Create README.md

fix: 16-token limit bug + smart web search routing

feat: always-on web search — every query gets real-time data

fix: web search now runs without workspace connection

feat: parallel multi-engine search with result merging

feat: 6 search engines with automatic fallback chain

feat: add Bing RSS + Google Custom Search (free, no API key needed)

feat: add web search for real-time data (Jarvis mode)

feat: integrate LangChain4j + Google ADK framework

feat: multi-agent system with local Ollama LLM integration