vocall_sdk 1.0.1

Flutter SDK for the Vocall Agent-Application Protocol (AAP). Let an AI agent see and control your app's UI — navigate screens, fill forms, click buttons, and interact via voice or text.

Changelog

All notable changes to the Jarvis SDK for Flutter will be documented in this file.

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

1.0.1 - 2026-03-21

Fixed

• JarvisOverlay dispose race condition — when navigating between pages that both use JarvisOverlay, the old overlay's dispose could erase the new overlay's callbacks. Now checks ownership before clearing.

1.0.0 - 2026-03-21

Added

• TTS audio queue — streaming audio playback with queue management
• audio_clear — prevents TTS overlap after auto-narration
• LLM observability — real-time LLM token streaming and response events
• Chat UI polish — auto-focus, adjustable height, clear button, font fixes
• Material Icons — copy chat and UI refinements
• Voice command execution — voice commands now trigger UI actions (navigate, fill, etc.)
• Conversation reset — status resets on conversation_end
• Narration text fix — replaces narration text with final LLM response on llm_end

0.1.0 - 2026-03-19

Added

• JarvisClient — WebSocket client with automatic reconnection (max 10 attempts, exponential backoff)
• ManifestBuilder — fluent API for constructing app manifests
• FieldRegistry — register form fields and actions for agent-driven UI automation
• JarvisControllable — mixin for StatefulWidget pages with automatic field registration and cleanup
• JarvisOverlay — floating action button with expandable chat panel widget
• ChatPanel — message thread UI with streaming tokens, pagination, and copy-to-clipboard
• VoiceVisualizer — animated robot head with spectrum analyzer and status visualization
• StatusIndicator — status badge widget
• showJarvisConfirmDialog — confirmation dialog for ask_confirm commands
• VoiceService — Web Audio API implementation for microphone capture and TTS playback
• FrameSplitter — PCM frame normalization for consistent 20ms audio frames
• Always-listening mode — continuous voice with wake word detection
• Click-to-talk mode — on-demand voice input with direct mode
• Barge-in (interrupt) — stop agent mid-response
• Echo suppression — automatic mic muting during TTS playback
• Typewriter animation — character-by-character field filling
• Command execution engine — sequential and parallel action execution with retry logic
• Session continuity — persistent visitor ID with localStorage on web
• Chat history — paginated message loading with 6-message visible window
• Protocol types — full Dart mirror of the AAP protocol (ManifestMessage, ScreenDescriptor, FieldDescriptor, ActionDescriptor, ModalDescriptor, UIAction, CommandMessage, etc.)

Platform Support

• Flutter Web — full support (chat + voice)
• Android / iOS / Desktop — chat and UI automation only (voice planned)