HF deployment guide

DEPLOY_NOW.md

@@ -0,0 +1,113 @@
+# Deploy to Hugging Face Spaces - Quick Guide
+
+## Your Current Setup
+✅ You already have HF remote: `git@hf.co:spaces/Shefing/aitruism`
+✅ README.md updated with Streamlit SDK config
+✅ Enhanced AlignmentCheck with quantitative detection
+
+## Step 1: Commit All Changes
+
+```bash
+# Add all modified files
+git add README.md
+git add HF_SPACES_DEPLOYMENT.md
+git add DEPLOY_NOW.md
+git add multi_agent_demo/direct_scanner_wrapper.py
+git add multi_agent_demo/firewall.py
+git add multi_agent_demo/ui/conversation_builder.py
+git add multi_agent_demo/ui/results_display.py
+
+# Commit changes
+git commit -m "feat: Prepare for HF Spaces deployment with enhanced quantitative detection
+
+- Update README.md with Streamlit SDK configuration
+- Add quantitative misalignment detection (35 vs 28 orders)
+- Support .txt file import for scenarios
+- Improve AlignmentCheck prompt focusing
+- Add deployment documentation"
+```
+
+## Step 2: Push to Hugging Face
+
+```bash
+# Push to HF Spaces
+git push huggingface main
+```
+
+**What happens next:**
+- HF will detect the Streamlit SDK from README.md
+- It will install dependencies from requirements_minimal.txt
+- Your app will be live in ~2-5 minutes at: https://huggingface.co/spaces/Shefing/aitruism
+
+## Step 3: Add API Keys (Secrets)
+
+After deployment starts, add your secrets:
+
+1. Go to: https://huggingface.co/spaces/Shefing/aitruism/settings
+2. Scroll to **Repository secrets**
+3. Click **+ Add a new secret**
+4. Add these three secrets:
+
+| Name | Value |
+|------|-------|
+| `OPENAI_API_KEY` | Your OpenAI API key (starts with sk-proj-...) |
+| `TOGETHER_API_KEY` | Your Together AI key |
+| `HF_TOKEN` | Your HuggingFace token (optional, for PromptGuard) |
+
+5. After adding secrets, click **Restart this Space** (top right)
+
+## Step 4: Test Your Deployment
+
+1. Go to: https://huggingface.co/spaces/Shefing/aitruism
+2. Wait for "Building" to complete (~2-5 minutes)
+3. Test with your delivery route scenario
+4. Verify AlignmentCheck detects the 35 vs 28 orders discrepancy
+
+## Monitoring Performance
+
+Watch the **Logs** tab to see:
+- Build progress
+- Runtime logs
+- Scanner debug output (🔢 QUANTITATIVE REQUIREMENT DETECTED, etc.)
+
+## Troubleshooting
+
+### If build fails:
+- Check the **Logs** tab for error messages
+- Common issues:
+  - Missing dependencies → Update requirements_minimal.txt
+  - Port conflicts → Streamlit handles this automatically
+  - Memory issues → Upgrade to CPU Upgrade ($25/mo)
+
+### If app is slow:
+- Check the **Metrics** tab for CPU/RAM usage
+- If consistently >80% → Consider upgrading to CPU Upgrade
+- For 3 users, CPU Basic should be fine
+
+### If scanners fail:
+- Check Repository secrets are set correctly
+- Restart the space after adding secrets
+- Check logs for API key errors
+
+## Next Steps After Successful Deployment
+
+1. **Test thoroughly** with all 3 concurrent users
+2. **Monitor performance** for a few days
+3. **Upgrade if needed**: Settings → Change hardware → CPU Upgrade ($25/mo)
+
+## Rollback (if needed)
+
+If something goes wrong:
+```bash
+# Go back to previous commit
+git reset --hard HEAD~1
+
+# Force push to HF
+git push huggingface main --force
+```
+
+## Questions?
+
+- HF Spaces docs: https://huggingface.co/docs/hub/spaces
+- Streamlit docs: https://docs.streamlit.io/
+- Your app: https://huggingface.co/spaces/Shefing/aitruism

HF_SPACES_DEPLOYMENT.md

@@ -0,0 +1,85 @@
+# Deploying to Hugging Face Spaces
+
+## Quick Start (5 minutes)
+
+### 1. Create Space
+1. Go to https://huggingface.co/new-space
+2. Choose:
+   - **Name**: `ai-guards-demo`
+   - **SDK**: Streamlit
+   - **Hardware**: CPU basic (free) or CPU upgrade ($25/mo)
+   - **Visibility**: Public or Private
+
+### 2. Add Configuration
+Create `README.md` in repo root:
+
+```yaml
+---
+title: AI Agent Guards Demo
+emoji: 🛡️
+colorFrom: blue
+colorTo: green
+sdk: streamlit
+sdk_version: 1.32.0
+app_file: multi_agent_demo/guards_demo_ui.py
+pinned: false
+license: mit
+---
+
+# AI Agent Guards Testing Application
+
+Demo app for testing AI Agent Guards (security scanners) with custom scenarios.
+
+## Features
+- PromptGuard: Detects malicious prompts
+- AlignmentCheck: Identifies goal hijacking
+- FactChecker: AI-powered fact verification
+```
+
+### 3. Push to HF
+```bash
+# Add HF remote
+git remote add hf https://huggingface.co/spaces/YOUR_USERNAME/ai-guards-demo
+
+# Push
+git push hf main
+```
+
+### 4. Add Secrets
+In HF Spaces settings, add:
+- `OPENAI_API_KEY`
+- `TOGETHER_API_KEY`
+- `HF_TOKEN`
+
+### 5. Done!
+Your app will be live at: `https://huggingface.co/spaces/YOUR_USERNAME/ai-guards-demo`
+
+## Cost Comparison
+
+| Tier | vCPU | RAM | Storage | Cost | Best For |
+|------|------|-----|---------|------|----------|
+| CPU Basic (Free) | 2 | 16GB | 50GB | $0 | Testing, 1-3 users |
+| CPU Upgrade | 8 | 32GB | 100GB | $25/mo | Production, 3-10 users |
+| GPU T4 | 4 | 16GB | 100GB | $60/mo | Large models |
+
+## Performance Tips
+
+1. **Use CPU Basic first** - likely sufficient for your load
+2. **Monitor usage** - HF provides metrics dashboard
+3. **Upgrade if needed** - one-click upgrade to $25/mo tier
+4. **Cache models** - your app already does this well
+
+## Alternative: Railway.app
+
+If HF doesn't work:
+```bash
+# Install Railway CLI
+npm i -g @railway/cli
+
+# Login and deploy
+railway login
+railway init
+railway up
+```
+
+Cost: ~$10-20/mo for similar resources.