flutter_local_agent_kit 1.0.1

Offline-first AI framework for Flutter. Resident LLM inference, local RAG, and autonomous agents with a Material 3 Chat UI.

Changelog

1.0.1

• Lifecycle: Reinitializing FlutterLocalAgentKit now disposes existing LLM and RAG sessions before recreating them.
• Reliability: Llama3Template now serializes a single BOS token per prompt for correct multi-turn formatting.
• Capability State: Added isRagReady and ragInitializationError so apps can detect LLM-only fallback when optional RAG startup fails.
• Testing: Added regression coverage for prompt formatting, reinitialization cleanup, degraded RAG startup, and chat-stream error handling.
• UI: AgentChatView now replaces failed response placeholders with a visible error message instead of leaving an empty assistant bubble.
• Dependency Cleanup: Removed unused google_generative_ai and migrated from discontinued flutter_markdown to flutter_markdown_plus.
• Pub.dev Optimization: Improved metadata, shortened description, and updated dependency constraints.
• Documentation: Added missing library and constructor documentation to reach 100% coverage.
• Visuals: Replaced placeholder screenshots with high-fidelity, premium Material 3 UI assets.
• Bug Fixes: Updated markdown rendering dependency for continued package support.

1.0.0

• Initial Release: Production-ready core for on-device AI.
• Orchestration: Implemented FlutterLocalAgentKit unified facade for LLM, RAG, and Agents.
• LLM Core: High-performance GGUF inference via llamadart.
• Private RAG: Offline-first vector search via mobile_rag_engine.
• Autonomous Agents: ReAct-based reasoning loop for tool-augmented intelligence.
• Premium UI: Included AgentChatView with high-speed streaming and Markdown support.
• Management: Robust ModelManager for background model orchestration and integrity.
• Compliance: 100/100 pub.dev health score ready.