Nested Private Services: a Microservice node inside a Project Workspace now recursively simulates the referenced service's full internal architecture against the exact RPS it receives — no more static benchmark snapshots or infinite-capacity black boxes.

Project Simulation Consistency: a nested service's error rate now drives both its own node metrics and the request path, so per-node and overall error rates always agree (previously a healthy node could coexist with a 100% overall error rate).

Request Absorption Rate: configurable per Microservice node to model how much inbound traffic cascades to internally-referenced services (caching, early returns, validation drops).

Cross-Service Traffic: a service referenced from multiple parents now correctly accumulates load from every path (direct + each cascade), not just its direct share.

Simulation Accuracy: Implemented probabilistic error propagation and Bootstrap P99 confidence intervals for refined performance modeling.

Virtual Time & LRU Caching: Introduced discrete virtual time tracking and LRU cache modeling to simulate complex workload behaviors.

Markdown Reports: Added functionality to generate, copy, and export detailed architectural simulation reports in Markdown format.

Node Physics: Implemented internal hit ratios for leaf nodes and cascade depth tracking to model local request absorption and call stack overhead.

Modular Engine: Extracted latency and utilization logic into dedicated modules and centralized component definitions for better maintainability.

Export Branding: Added metadata and footer branding to simulation exports for enhanced report professionalism.

Infrastructure Expansion: 10 new components including GraphQL, RabbitMQ, and Distributed Databases with deep-dive documentation formulas.

Architectural Search: Real-time search functionality added to the Dashboard to find and filter architectural designs instantly.

Simulation Engine: Modularized logic into TrafficWeight/MetricsAggregator and implemented a 3-iteration saturation loop for better network convergence.

Request Propagation: Failures in parallel branches are now isolated, preventing error-leakage into independent sibling request paths.

Engine Observability: Refactored core visit logic to return pure VisitResult objects { latencyMs, isError, depth }, removing side-effect mutations.

UX & Accessibility: Resolved layout visibility issues in the Dashboard grid and enforced 100dvh viewport stability for mobile sidebars.

Messaging Engine: Decoupled producer-consumer latency for Kafka, SQS, and Pub/Sub. Producers now properly "fire and forget", ignoring downstream consumer latency in the primary request path.

Mail Server physics: Implemented SMTP-aware bottleneck logic including concurrent connection ceilings and daily send quotas.

Kafka Scalability: Throughput capacity now scales linearly with partition count (2k msg/s per partition), and latency accounts for broker replication overhead.

SQS Soft-Limits: Differentiated FIFO (300 msg/s) vs Standard (1k msg/s) AWS throughput caps and ordering overhead penalties.

Enhanced VectorDB & AI models: Implemented dimension-aware search latency and refined LLM autoregressive token generation speeds.

Service Worker component: background worker pool with M/D/1 (Pollaczek–Khinchine) queueing model. Configurable worker count, job duration, queue depth, memory per worker, and retry policy.

Simulation engine: per-node average latency (latencyMs) now tracked and reported in simulation results.

Workload type modifiers: read/write/cpu/io/memory profiles now actively modify storage IOPS, cache capacity, and compute saturation thresholds.

Client node network cost: the client/origin node no longer incorrectly pays a 15–25ms internet hop on every request.

Multi-client RPS: total RPS now correctly sums all client nodes; traffic is distributed proportionally per client RPS share.

gRPC latency benefit now correctly applied to network transmission cost (−30% networkLat), not the 2ms processing base.

Shard count added for SQL, NoSQL, Redis, and Memcached: multiplies capacity linearly via horizontal sharding.

Updated cached logic: fixed cache-miss routing for Load Balancers and API Gateways to prevent unintended parallel backend requests.

Memcached component: multi-threaded volatile cache with memory, threads, hit ratio, and sharding config.

Conditional field visibility (showIf): Shard Count now only appears in the config panel when Enable Sharding is toggled on.

Right config sidebar scrolling: sidebar now scrolls within itself, fixing layout overflow issues.

Sharding enabled in Redis: simulated horizontal scaling for high-throughput caching.

Memcache component: added to the architecture library with specific TTL and LRU logic.

Infinite Drag-and-Drop Canvas with high-performance infrastructure rendering.

Simulation Engine: High-fidelity request propagation with CPU/RAM cost modeling.

Network Physics: Region-aware latency, VPC boundaries, and cross-region penalties.

Advanced Load Balancing: Round-Robin, Weighted, and Resource-Based algorithms.

Bottleneck Detection: Automatic identification of saturation points and "Hot Nodes".

Support for 24+ Components: Compute, SQL/NoSQL, Kafka, SQS, AI Models, and more.

Live Metrics Dashboard: Real-time P50/P99 latency, RPS, and Error rate visualization.

VPC Grouping: Logical isolation of nodes with automated CIDR and Region handling.

Sticky Notes: Markdown-ready annotations pinned to nodes, edges, and VPCs.

State Persistence: LocalIndexedDB integration for persistent project drafts.

Collaboration: One-click serialization to URL hash for instant sharing.

Technical Documentation: Integrated "Engine Docs" explaining component physics.

PWA & Mobile Support: Optimized touch-anchors and responsive tool navigation.