Vitruvyan Core — Changelog

Last updated: February 17, 2026

Cronologia consolidata di tutte le milestone architetturali di Vitruvyan Core, dalla fondazione al V1.0.

---

Feb 17, 2026 — Android Oculus Prime App Roadmap

Planning document created for Android edge client development (docs/planning/ANDROID_OCULUS_PRIME_APP_ROADMAP_FEB17_2026.md):

• Analyzed existing Edge Gateway + Oculus Prime API architecture for Android integration.
• Proposed 3-module Android SDK: sdk-core (networking + contracts), sdk-offline (SQLite outbox + sync), sdk-sensors (camera, audio, GPS helpers).
• Designed offline-first architecture with local queuing, background sync via WorkManager, and retry policies.
• Specified security layer: device identity, token encryption (EncryptedSharedPreferences), HMAC payload signing, optional mTLS.
• Defined 5-sprint implementation roadmap (SDK foundation → demo app → IoT sensors → security → E2E validation).
• Reference implementation target: Jetpack Compose app with camera capture, gallery upload, GPS tracking, and real-time status dashboard.

Feb 17, 2026 — LangGraph 1.0.8 Production Upgrade

Graph Orchestrator upgraded from LangGraph 0.5.4 to 1.0.8 (major version jump). Isolated test container validated compatibility. Dependencies upgraded: langgraph-checkpoint 2.1.2 → 4.0.0, langgraph-prebuilt 0.5.2 → 1.0.7, langgraph-sdk 0.1.74 → 0.3.6. All functional tests passed. Production deployment successful (port 9004, service healthy, dispatch endpoint validated).

Feb 17, 2026 — Edge Intake Compliance Refactor (Service Layer)

services/api_edge_oculus_prime aligned to LIVELLO 2 service logic for service-side I/O:

• Added adapters/persistence.py as dedicated DB read adapter for health/evidence/pipeline/events endpoints.
• Removed downstream coupling from Intake API read models (cognitive_entities queries removed).
• Kept Streams-native emission path unchanged (intake.evidence.created) with append-only persistence.
• Updated intake docs to reflect intake_event_failures naming and Streams consumer-group semantics.
• Added tests for persistence adapter, routes, and document ingest flow (insert + emit behavior).

Feb 17, 2026 — Oculus Prime Event Naming Migration (Versioned)

Versioned migration introduced for event naming consistency in edge ingestion:

• Added canonical channel oculus_prime.evidence.created with schema vitruvyan://oculus_prime/events/evidence_created/v2.0.
• Preserved legacy alias intake.evidence.created for backward compatibility.
• Implemented rollout modes via OCULUS_PRIME_EVENT_MIGRATION_MODE: dual_write (default), v2_only, v1_only.
• Updated emitter audit metadata to include migration mode and emitted channels.
• Added v2 schema file and migration tests for dual-write/rollback behavior.

Jan 25–26, 2026 — Herald → Redis Streams Migration: 100% Complete

All 7 listeners migrated from pub/sub Herald to Redis Streams. ReadOnlyError fix (2-line fix in listener_adapter.py). Shadow Traders + MCP Listener migrated (Phase 3 of 3).

Jan 24, 2026 — Cognitive Bus Phase 6: Plasticity System

Bounded, auditable, reversible parameter adaptation. Consumers learn from outcomes with 5 structural guarantees (bounded, auditable, reversible, opt-in, governance-validated). 6/6 tests passing.

Jan 24, 2026 — Cognitive Bus Phase 0: Bus Hardening (BREAKING)

Redis Streams made canonical transport. Pub/Sub archived. Fixed BaseConsumer (broken async), unified 4 incompatible event models into TransportEvent / CognitiveEvent / EventAdapter.

Jan 20, 2026 — Cognitive Bus Phase 4: Working Memory System

Distributed working memory for consumers. Octopus-mycelium architecture: isolated local memory + optional mycelial sharing via events. Memory Inspector API for debugging.

Jan 19, 2026 — Cognitive Bus Phase 3: Socratic Pattern (non_liquet)

Orthodoxy Wardens gained epistemic gatekeeping: 5-state verdict system (blessed, purified, heretical, non_liquet, clarification_needed). Vitruvyan can now explicitly say "I don't know" instead of hallucinating.

Dec 30, 2025 — v1.0.0 Milestone: Domain-Agnostic Framework Complete

Vitruvyan Core declared production-ready. Phases 1–3D complete. Mercator (finance) validated as PoC. Vitruvyan (governance) ready to proceed.

Dec 30, 2025 — Phase 3D: Neural Engine Integration Pattern

Canonical integration pipeline: NE → VWRE → VARE → VEE. VerticalOrchestrator, BatchProcessor, ResultAggregator utilities. Mercator-lite demo vertical as proof-of-concept.

Dec 30, 2025 — Phase 3: Domain Abstraction (VEE/VARE/VWRE)

VEE, VARE, VWRE engines refactored from finance-specific to domain-agnostic. 3 abstract contracts (Explainability, Risk, Aggregation) + 3 finance provider implementations for backward compatibility. COO approval granted.

Dec 29, 2025 — Phase 1E: Neural Engine Abstraction

Domain-agnostic Neural Engine: AbstractFactor, NormalizerStrategy, AggregationProfile contracts + EvaluationOrchestrator + 3 built-in normalizers (ZScore, MinMax, Rank).

Dec 29, 2025 — Phase 1D: Node Abstraction (Finance Logic Removal)

Finance-specific logic removed from 5 LangGraph nodes. Boot test validated all containers (zero import errors, APIs responsive).

Dec 28, 2025 — Phase 1A–1C: Foundation & Domain Contracts

Repository created (vitruvyan_os → vitruvyan_core), package renamed, domain contracts defined (EntitySchema, SignalSchema, DomainPolicy), GenericDomain fallback implemented.

---

Note: This file consolidates 21 per-phase reports that existed in this directory prior to Feb 16, 2026. The original detailed reports are preserved in git history.