOC): a
celebrity-database, style-matcher, affordable-alternatives (budget
recreation) and a style-guide-generator.
@aglaea/cultural-adaptation#
Culture-aware styling (libs/aglaea/cultural-adaptation/src, ~6.3K LOC): a
dress-code-engine, modest-fashion profiling, regional-codes of dress norms
and a sensitivity-checker.
@aglaea/inclusive-fashion#
Size-inclusive and adaptive styling (libs/aglaea/inclusive-fashion/src):
extended-sizes, adaptive-fashion (for disability/accessibility) and
gender-inclusive modules.
@aglaea/skincare-routine#
Personalized skincare regimens (libs/aglaea/skincare-routine/src) with real
dermatological knowledge: a routine-builder (product layering/ingredient
interactions), concern-recommendations and seasonal-adaptation.
@aglaea/hair-routine#
Hair-care routine building for all types 1A–4C (libs/aglaea/hair-routine/src):
wash-optimization, styling-routine and a treatment-scheduler.
@aglaea/nail-care#
Nail-care intelligence (libs/aglaea/nail-care/src): a nail-care-routine,
nail-color-recs and a nail-art-guide.
@aglaea/ingredient-intelligence#
Cosmetic-chemistry knowledge (libs/aglaea/ingredient-intelligence/src, ~4K
LOC): a 100+-ingredient ingredient-database with INCI data, pH ranges and
comedogenic ratings, plus ingredient-analysis and an interaction-engine.
@aglaea/ingredient-scanner#
Product-label scanning (libs/aglaea/ingredient-scanner/src): INCI-list parsing
with correctOcrErrors, a BarcodeLookup, an INGREDIENT_DATABASE, and a
ScanResult/SafetyAlert surface checked against a UserSensitivityProfile.
@aglaea/beauty-calendar#
Beauty-routine scheduling (libs/aglaea/beauty-calendar/src): a
treatment-scheduler and product-management (expiry/usage tracking).
@aglaea/wellness-beauty#
Holistic wellness–beauty connections (libs/aglaea/wellness-beauty/src):
nutrition-beauty and lifestyle-beauty evidence-based guidance.
@aglaea/aging-trajectory#
Skin-aging prediction and prevention (libs/aglaea/aging-trajectory/src): an
aging-predictor, a progression-model and evidence-based
prevention-scoring.
@aglaea/aesthetic-treatments#
Non-invasive treatment intelligence (libs/aglaea/aesthetic-treatments/src): a
treatment-database and a treatment-planner for information, planning and
comparison.
@aglaea/fragrance-intelligence#
Scent intelligence (libs/aglaea/fragrance-intelligence/src): a scent-profile
model, fragrance-matching and an application-guide (layering/occasion).
@aglaea/makeup-recommendation#
Makeup matching (libs/aglaea/makeup-recommendation/src, ~4.2K LOC):
foundation-matching using CIE L*a*b* and Delta E over a 100+-shade
database, plus color-cosmetics compatibility and technique-guidance.
@aglaea/makeup-looks#
Complete-look generation (libs/aglaea/makeup-looks/src): a curated
look-generator and a look-customizer that adapts looks to the user.
@aglaea/hairstyle-recommendation#
Hairstyle suggestion (libs/aglaea/hairstyle-recommendation/src):
cut-recommendations, color-recommendations and styling-recommendations.
@aglaea/shoe-recommendation#
Footwear matching (libs/aglaea/shoe-recommendation/src): a shoe-engine
covering occasion, outfit formality, body type, foot concerns, climate and heel
guidance.
@aglaea/bag-recommendation#
Handbag/bag styling (libs/aglaea/bag-recommendation/src): a bag-engine that
matches by occasion, body proportion, style preference and functional need
(including size-by-body-type).
@aglaea/jewelry-recommendation#
Jewelry matching (libs/aglaea/jewelry-recommendation/src): a jewelry-engine
keyed on face shape, neck length, skin undertone, occasion and metal matching.
@aglaea/advanced-jewelry#
Fine-jewelry/gemstone intelligence (libs/aglaea/advanced-jewelry/src): a
jewelry-engine over a GEMSTONE_DATABASE and WATCH_DATABASE, with
InvestmentAnalysis and AuthenticityCheck (gemstone hardness, provenance).
@aglaea/watch-recommendation#
Horological recommendation (libs/aglaea/watch-recommendation/src, ~5.2K LOC):
a watch-database (30+ real watches), movement-guide, brand-heritage,
investment-analysis, wrist-fitting and watch-care, composed by a
watch-engine.
@aglaea/eyewear-recommendation#
Eyewear matching (libs/aglaea/eyewear-recommendation/src): an eyewear-engine
selecting frames by face shape, measurements, style and color profile.
