Hugging Face's logo

Raiff1982
/
Codette-Reasoning

PEFT
English
codette
multi-perspective-reasoning
ethical-ai
lora
qlora
llama-3.1
recursive-cognition
rc-xi
behavioral-locks
cognition-cocooner
Eval Results (legacy)
Model card Files
xet
 Community
1
Codette-Reasoning / docs /deployment /SELF_HOST_CODETTE.md
Jonathan Harrison
Full Codette codebase sync — transparency release
74f2af5
preview code
|
raw
history blame contribute delete
3.55 kB

Deploy Codette Widget to horizoncorelabs.studio

TL;DR

  1. Take index-selfhosted.html from this repo
  2. Upload to horizoncorelabs.studio at /horizoncoreai/index.html
  3. Access at https://horizoncorelabs.studio/horizoncoreai/

What Changed

The self-hosted version points to the HF Space API instead of trying to call a local /api/chat:

// ORIGINAL (doesn't work on your domain):
const API_BASE = window.location.origin;  // → https://horizoncorelabs.studio/api/chat (404)

// SELF-HOSTED (works):
const API_BASE = "https://huggingface.co/spaces/Raiff1982/codette-ai";  // → HF Space API

Deployment Steps

Option A: Direct File Upload (Easiest)

  1. Download index-selfhosted.html from the repo
  2. Rename to index.html
  3. Upload to your web server at:
    /horizoncoreai/index.html
  4. Access at:
    https://horizoncorelabs.studio/horizoncoreai/

Option B: Git Deployment

If you're deploying via git:

# In your horizoncorelabs.studio repo
cp path-to-codette-repo/codette-ai-space/index-selfhosted.html horizoncoreai/index.html
git add horizoncoreai/index.html
git commit -m "Add Codette AI widget"
git push

Option C: Docker/Container Deployment

If you're running a container to serve static files:

FROM nginx:alpine
COPY index-selfhosted.html /usr/share/nginx/html/horizoncoreai/index.html
EXPOSE 80

Then access at: https://horizoncorelabs.studio/horizoncoreai/

Testing

Open your browser console (F12) and test:

fetch("https://huggingface.co/spaces/Raiff1982/codette-ai/api/chat", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({ messages: [{ role: "user", content: "hello" }] })
})
.then(r => r.json())
.then(d => console.log("API OK:", d))
.catch(e => console.error("API Error:", e))

Should return JSON with metadata + streaming response.

Features Included

  • ✓ Real-time streaming chat UI
  • ✓ Message history
  • ✓ Adapter/complexity metadata display
  • ✓ AEGIS security status
  • ✓ Cocoon memory counter
  • ✓ Suggestion chips (auto-populated)
  • ✓ Dark theme (lime green + cyan accents)
  • ✓ Mobile responsive

Troubleshooting

"Cannot reach Codette" error:

  • HF Space may be waking up (first request is slow)
  • Try again in 30 seconds
  • Check browser console for CORS errors (should not happen — CORS is enabled)

Widget loads but no response:

  • Check API endpoint in browser console (should be https://huggingface.co/spaces/Raiff1982/codette-ai/api/chat)
  • Open DevTools → Network tab → check /api/chat request
  • Verify HF Space is running at https://huggingface.co/spaces/Raiff1982/codette-ai

Styling looks wrong:

  • Make sure fonts load (Google Fonts CDN should work)
  • Check CSS variables in browser DevTools if colors are off

What's Live Behind This Widget

  • Model: Llama 3.1 8B via HF Inference API
  • 9 Adapters: Newton, DaVinci, Empathy, Philosophy, Quantum, Consciousness, Multi-Perspective, Systems, Orchestrator
  • Hallucination Guard: Real-time detection across all domains
  • Music Production Expert: Grounded in real DAWs, plugins, frequencies
  • Ethical Governance: AEGIS-lite pattern-based safety checks
  • Behavioral Locks: 4 permanent constraints on all responses
  • Cocoon Memory: In-memory reasoning history (up to 500 cocoons)

Live Demo

Test the widget live at: https://huggingface.co/spaces/Raiff1982/codette-ai