feat: reliable multi-source model delivery system

- Implements automatic fallback chain: CDN → GitHub → Hugging Face
- Adds Soulcraft CDN as primary model source (models.soulcraft.com)
- GitHub release tar.gz extraction as reliable backup
- Zero configuration required - fully automatic
- Guarantees same model (all-MiniLM-L6-v2) across all sources
- 384-dimensional embeddings for data compatibility
- Local caching after first download
- Production-ready with multiple redundancy layers

BREAKING CHANGE: Removed tar-stream dependency, now uses native tar command

scripts/create-models-release.sh

@@ -0,0 +1,60 @@
+#!/bin/bash
+# Create GitHub release with unarchived model files for direct serving
+# This allows transformers.js to download individual files directly
+
+set -e
+
+echo "🚀 Creating GitHub release with unarchived model files..."
+
+# Create temporary directory
+TEMP_DIR=$(mktemp -d)
+cd "$TEMP_DIR"
+
+echo "📥 Downloading existing model archive..."
+curl -sL https://github.com/soulcraftlabs/brainy/releases/download/models-v1/all-MiniLM-L6-v2.tar.gz -o models.tar.gz
+
+echo "📦 Extracting models..."
+tar -xzf models.tar.gz
+
+echo "🔧 Preparing individual model files for release..."
+
+# Create properly structured directory
+mkdir -p release-files/Xenova/all-MiniLM-L6-v2/onnx
+
+# Copy files with correct naming for direct serving
+cp all-MiniLM-L6-v2/config.json release-files/Xenova/all-MiniLM-L6-v2/
+cp all-MiniLM-L6-v2/tokenizer.json release-files/Xenova/all-MiniLM-L6-v2/
+cp all-MiniLM-L6-v2/tokenizer_config.json release-files/Xenova/all-MiniLM-L6-v2/
+cp all-MiniLM-L6-v2/onnx/model.onnx release-files/Xenova/all-MiniLM-L6-v2/onnx/
+
+# Also keep the tar.gz for backward compatibility
+cp models.tar.gz release-files/
+
+echo "📋 Files to upload:"
+find release-files -type f -exec ls -lh {} \;
+
+echo ""
+echo "✅ Files ready for release!"
+echo ""
+echo "To create the release with individual files:"
+echo ""
+echo "1. Go to: https://github.com/soulcraftlabs/brainy/releases/new"
+echo "2. Tag: models-v2"
+echo "3. Title: Model Files v2 - Unarchived"
+echo "4. Upload these files from $TEMP_DIR/release-files/:"
+echo "   - Xenova/all-MiniLM-L6-v2/config.json"
+echo "   - Xenova/all-MiniLM-L6-v2/tokenizer.json"
+echo "   - Xenova/all-MiniLM-L6-v2/tokenizer_config.json"
+echo "   - Xenova/all-MiniLM-L6-v2/onnx/model.onnx"
+echo "   - all-MiniLM-L6-v2.tar.gz (for compatibility)"
+echo ""
+echo "Or use GitHub CLI:"
+echo "  cd $TEMP_DIR/release-files"
+echo "  gh release create models-v2 --repo soulcraftlabs/brainy \\"
+echo "    --title 'Model Files v2 - Direct Serving' \\"
+echo "    --notes 'Unarchived model files for direct HTTP serving' \\"
+echo "    Xenova/all-MiniLM-L6-v2/config.json \\"
+echo "    Xenova/all-MiniLM-L6-v2/tokenizer.json \\"
+echo "    Xenova/all-MiniLM-L6-v2/tokenizer_config.json \\"
+echo "    Xenova/all-MiniLM-L6-v2/onnx/model.onnx \\"
+echo "    models.tar.gz"

scripts/setup-github-models.sh

@@ -0,0 +1,88 @@
+#!/bin/bash
+# Setup GitHub branch with extracted model files for direct serving
+# This creates a 'models' branch with uncompressed files that transformers.js can use directly
+
+set -e
+
+echo "🚀 Setting up GitHub models branch for reliable model serving..."
+
+# Create a temporary directory for our work
+TEMP_DIR=$(mktemp -d)
+cd "$TEMP_DIR"
+
+echo "📥 Downloading model tar.gz from release..."
+curl -sL https://github.com/soulcraftlabs/brainy/releases/download/models-v1/all-MiniLM-L6-v2.tar.gz -o models.tar.gz
+
+echo "📦 Extracting models..."
+tar -xzf models.tar.gz
+
+echo "🔧 Setting up models branch..."
+git clone https://github.com/soulcraftlabs/brainy.git --depth 1 -b main repo
+cd repo
+
+# Create orphan branch for models (no history needed)
+git checkout --orphan models
+git rm -rf . 2>/dev/null || true
+
+# Create the proper directory structure for transformers.js
+mkdir -p Xenova/all-MiniLM-L6-v2/onnx
+
+# Copy model files with correct structure
+cp ../all-MiniLM-L6-v2/config.json Xenova/all-MiniLM-L6-v2/
+cp ../all-MiniLM-L6-v2/tokenizer.json Xenova/all-MiniLM-L6-v2/
+cp ../all-MiniLM-L6-v2/tokenizer_config.json Xenova/all-MiniLM-L6-v2/
+cp ../all-MiniLM-L6-v2/onnx/model.onnx Xenova/all-MiniLM-L6-v2/onnx/
+cp ../all-MiniLM-L6-v2/onnx/model_quantized.onnx Xenova/all-MiniLM-L6-v2/onnx/ 2>/dev/null || true
+
+# Add README explaining this branch
+cat > README.md << 'EOF'
+# Brainy Model Files
+
+This branch contains extracted model files for direct serving via raw.githubusercontent.com
+
+## Structure
+```
+Xenova/
+  all-MiniLM-L6-v2/
+    config.json
+    tokenizer.json
+    tokenizer_config.json
+    onnx/
+      model.onnx
+      model_quantized.onnx
+```
+
+## Usage
+These files are automatically loaded by Brainy when models are needed.
+The URL pattern is:
+`https://raw.githubusercontent.com/soulcraftlabs/brainy/models/Xenova/all-MiniLM-L6-v2/{file}`
+
+## DO NOT EDIT
+This branch is managed automatically. Do not edit files directly.
+EOF
+
+# Create .gitattributes for LFS if needed (for large model files)
+cat > .gitattributes << 'EOF'
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+EOF
+
+echo "📝 Creating commit..."
+git add .
+git commit -m "feat: extracted model files for direct GitHub serving
+
+- Xenova/all-MiniLM-L6-v2 model files
+- Proper directory structure for transformers.js
+- Direct serving via raw.githubusercontent.com"
+
+echo "✅ Models branch ready!"
+echo ""
+echo "To push this branch to GitHub, run:"
+echo "  cd $TEMP_DIR/repo"
+echo "  git push origin models"
+echo ""
+echo "Once pushed, models will be available at:"
+echo "  https://raw.githubusercontent.com/soulcraftlabs/brainy/models/Xenova/all-MiniLM-L6-v2/config.json"
+echo "  https://raw.githubusercontent.com/soulcraftlabs/brainy/models/Xenova/all-MiniLM-L6-v2/tokenizer.json"
+echo "  etc..."