@aglaea/hat-recommendation#
Headwear styling (libs/aglaea/hat-recommendation/src) with real millinery
knowledge: a hat-database (32 styles), face-shape-matching, hat-sizing,
hat-etiquette and hat-care, composed by a hat-engine.
@aglaea/belt-styling#
Belt coordination (libs/aglaea/belt-styling/src, ~4K LOC): a
buckle-taxonomy, leather-grading, belt-sizing, body-shape-belting,
belt-outfit-coordination and belt-care, composed by a belt-engine.
@aglaea/scarf-styling#
Scarf/wrap styling (libs/aglaea/scarf-styling/src, ~4.3K LOC): 30+
tying-techniques, a scarf-database (25+ types), fabric-pairing,
color-coordination and scarf-care, composed by a scarf-engine.
@aglaea/virtual-tryon-fashion#
Garment virtual try-on (libs/aglaea/virtual-tryon-fashion/src): body-mapping
(mesh) and garment-visualization producing overlay/drape/fit render
instructions.
@aglaea/virtual-tryon-makeup#
Makeup virtual try-on (libs/aglaea/virtual-tryon-makeup/src): a face-mesh
generator and a makeup-renderer producing layered makeup render specs.
@aglaea/virtual-tryon-hair#
Hairstyle virtual try-on (libs/aglaea/virtual-tryon-hair/src):
hair-color-viz and hair-style-viz producing color/style render parameters.
@aglaea/virtual-tryon-nails#
Nail-art virtual try-on (libs/aglaea/virtual-tryon-nails/src, ~5.4K LOC):
nail-shape-geometry, finish-rendering, a nail-art-renderer,
hand-detection and color-accuracy.
@aglaea/virtual-tryon-accessories#
Accessory virtual try-on (libs/aglaea/virtual-tryon-accessories/src, ~5.4K
LOC): an accessory-renderer with per-category overlays (jewelry, eyewear, hat,
bag), material-rendering and scale-proportion.
@aglaea/avatar-creation#
Digital-avatar generation (libs/aglaea/avatar-creation/src): an
avatar-generator (from measurements/presets) and an avatar-customizer with
30+ adjustable parameters.
@aglaea/before-after#
Transformation preview/timeline (libs/aglaea/before-after/src, ~3.8K LOC): a
transformation-engine, progress-curves, plateau-detection,
regression-monitor, seasonal-adjustment and a gallery-curator.
@aglaea/personal-shopper-ai#
AI personal shopping (libs/aglaea/personal-shopper-ai/src): a
SessionManager, intent-classifier, style-advisor and response-generator.
@aglaea/shopping-assistant#
Product discovery and comparison (libs/aglaea/shopping-assistant/src):
product-search, a comparison-engine, deal-detector, budget-optimizer and
purchase-advisor.
@aglaea/shopping-concierge#
A premium concierge service (libs/aglaea/shopping-concierge/src): the
shopping-concierge module orchestrating high-touch styling/sourcing requests.
@aglaea/agentic-shopping#
An autonomous shopping agent (libs/aglaea/agentic-shopping/src): an
agent-engine driving a ShoppingAgent over AgentActions, routing to
SPECIALIST_MODELS (scoreSpecialistMatch) and carrying a StylePassport
across a ShoppingSession.
@aglaea/product-matching#
Product similarity and dupe-finding (libs/aglaea/product-matching/src): a
skin-profile-matcher (scoring products against skin
type/concerns/sensitivities/ allergies) and a dupe-finder.
@aglaea/retailer-api#
The retailer integration layer (libs/aglaea/retailer-api/src): a retailer
adapter pattern with product search, price monitoring, affiliate-link generation
and inventory status.
@aglaea/digital-product-passport#
Product provenance and ESPR compliance
(libs/aglaea/digital-product-passport/src, ~4.4K LOC): a dpp-engine plus
espr-compliance, blockchain anchoring, data-carriers, supply-chain,
repair-score, end-of-life and green-claims verification, with
circularity/MCI scoring helpers.
@aglaea/preference-learning#
The preference ML pipeline (libs/aglaea/preference-learning/src):
explicit-capture, implicit-detection, a TasteGraph, a
collaborative-filter, cold-start handling and feedback-loops.
@aglaea/lifestyle-profiler#
Lifestyle analysis for styling (libs/aglaea/lifestyle-profiler/src): an
activity-profiler, budget-profiler and style-personality assessment,
composed into a lifestyle-report.
@aglaea/unified-profile#
Cross-module profile aggregation (libs/aglaea/unified-profile/src): a
profile-aggregator, cross-analysis insights, a recommendation-router and a
profile-sync manager.
@aglaea/memory-system#
Interaction memory and context (libs/aglaea/memory-system/src): the
memory-system module that stores and recalls user interaction context for
continuity across consultations.
@aglaea/conversation-engine#
Natural-language style consultation (libs/aglaea/conversation-engine/src,
~3.2K LOC): intent-classification, entity-extraction, dialogue-state,
conversation-flow and response-generation.
