L.U.M.A. V2 // Lau’s Universal Management Agent

The Central Intelligence Hub for Personal Engineering Operations

---

L.U.M.A. is a localized, high-performance AI OS designed for engineering project management and daily task orchestration. Taking inspiration from the F.R.I.D.A.Y. standard, she features a grounded Irish-lilted personality, technical fluidity, and a real-time HUD visual cortex.

🛠️ System Architecture

L.U.M.A. operates on a modular "Nerve-Link" architecture, anchoring its logic in a dual-directory structure to maintain environmental isolation:

• The Nerve Center (LUMA001/): The root directory housing the primary .venv and shared binary dependencies (FFmpeg DLLs).
• The Cognitive Core (luma-orb/): The engine's heart, containing the LLM interface, voice engine, and localized knowledge archives.

📂 Module Breakdown

1. main.py: The Mission Control. Orchestrates the sensory loop and event handling.
2. luma.py: The Router. Manages cognitive state, mode switching (e.g., DEEPWORK), and LLM uplinks.
3. voice_engine.py: The Sensory Array. Handles asynchronous speech queuing and neural XTTS-v2 synthesis.
4. energy_orb.py: The Visual Cortex. Renders the real-time HUD, CPU pulse, and Location station telemetry.
5. luma_ops.py: The Archive Manager. Handles YAML-native data migration and environment grounding.
6. config.py: The Central Core. Defines the Nordic Tech Palette and operational constraints.

---

🧠 Cognitive Grounding (YAML & Markdown)

L.U.M.A. has transitioned from rigid JSON to a fluid YAML/Markdown hybrid knowledge base located in luma-orb/knowledge/:

• projects.yaml: Active engineering logs for the station.
• session.yaml: Contextual memory of recent interactions.
• persona.md: Behavioral constraints (The Irish-lilted F.R.I.D.A.Y. vibe).
• user.md: Direct grounding for Master Lau’s role and preferences.

---

🖥️ Operational Telemetry (The HUD)

The Energy Orb interface utilizes a real-time top-bar ribbon for station diagnostics:

• PULSE: System uptime and CPU load monitoring.
• HERNING_STN: Local weather grounding (via Open-Meteo) and synchronized timestamping.
• THOUGHT STREAM: Real-time ghost-text display of the LLM’s token generation.

---

⚡ Technical Specifications

• Core Model: Phi-3 (Quantized 4-bit) via Ollama.
• STT: Faster-Whisper (Tiny).
• TTS: Neural XTTS-v2 with asynchronous sentence queuing.
• Environment: Python 3.11 with pyyaml, psutil, pygame, and requests.

---

⚖️ License & Constraints

Designed exclusively for personal use. Unauthorized redistribution is prohibited.