diff --git a/configs/llama-server.service b/configs/llama-server.service
new file mode 100644
index 0000000..ecc5a5c
--- /dev/null
+++ b/configs/llama-server.service
@@ -0,0 +1,22 @@
+[Unit]
+Description=llama.cpp inference server for Timmy
+After=network.target
+
+[Service]
+Type=simple
+User=root
+WorkingDirectory=/root/timmy
+ExecStart=/root/timmy/llama-server \
+    -m /root/timmy/models/hermes-3-8b.Q4_K_M.gguf \
+    --host 127.0.0.1 \
+    --port 8081 \
+    -c 8192 \
+    -np 1 \
+    --jinja \
+    -ngl 0
+Restart=always
+RestartSec=10
+Environment="HOME=/root"
+
+[Install]
+WantedBy=multi-user.target
diff --git a/configs/timmy-agent.service b/configs/timmy-agent.service
new file mode 100644
index 0000000..e4e0d76
--- /dev/null
+++ b/configs/timmy-agent.service
@@ -0,0 +1,17 @@
+[Unit]
+Description=Timmy Agent Harness
+After=llama-server.service
+Requires=llama-server.service
+
+[Service]
+Type=simple
+User=root
+WorkingDirectory=/root/timmy
+ExecStart=/root/timmy/venv/bin/python /root/timmy/timmy-home/agent/agent_daemon.py
+Restart=always
+RestartSec=30
+Environment="HOME=/root"
+Environment="TIMMY_MODEL_URL=http://127.0.0.1:8081"
+
+[Install]
+WantedBy=multi-user.target
diff --git a/scripts/provision-timmy-vps.sh b/scripts/provision-timmy-vps.sh
new file mode 100644
index 0000000..b9978bc
--- /dev/null
+++ b/scripts/provision-timmy-vps.sh
@@ -0,0 +1,260 @@
+#!/bin/bash
+# Timmy VPS Provisioning Script
+# Transforms fresh Ubuntu 22.04+ VPS into sovereign local-first wizard
+
+set -e
+
+TIMMY_USER="${TIMMY_USER:-root}"
+TIMMY_HOME="${TIMMY_HOME:-/root}"
+TIMMY_DIR="$TIMMY_HOME/timmy"
+REPO_URL="${REPO_URL:-http://143.198.27.163:3000/Timmy_Foundation/timmy-home.git}"
+MODEL_URL="${MODEL_URL:-https://huggingface.co/TheBloke/Hermes-3-Llama-3.1-8B-GGUF/resolve/main/hermes-3-llama-3.1-8b.Q4_K_M.gguf}"
+MODEL_NAME="${MODEL_NAME:-hermes-3-8b.Q4_K_M.gguf}"
+
+echo "========================================"
+echo "  Timmy VPS Provisioning"
+echo "========================================"
+echo ""
+
+# Colors
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+NC='\033[0m' # No Color
+
+log() {
+    echo -e "${GREEN}[TIMMY]${NC} $1"
+}
+
+warn() {
+    echo -e "${YELLOW}[WARN]${NC} $1"
+}
+
+error() {
+    echo -e "${RED}[ERROR]${NC} $1"
+}
+
+# Check if running as root
+if [ "$EUID" -ne 0 ]; then
+    error "Please run as root"
+    exit 1
+fi
+
+# Check Ubuntu version
+if ! grep -q "Ubuntu 22.04\|Ubuntu 24.04" /etc/os-release; then
+    warn "Not Ubuntu 22.04/24.04 - may not work correctly"
+fi
+
+log "Step 1/8: Installing system dependencies..."
+export DEBIAN_FRONTEND=noninteractive
+apt-get update -qq
+apt-get install -y -qq \
+    build-essential \
+    cmake \
+    git \
+    curl \
+    wget \
+    python3 \
+    python3-pip \
+    python3-venv \
+    libopenblas-dev \
+    pkg-config \
+    ufw \
+    jq \
+    sqlite3 \
+    libsqlite3-dev \
+    2>&1 | tail -5
+
+log "Step 2/8: Setting up directory structure..."
+mkdir -p "$TIMMY_DIR"/{soul,scripts,logs,shared,models,configs}
+mkdir -p "$TIMMY_HOME/.config/systemd/user"
+
+log "Step 3/8: Building llama.cpp from source..."
+if [ ! -f "$TIMMY_DIR/llama-server" ]; then
+    cd /tmp
+    git clone --depth 1 https://github.com/ggerganov/llama.cpp.git 2>/dev/null || true
+    cd llama.cpp
+    
+    # Build with OpenBLAS for CPU optimization
+    cmake -B build \
+        -DGGML_BLAS=ON \
+        -DGGML_BLAS_VENDOR=OpenBLAS \
+        -DLLAMA_BUILD_TESTS=OFF \
+        -DLLAMA_BUILD_EXAMPLES=OFF \
+        -DCMAKE_BUILD_TYPE=Release
+    
+    cmake --build build --config Release -j$(nproc)
+    
+    # Copy binaries
+    cp build/bin/llama-server "$TIMMY_DIR/"
+    cp build/bin/llama-cli "$TIMMY_DIR/"
+    
+    log "llama.cpp built successfully"
+else
+    log "llama.cpp already exists, skipping build"
+fi
+
+log "Step 4/8: Downloading model weights..."
+if [ ! -f "$TIMMY_DIR/models/$MODEL_NAME" ]; then
+    cd "$TIMMY_DIR/models"
+    wget -q --show-progress "$MODEL_URL" -O "$MODEL_NAME" || {
+        error "Failed to download model. Continuing anyway..."
