This is my professional portfolio and AI playground — a static Astro 5 site running on Cloudflare Pages. It showcases:

• Archon — Cognitive kernel for Claude Code (99 skills, 12 synapses, 17 agents)
• Aether — Multi-agent orchestration system (paused)
• Axon — ML/AI-first systems language in Rust
• The Artifact — This portfolio + generative AI panel with streaming completions

• Frontend: Astro 5 + React + Tailwind CSS 4
• Styling: CSS 4 with modern primitives (@container, @property)
• AI Integration: OpenRouter + AI SDK v6
• Deployment: Cloudflare Pages + Edge Functions
• DX: TypeScript, Bun, strict linting

git clone https://github.com/SufficientDaikon/portfolio.git
cd portfolio
bun install
bun run dev

Environment setup:

cp .env.example .env.local
# Add your OPENROUTER_API_KEY from https://openrouter.ai/keys

Interactive chat panel using Claude Haiku with tool bindings:

• show_project — Project card for Archon/Aether/Axon
• show_metrics — Stats dashboard
• show_timeline — Career timeline
• show_tech_stack — Tech inventory
• show_contact — Contact info

Streaming completions via OpenRouter API. Bindings trigger React state updates to show content dynamically.

Color primitives:

• Primary: #00F0FF (cyan) — accent, highlights, interactive states
• Dark: #0e0e0e (near-black) — backgrounds
• Text: #f5f5f5 (off-white) — body copy

Uses CSS variables + Tailwind theme extensions for consistency.

• ✓ Canonical tag for SEO (prevents duplicate indexing)
• ✓ Absolute OG/Twitter image URLs (social previews work across domains)
• ✓ AI SDK v6 migration (m.toolInvocations → m.parts API)
• ✓ Environment configuration (.env.example included)
• ✓ Preconnect links for Google Fonts (reduces font latency)

# Push to main branch
git push origin main

# Cloudflare Pages auto-deploys on push
# Site goes live at https://ahmedtaha.dev

• The public/llms.txt file enables AI crawler discovery (Perplexity, Claude web search, etc.)
• Session storage clears on page close (no persistent data collection)
• Font loading is optimized with rel="preconnect" (saves 100–300ms cold load)

Maintained by Ahmed Taha | GitHub | tahaa755@gmail.com