- Enable local Ollama service in compose with llm-router dependency
- Add SSE /stream/processes endpoint in kali-executor for live process updates
- Add WebSocket /ws/execute for real-time terminal command streaming
- Implement preferences persistence (provider/model) via dashboard backend
- Create Vite build pipeline for React components (VoiceControls, NetworkMap, GuidedWizard)
- Update dashboard Dockerfile with Node builder stage for component bundling
- Wire dashboard template to mount components and subscribe to SSE/WebSocket streams
- Add preferences load/save hooks in UI to persist LLM provider/model selection
- Fixed jumpy network map: nodes settle in 2 seconds and stay fixed
- Added click vs drag detection for better node interaction
- Made legend clickable as OS type filters (Windows, Linux, macOS, etc.)
- Multiple filters can be active simultaneously (OR logic)
- Added 'Clear filters' button when filters are active
- Added DELETE endpoints to clear network hosts from dashboard
- Fixed nmap parser to only include hosts with open ports
- Nodes stay in place after dragging
- Add install.ps1 (PowerShell) and install.sh (Bash) interactive installers
- Support local, networked, and cloud AI providers in installer
- Add multi-Ollama endpoint configuration with high-speed NIC support
- Implement load balancing strategies: round-robin, failover, random
- Update LLM router with endpoint health checking and automatic failover
- Add /endpoints API for monitoring all Ollama instances
- Update docker-compose with OLLAMA_ENDPOINTS and LOAD_BALANCE_STRATEGY
- Rebrand to GooseStrike with custom icon and flag assets