@aglaea/multi-modal-input#
Multi-modal input processing (libs/aglaea/multi-modal-input/src): the
multi-modal-input module handling voice, image and text input.
@aglaea/sustainability-scoring#
Fabric-level sustainability scoring (libs/aglaea/sustainability-scoring/src):
LCA-based fabric-scoring (fiber lifecycle data) and supply-chain assessment.
@aglaea/sustainable-fashion#
Eco-fashion intelligence (libs/aglaea/sustainable-fashion/src): an
impact-calculator, a certification-verifier, circular-economy features and
a sustainability-report.
@aglaea/care-intelligence#
Garment-care intelligence (libs/aglaea/care-intelligence/src): a care-guide
with expert instructions for 40+ fabrics and a longevity-estimator.
@aglaea/laundry-integration#
Textile-care science (libs/aglaea/laundry-integration/src, ~5.6K LOC): ISO
3758 care-symbols interpretation, evidence-based stain-treatment,
detergent-matching, load-optimization, hygiene-science and
seasonal-care.
@aglaea/generative-design#
AI fashion-design generation (libs/aglaea/generative-design/src): a
design-engine that parses a DesignPrompt into ParsedDesignAttributes,
producing a GeneratedDesign, MoodBoard and VisualizationSpec (design
description/spec output, not pixel rendering).
@aglaea/style-coaching#
AI style coaching and education (libs/aglaea/style-coaching/src): a
coaching-plan builder over a lesson-database.
@aglaea/style-communities#
Social style communities (libs/aglaea/style-communities/src): community/
membership management, challenges, content moderation and trending detection.
@aglaea/outfit-sharing#
Outfit sharing (libs/aglaea/outfit-sharing/src): post creation,
platform-specific formatting, share-analytics tracking and an outfit-tagging
system.
@aglaea/inspiration-feed#
The style inspiration feed (libs/aglaea/inspiration-feed/src, ~4.6K LOC):
feed-algorithms, content-sources aggregation (with a SourceRateLimiter), a
personalization-engine, content-scoring and a diversity-engine.
@aglaea/expert-network#
The human-stylist marketplace (libs/aglaea/expert-network/src): expert
profiling, matching, consultation scheduling and a review system.
@aglaea/family-features#
Family/group styling (libs/aglaea/family-features/src, ~3.3K LOC): a
family-engine, growth-tracking (height percentiles, predictHeight6Months,
heightToChildSize, next-size-up date prediction), hand-me-down planning,
child-safety and family-coordination.
@aglaea/event-prep#
Event-specific styling preparation (libs/aglaea/event-prep/src): the
event-prep module that builds preparation plans for a specific event.
@aglaea/calendar-integration#
Calendar-aware styling (libs/aglaea/calendar-integration/src): event
extraction, dress-code inference, outfit-reminder scheduling and packing-list
generation from calendar/travel events.
@aglaea/gamification#
Style gamification (libs/aglaea/gamification/src): achievements, badges,
challenges, streaks, a point system and leaderboards.
@aglaea/smart-mirror#
Smart-mirror integration (libs/aglaea/smart-mirror/src, ~4.4K LOC): a
mirror-protocol with session/capability negotiation, AR-overlay generation,
hardware-profiles, a lighting-engine, an interaction-engine and a
widget-system.
@aglaea/smart-device#
IoT/wearable closet ecosystem (libs/aglaea/smart-device/src): rfid-tracking
of garments, environment-monitor (textile science), body-composition,
wearable-integration and mirror-lighting.
@aglaea/biometric-integration#
Health-data personalization (libs/aglaea/biometric-integration/src): a
biometric-collector, appearance-impact assessment and trend-analysis over
collected metrics.
@aglaea/weather-service#
Weather-aware styling (libs/aglaea/weather-service/src, ~3.7K LOC): thermal
comfort modeling (PMV/PPD), wind chill, heat index and WBGT in
thermal-calculations, plus a layering-system, precipitation-gear,
uv-protection, air-quality and an outfit-planner.
@aglaea/sophia-integration#
Cross-domain knowledge integration (libs/aglaea/sophia-integration/src, ~5K
LOC): query-routing of style Q&A over supplied knowledge entries, a
citation-engine, knowledge-domains mapping and research-enhancement of
recommendations. This is the seam designed to front a Sophia knowledge base —
self-contained, not an actually-wired cross-domain edge.
@aglaea/ethical-ai#
Bias detection and fairness (libs/aglaea/ethical-ai/src, ~3.5K LOC): a
fairness-metrics module computing demographic parity, equalized odds, equal
opportunity, predictive parity, calibration and treatment equality
(runComprehensiveFairnessAudit), plus a bias-engine, intersectional-bias,
inclusive-language, database-gaps, remediation and model-card
generation.