synheart_behavior 0.1.4

A lightweight, privacy-preserving mobile SDK that collects digital behavioral signals from smartphones.

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

0.0.1 - 2025-12-26

Added

• Initial release of Synheart Behavioral SDK for Flutter
• Core SDK classes: SynheartBehavior, BehaviorConfig, BehaviorEvent, BehaviorSession, BehaviorStats
• Streaming API for real-time behavioral events
• Session tracking with summaries
• Manual stats polling
• Platform channel interfaces for iOS and Android
• Example Flutter app demonstrating SDK usage
• Comprehensive documentation and README

Features

• Input interaction signals (tap, scroll, swipe gestures)
• Attention & multitasking signals (app switching, idle gaps, session stability)
• Privacy-preserving design (no text, content, or PII collected)
• Lightweight implementation (<150 KB compiled)
• Low resource usage (<2% CPU, <500 KB memory)
• Optional notification and call tracking (requires permissions)

Platform Support

• iOS 12.0+
• Android API 21+ (Android 5.0+)
• Flutter 3.10.0+

0.1.0 - 2025-12-29

Added

• Motion state inference with ML model (LAYING, MOVING, SITTING, STANDING)
• Typing session tracking and comprehensive typing metrics
• Emotion metrics integration
• Motion feature extractor for device motion signals
• Enhanced behavior session with expanded metrics
• ONNX model support for motion state prediction
• Label mapping for motion state classification

Features

• Real-time motion state prediction using on-device ML inference
• Typing activity ratio, cadence, and burstiness metrics
• Deep focus block detection
• Enhanced behavioral metrics (focus hint, distraction score)
• Improved session summaries with motion state information

0.1.1 - 2025-12-30

Changed

• Removed documentation file from repository

0.1.2 - 2025-12-31

Fixed

• Fixed null casting error in Android build.gradle.kts by making signing config conditional
• Updated .gitignore to exclude Gradle cache files (.gradle/ directory)

Changed

• Replaced onnxruntime with flutter_onnxruntime in pubspec.yaml
• Build configuration now gracefully handles missing key.properties file for debug builds

0.1.3 - 2026-01-08

Added

• On-Demand Metrics Calculation: New calculateMetricsForTimeRange() method to calculate behavioral metrics for custom time ranges within a session
• Time range validation to ensure selected ranges are within session bounds
• Session data persistence improvements - data now persists until next session starts, enabling on-demand queries for ended sessions
• Motion state inference support for on-demand calculations

Fixed

• Fixed session data being cleared too early (now cleared when new session starts instead of when session ends)

Changed

• Enhanced session data management to support on-demand metric calculations for ended sessions
• Improved motion data handling with proper type conversions for nested maps

0.1.4 - 2026-01-23

Changed

• Flux Integration: All behavioral and typing metric calculations now exclusively use synheart-flux (Rust library)
• Removed native Kotlin/Swift calculation implementations for behavioral metrics
• Removed redundant comparison fields (behavioralMetricsFlux, typingSessionSummaryFlux) from Dart models
• Updated data flow: All metrics now come directly from Flux, ensuring HSI compliance and cross-platform consistency
• Simplified codebase by removing ~1000+ lines of native calculation code

Removed

• Native calculation functions: computeBehavioralMetrics(), computeTypingSessionSummary(), computeIdleRatio(), computeFragmentedIdleRatio(), computeScrollJitterRate(), computeBurstiness(), computeDeepFocusBlocks(), and related helper functions

Breaking Changes

• Flux is now required: The SDK requires synheart-flux libraries to be present. If Flux is unavailable, session ending will fail with an error instead of falling back to native calculations
• Removed behavioralMetricsFlux and typingSessionSummaryFlux fields from BehaviorSessionSummary model (use behavioralMetrics and typingSessionSummary instead, which now contain Flux data)

Technical Details

• All behavioral metrics (interaction intensity, distraction score, focus hint, deep focus blocks, etc.) are computed by Flux
• All typing session metrics (typing session count, average keystrokes, typing speed, etc.) are computed by Flux
• Performance information now only tracks Flux execution time
• Native code now only handles event collection, session management, and Flux integration