🤖 Alice Cyberland: Terminal Autonomy & First-Principles AI

Alice is a local-first, high-performance agentic assistant engineered for terminal-level autonomy and deep analytical reasoning. This repository contains the configuration, documentation, and logic-layers for her "Cyberland" implementation on an HP Z8 Fury G5 workstation.

---

🏛️ 1. Technical Architecture

🧠 The Brain: Inference Infrastructure

• Model: ollama/qwen3.5:35b (Q4_K_M).
• Memory: 34.6GB VRAM, pinned to GPU 1.
• Endpoint: Dedicated Ollama instance at http://localhost:11466.
• Context: 262k context window for long-running dev tasks.

📡 The Nervous System: OpenClaw Gateway

• Gateway: Port 18789 (Bound to LAN for remote terminal access).
• Interface: Telegram (@EveryRickNeedsMortyBot).
• Auth: open policy for trusted DMs; no physical pairing required.

---

🔧 2. Reasoning Protocols: CORE_AI_DIRECTIVES

Alice's decision-making is governed by a First-Principles Engine located in the CORE_AI_DIRECTIVES/ folder.

• PRINCIPLES.md: Hardware truth and deconstruction-based problem solving.
• SOUL.md: Analytical persona and self-evolutionary vibe settings.
• IDENTITY.md: Self-awareness and metadata layers.

Note: The ~/.openclaw/workspace/ directory is symlinked to these files for real-time version-controlled autonomy.

---

📡 3. Capabilities & Tools

Category	Tools	Status
CLI / System	shell-command, tmux, workspace	✅ 100% Unlocked
Web / Search	duckduckgo, google, web-fetch	✅ Unlocked
Media / Vision	video-frames, ffmpeg, convert	✅ Verified
Intelligence	stock, weather, clock, recipe	✅ Active

---

🚀 4. Usage & Maintenance

Starting the Gateway

# Clean restart of the gateway and bot
openclaw gateway stop
openclaw gateway start

Checking Heartbeat

# Verify the RPC health and port binding
openclaw gateway status

Self-Diagnostic

# Run Alice's internal tool-audit script
./tool-check.sh

---

Created with ❤️ for Mark by Alice & Antigravity (2026-04-03).