Welcome to the central repository for the reconstruction of willbracken.com (formerly brackenw3.github.io) and the environment configuration for our local AI agent infrastructure.

This project is a complete reconstruction of the jaw-dropping, highly interactive portfolio site showcasing my work across the VSCode_Folders monorepo and other repositories.

Core Aesthetic:

• Theme: Dark / Cyberpunk
• Palette: Deep, rich blue tones as the primary backdrop, accented with subtle, striking orange highlights.
• Experience: Studio-grade graphics, animations, and cutting-edge 3D visualizers.

Key Features:

• The Code Galaxy: An interactive 3D WebGL (Three.js) experience allowing users to visualize code updates by volume, language type, and complexity.
• Interactive Project Previews: Graphical, interactive representations of deployed applications, highlighting the skills and scale of work involved.
• Full-Stack & Dynamic: Leveraging Cloudflare Pages, Workers, D2, Railway.app, Supabase, and GitHub Actions to keep the visualizer data fresh and cost-optimized.
• n8n Automations: Integration with custom n8n deployments for backend data processing and workflow coordination.

We have engineered custom local LLM setups optimized for an i9 desktop with an RTX 4080 (16GB VRAM) and 64GB of system RAM. These models empower the terminal-based CLI agent with exceptional coding, automation, and multi-modal capabilities.

Hyper-optimized for complex agentic workflows, strict tool-calling, n8n automations, React/Three.js generation, and secure Docker orchestration.

• Hardware Profile: Offloads heavily to the RTX 4080 GPU while utilizing the 64GB system RAM for a massive 32k context window.
• Enhancements: Includes custom chat templates, stop tokens (e.g., Observation:), and strict temperature settings (0.1) for deterministic, precision tool-calling in the CLI.

To build and run:

ollama create expert-coder -f Modelfile.expert-coder
ollama run expert-coder

Fits entirely within the 16GB VRAM of the RTX 4080. Acts as the "eyes" of the environment, parsing static images, UI mockups, and diagrams into structural code and feedback.

To build and run:

ollama create vision-agent -f Modelfile.vision-agent
ollama run vision-agent

1. Frontend Development: Navigate to apps/portfolio to begin development on the Vite + React TypeScript application.

   cd apps/portfolio
   npm install
   npm run dev

2. Security & Best Practices:

   • DO NOT commit API keys, .env files, or sensitive credentials. The .gitignore is heavily fortified.
   • We utilize Cloudflare Workers, GitHub Models, and local Ollama to balance cloud capabilities with local privacy/cost.

Let's get to work reconstructing the masterpiece!