How fatihmakes/mark-lv works
Mark LV is a cross-platform voice-first personal AI assistant that can perceive the screen, speak, control the computer, and expose a phone dashboard. The principal workflow runs from user input through the live assistant in main.py, which coordinates audio, vision, memory, and dynamically loaded skills; results appear in the HUD or are spoken back. The graph preserves the documented capability branches. Several action modules and runtime integrations were not excerpted, so their internal links are omitted where the evidence does not establish them.
Main entry points: main.py Live assistant · action_loader.py Skill discovery · actions Computer and desktop control · ui.py HUD and controls
Components 27
HUD and audio
- HUD and controls
ui.py - Holographic avatar
core/avatar.py - Face geometry
core/avatar_mesh.py - Audio devices
core/audio_devices.py - Wake word
core/wake_word.py - Push to talk
core/hotkey.py - Echo guard
core/echo.py - Lip sync
core/viseme.py - Remote dashboard
dashboard/server.py
Assistant runtime
- Live assistant
main.py - Gemini live and ladder
core/gemini.py - One-shot model client
core/llm_client.py - Speech recognition
core/stt.py - Speech synthesis
core/tts.py
Skills and tasks
- Skill discovery
core/action_loader.py - Web search
actions/web_search.py - Computer and desktop control
actions - File operations
actions - Video playback
actions - Proactive tasks
actions - Plugin discovery
core/plugin_loader.py - System monitoring
actions/system_monitor.py
Memory and access
- Persistent memory
memory/memory_manager.py - Assistant configuration
memory/config_manager.py - Undo history
core/undo.py - Action confirmation
core/confirm.py
Other components
- User
Connections 29
- User → Live assistant: speaks or requests
- Live assistant → HUD and controls: updates display
main.py - Live assistant → Gemini live and ladder: runs live session
main.py - Live assistant → Audio devices: configures devices
main.py - Live assistant → Persistent memory: loads context
main.py - Live assistant → Assistant configuration: reads settings
main.py - Live assistant → Skill discovery: loads skills
main.py - Live assistant → Plugin discovery: loads plugins
main.py - HUD and controls → Holographic avatar: shows face
ui.py - Holographic avatar → Face geometry: uses geometry
core/avatar.py - Live assistant → Wake word: gates listening
main.py - Live assistant → Push to talk: handles hotkey
main.py - Live assistant → Echo guard: filters self-echo
main.py - Live assistant → Lip sync: drives mouth
main.py - Live assistant → Speech recognition: transcribes audio
- Live assistant → Speech synthesis: speaks replies
- Skill discovery → Web search: discovers skill
- Skill discovery → Computer and desktop control: discovers skills
- Skill discovery → File operations: discovers skill
- Skill discovery → Video playback: discovers skill
- Live assistant → Web search: runs search
main.py - Live assistant → System monitoring: runs monitoring
main.py - Live assistant → Proactive tasks: runs proactive tasks
main.py - Computer and desktop control → Action confirmation: requests confirmation
actions/computer_settings.py - Computer and desktop control → Undo history: records reversible changes
actions/computer_settings.py - File operations → Undo history: records file changes
actions/file_controller.py - Plugin discovery → Assistant configuration: reads configuration
core/plugin_loader.py - User → Remote dashboard: pairs phone
- Remote dashboard → Live assistant: submits commands
main.py