feat: stabilize AI engine, harden Zod schemas, and implement UX guardrails

feat: Implement fuzzy matching for commands and input guardrails in whatsapp.ts

fix: Update Gemini model names and harden transcription body reader

fix: Decouple A/V providers in AIService and refactor routes for robustness

feat: Implement Multi-AI Architecture with Google Gemini provider and failover logic

fix(whatsapp): Sprint 42 - Sécurisation routage Deep Dive et immuabilité secteur

fix(ai): Sprint 41 - Détanchéité linguistique Wolof stricte et réinjection boucle Deep Dive

feat(Sprint 40): Boucle interactive Deep Dive & persistences des réponses

feat(ai): refactor coach feedback to 3-pillar format with mandatory browsing (Sprint 39)

chore: execute Sprint 38 technical debt resolution (Type Safety, Zod validation, Vitest, Mock LLM extracted)

feat: Genspark-Standard upgrade, MLOps audit fixes, and XAMLÉ branding

feat(ai): integrate Whisper STT confidence score for auto-validation

fix(core): remove AI jus-de-fruits hallucination, expand onboarding sectors to 8, streamline pedagogy menu and lock REPRISE badge

fix(ai): implement pedagogical leap fixes, pending_remediation, coherence locks and cache wipes

feat(ai): integrate XAMLÉ v2 prompt guidelines including anti-hallucination constraints and strict Wolof normalization

fix(ai): remove hardcoded sector examples from system prompts to prevent AI hallucinations across mismatched domains

fix(ai): add imageUrl field to Prisma schema and correctly route language selection inside whatsapp.ts

fix(ai): fully wipe BusinessProfile and Response history on INSCRIPTION to prevent AI context leaks

feat(ai): expert coaching system with persistent business context and explicit SUITE progression fix

feat(sprint19-20): production-ready video integration & STT hardening

chore: final integration audit and logic hardening

fix(syntax): resolve unbalanced braces and scoping in WhatsAppService

fix(logic): prioritize commands over exercises and strictly lock gateway routing

feat(hardening): implement anti-silence, gateway loop protection and deep trace logging

Fix Post-STT Blockage: Async chain, Intent detection, and Feedback fallback

Audit Final: Fix feedback message construction & Global Hardening

feat(wolof): standardized tracks T1-T5 with glossary v4.0 and implemented adaptive multi-sector architecture

chore: final logic hardening, AI prompt reinforcement and SUITE command fix

feat(ai): fully operational v1.0 criteria engine and 5-level curriculum (10 modules)

feat(pedagogy): implementation of dynamic content infrastructure and evaluation criteria

fix: final hardening audit fixes (normalizeCommand, activityLabel key, Zod schemas)

feat: localize AI documents (FR/WO) and fix Zod warnings

fix: resolve ESM module not found error in database package

feat: implement adaptive pedagogy, visuals, and remediation (WOW Phase 1)

feat: implement WOW Sprint 1 (BusinessProfile, Structured Feedback, STT Hardening, Progress Bar)

feat: audit and hardening of audio pipeline, AI localization, and WhatsApp UX improvements

feat: complete audio pipeline hardening and audit requirements

build: force deep rebuild v4 with improved logging

debug: add detailed logging for transcription connection error

feat(audio): Enforce strict AI_API_BASE_URL and add FFmpeg MP3 STT conversion

feat(audio): Complete robust outbound/inbound pipeline and async AI generation

fix: Split Webhook into Gateway (HuggingFace) and Processor (Railway), fix missing https:// in API_URL

fix(whatsapp): resolve duplicate CONTINUE action match and remote 120s blocking delay on feedback

feat: implement STT inbound audio, TTS fallback, and Interactive Lists

fix(deploy): make Stripe keys optional at initialization

refactor(prod): unify currentDay exclusively in Enrollment model to fix scheduler sync

fix(audit): 6 bugs fixed — static imports, store-audio endpoint, CONTINUE fallthrough, SUITE regex, no more require() calls

feat(railway): combined API+Worker Dockerfile + fix list char limits + 10s AI timeout

feat(ux): interactive buttons for language choice + LIST menu for sector selection (8 sectors FR/Wolof)

fix(comp5): add 429 catch on generateFeedback, use proper imports for whatsappQueue/scheduleTrackDay + Wolof translations