Fix agent API issues on Hugging Face Spaces

Add defensive None check for case_context in BaseAgent._respond_with_claude

Fix multi-agent orchestration: shared ward context, dynamic vitals, persona guardrails

Use adaptive thinking (Opus 4.6 recommended) instead of deprecated enabled+budget_tokens

Use claude-opus-4-6 for all agent and knowledge builder API calls

Fix agent pipeline: thinking config, cache keys, and conversation compression

Fix 500 error on /api/agents/action and NaN in vitals sparklines

Fix agent routing bug: Dr. Sharma responding instead of Patient/Family

Fix HF Spaces runtime errors: missing SimulationOrchestrator, httpx proxies, ChromaDB telemetry

Fix HF Spaces deployment: Dockerfile, router registration, cleanup backup files

Add Hugging Face Spaces deployment configuration

🚀 Production-ready multi-agent medical simulation with major improvements

Merge enhanced clinical simulation with multi-agent system, WhatsApp UI, and examination module

Complete 5-phase clinical simulation: agents, complications, WhatsApp UI, exam module

Add full hospital simulation engine with time progression, treatment effects, and multi-agent dynamics

Enforce strict source grounding, accuracy constraints, and Indian hospital realism

Add DynamicKnowledgeBuilder for per-case agent specialization via RAG + Claude

Add multi-agent hospital ecosystem with patient, nurse, and senior doctor

feat: Fix case routing, add session tracker for dynamic analytics

feat: Complete Clinical-Mind v2.0 - Full-Stack AI Patient Simulator

fix: Switch model from Sonnet to Claude Opus 4.6 across all API calls

feat: Wire frontend to backend APIs and upgrade Socratic tutor to Claude

feat: Add dotenv loading for ANTHROPIC_API_KEY in backend startup

feat: Expand all specialties to 10+ cases, update API for 14 specialties

feat: Build complete RAG system with real Indian medical corpus + ChromaDB + Claude API

feat: Initial Clinical-Mind project setup with full-stack architecture