+    }
+    log "Model downloaded"
+else
+    log "Model already exists, skipping download"
+fi
+
+log "Step 5/8: Setting up llama-server systemd service..."
+cat > /etc/systemd/system/llama-server.service << EOF
+[Unit]
+Description=llama.cpp inference server for Timmy
+After=network.target
+
+[Service]
+Type=simple
+User=$TIMMY_USER
+WorkingDirectory=$TIMMY_DIR
+ExecStart=$TIMMY_DIR/llama-server \\
+    -m $TIMMY_DIR/models/$MODEL_NAME \\
+    --host 127.0.0.1 \\
+    --port 8081 \\
+    -c 8192 \\
+    -np 1 \\
+    --jinja \\
+    -ngl 0
+Restart=always
+RestartSec=10
+Environment="HOME=$TIMMY_HOME"
+
+[Install]
+WantedBy=multi-user.target
+EOF
+
+systemctl daemon-reload
+systemctl enable llama-server.service
+
+log "Step 6/8: Cloning timmy-home repo and setting up agent..."
+if [ ! -d "$TIMMY_DIR/timmy-home" ]; then
+    cd "$TIMMY_DIR"
+    git clone "$REPO_URL" timmy-home 2>/dev/null || warn "Could not clone repo"
+fi
+
+# Create minimal Python environment for agent
+if [ ! -d "$TIMMY_DIR/venv" ]; then
+    python3 -m venv "$TIMMY_DIR/venv"
+    "$TIMMY_DIR/venv/bin/pip" install -q requests pyyaml 2>&1 | tail -3
+fi
+
+log "Step 7/8: Setting up Timmy agent systemd service..."
+cat > /etc/systemd/system/timmy-agent.service << EOF
+[Unit]
+Description=Timmy Agent Harness
+After=llama-server.service
+Requires=llama-server.service
+
+[Service]
+Type=simple
+User=$TIMMY_USER
+WorkingDirectory=$TIMMY_DIR
+ExecStart=$TIMMY_DIR/venv/bin/python $TIMMY_DIR/timmy-home/agent/agent_daemon.py
+Restart=always
+RestartSec=30
+Environment="HOME=$TIMMY_HOME"
+Environment="TIMMY_MODEL_URL=http://127.0.0.1:8081"
+
+[Install]
+WantedBy=multi-user.target
+EOF
+
+systemctl daemon-reload
+systemctl enable timmy-agent.service
+
+log "Step 8/8: Configuring firewall..."
+# Reset UFW
+ufw --force reset 2>/dev/null || true
+ufw default deny incoming
+ufw default allow outgoing
+
+# Allow SSH
+ufw allow 22/tcp
+
+# Allow Syncthing (sync protocol)
+ufw allow 22000/tcp
+ufw allow 22000/udp
+
+# Allow Syncthing (discovery)
+ufw allow 21027/udp
+
+# Note: llama-server on 8081 is NOT exposed (localhost only)
+
+ufw --force enable
+
+log "Starting services..."
+systemctl start llama-server.service || warn "llama-server failed to start (may need model)"
+
+# Wait for llama-server to be ready
+log "Waiting for llama-server to be ready..."
+for i in {1..30}; do
+    if curl -s http://127.0.0.1:8081/health >/dev/null 2>&1; then
+        log "llama-server is healthy!"
+        break
+    fi
+    sleep 2
+done
+
+# Create status script
+cat > "$TIMMY_DIR/scripts/status.sh" << 'EOF'
+#!/bin/bash
+echo "=== Timmy VPS Status ==="
+echo ""
+echo "Services:"
+systemctl is-active llama-server.service && echo "  llama-server: RUNNING" || echo "  llama-server: STOPPED"
+systemctl is-active timmy-agent.service && echo "  timmy-agent: RUNNING" || echo "  timmy-agent: STOPPED"
+echo ""
+echo "Inference Health:"
+curl -s http://127.0.0.1:8081/health | jq . 2>/dev/null || echo "  Not responding"
+echo ""
+echo "Disk Usage:"
+df -h $HOME | tail -1
+echo ""
+echo "Memory:"
+free -h | grep Mem
+EOF
+chmod +x "$TIMMY_DIR/scripts/status.sh"
+
+# Create README
+cat > "$TIMMY_DIR/README.txt" << EOF
+Timmy Sovereign Wizard VPS
+==========================
+
+Quick Commands:
+  $TIMMY_DIR/scripts/status.sh    - Check system status
+  systemctl status llama-server   - Check inference service
+  systemctl status timmy-agent    - Check agent service
+
+Directories:
+  $TIMMY_DIR/models/     - AI model weights
+  $TIMMY_DIR/soul/       - SOUL.md and conscience files
+  $TIMMY_DIR/logs/       - Agent logs
+  $TIMMY_DIR/shared/     - Syncthing shared folder
+
+Inference Endpoint:
+  http://127.0.0.1:8081 (localhost only)
+
+Provisioning complete!
+EOF
+
+echo ""
+echo "========================================"
+log "Provisioning Complete!"
+echo "========================================"
+echo ""
+echo "Status:"
+"$TIMMY_DIR/scripts/status.sh"
+echo ""
+echo "Next steps:"
+echo "  1. Run syncthing setup: curl -sL $REPO_URL/raw/branch/main/scripts/setup-syncthing.sh | bash"
+echo "  2. Check inference: curl http://127.0.0.1:8081/health"
+echo "  3. Review logs: journalctl -u llama-server -f"
+echo ""