Initial metrics skill with aggregation

SKILL.md

@@ -0,0 +1,73 @@
+---
+name: metrics
+description: Metrics aggregator with system and skill metrics
+metadata:
+  version: "1.0.0"
+  vibestack:
+    main: false
+    metrics-port: 9090
+---
+
+# Metrics Skill
+
+Aggregates metrics from system (node_exporter) and other skills into a single `/metrics` endpoint.
+
+## Features
+
+- System metrics via node_exporter (CPU, memory, disk, network)
+- Auto-discovers skills that expose metrics
+- Single aggregated endpoint for Prometheus scraping
+- Auto-registers with Caddy if present
+
+## How Skills Expose Metrics
+
+Skills can expose metrics by adding to their SKILL.md:
+
+```yaml
+metadata:
+  vibestack:
+    metrics-port: 8080
+```
+
+Or by setting environment variable:
+```bash
+METRICS_PORT_myskill=8080
+```
+
+The aggregator will scrape `localhost:{port}/metrics` for each discovered skill.
+
+## Configuration
+
+| Variable | Description | Default |
+|----------|-------------|---------|
+| `METRICS_PORT` | Aggregator port | `9090` |
+| `NODE_EXPORTER_PORT` | node_exporter port | `9100` |
+| `METRICS_DOMAIN` | Domain for Caddy auto-config | (none) |
+| `METRICS_PATH` | URL path for metrics | `/metrics` |
+
+## Collected Metrics
+
+### System (node_exporter)
+- `node_cpu_*` - CPU usage
+- `node_memory_*` - Memory usage
+- `node_disk_*` - Disk I/O
+- `node_network_*` - Network I/O
+- `node_filesystem_*` - Filesystem usage
+
+### Per-Skill
+Each skill's metrics are prefixed with the skill name:
+```
+# From ttyd skill
+ttyd_connections_active 5
+ttyd_sessions_total 123
+```
+
+## Usage
+
+```bash
+# Scrape aggregated metrics
+curl http://localhost:9090/metrics
+
+# With Caddy proxy
+curl https://metrics.example.com/metrics
+```

scripts/autorun.sh

@@ -0,0 +1,124 @@
+#!/bin/bash
+set -e
+
+NODE_EXPORTER_VERSION="${NODE_EXPORTER_VERSION:-1.7.0}"
+SKILLS_DIR="${SKILLS_DIR:-/skills}"
+
+# Idempotent node_exporter installation
+install_node_exporter() {
+    if command -v node_exporter &>/dev/null; then
+        echo "node_exporter already installed"
+        return 0
+    fi
+
+    echo "Installing node_exporter v${NODE_EXPORTER_VERSION}..."
+
+    local arch
+    case "$(uname -m)" in
+        x86_64)  arch="amd64" ;;
+        aarch64) arch="arm64" ;;
+        armv7l)  arch="armv7" ;;
+        *)
+            echo "Unsupported architecture: $(uname -m)"
+            exit 1
+            ;;
+    esac
+
+    local url="https://github.com/prometheus/node_exporter/releases/download/v${NODE_EXPORTER_VERSION}/node_exporter-${NODE_EXPORTER_VERSION}.linux-${arch}.tar.gz"
+
+    curl -sSL "$url" -o /tmp/node_exporter.tar.gz
+    tar -xzf /tmp/node_exporter.tar.gz -C /tmp
+    mv /tmp/node_exporter-*/node_exporter /usr/local/bin/
+    rm -rf /tmp/node_exporter*
+
+    echo "node_exporter installed"
+}
+
+# Discover skills with metrics ports
+discover_metrics_targets() {
+    local targets_file="/tmp/metrics_targets.txt"
+    echo "# Auto-discovered metrics targets" > "$targets_file"
+
+    for skill_dir in "$SKILLS_DIR"/*/; do
+        local skill_name=$(basename "$skill_dir")
+        local skill_md="$skill_dir/SKILL.md"
+
+        # Skip self
+        [ "$skill_name" = "metrics" ] && continue
+
+        # Check for metrics-port in SKILL.md
+        if [ -f "$skill_md" ]; then
+            local port=$(yq -r '.metadata.vibestack."metrics-port" // empty' < <(sed -n '/^---$/,/^---$/p' "$skill_md" | sed '1d;$d') 2>/dev/null)
+            if [ -n "$port" ] && [ "$port" != "null" ]; then
+                echo "$skill_name localhost:$port" >> "$targets_file"
+                echo "  $skill_name: metrics on port $port"
+            fi
+        fi
+
+        # Check for METRICS_PORT_skillname env var
+        local env_var="METRICS_PORT_${skill_name//-/_}"
+        local env_port="${!env_var}"
+        if [ -n "$env_port" ]; then
+            echo "$skill_name localhost:$env_port" >> "$targets_file"
+            echo "  $skill_name: metrics on port $env_port (from env)"
+        fi
+    done
+
+    echo "Targets written to $targets_file"
+}
+
+# Configure Caddy if present
+configure_caddy() {
+    local caddy_dir="$SKILLS_DIR/caddy"
+    local metrics_port="${METRICS_PORT:-9090}"
+    local metrics_domain="${METRICS_DOMAIN:-}"
+
+    if [ ! -d "$caddy_dir" ]; then
+        echo "Caddy not found - metrics on port $metrics_port"
+        return 0
+    fi
+
+    echo "Caddy detected - configuring reverse proxy..."
+    mkdir -p "$caddy_dir/snippets.d"
+
+    local snippet="$caddy_dir/snippets.d/metrics.caddy"
+
+    if [ -n "$metrics_domain" ]; then
+        cat > "$snippet" << EOF
+# Auto-generated by metrics skill
+$metrics_domain {
+    reverse_proxy localhost:$metrics_port
+}
+EOF
+        echo "Caddy config: $metrics_domain -> localhost:$metrics_port"
+    else
+        cat > "$snippet" << EOF
+# Auto-generated by metrics skill
+# Add to your site block:
+# handle /metrics {
+#     reverse_proxy localhost:$metrics_port
+# }
+EOF
+        echo "Caddy snippet created (manual config needed)"
+    fi
+}
+
+# Install socat for HTTP server
+install_socat() {
+    if command -v socat &>/dev/null; then
+        echo "socat already installed"
+        return 0
+    fi
+
+    echo "Installing socat..."
+    apt-get update
+    apt-get install -y socat
+    echo "socat installed"
+}
+
+install_node_exporter
+install_socat
+discover_metrics_targets
+configure_caddy
+
+echo "Metrics setup complete"

scripts/run.sh

@@ -0,0 +1,107 @@
+#!/bin/bash
+set -e
+
+SCRIPT_DIR="$(dirname "$0")"
+METRICS_PORT="${METRICS_PORT:-9090}"
+NODE_EXPORTER_PORT="${NODE_EXPORTER_PORT:-9100}"
+TARGETS_FILE="/tmp/metrics_targets.txt"
+
+# Start node_exporter in background
+start_node_exporter() {
+    echo "Starting node_exporter on port $NODE_EXPORTER_PORT..."
+    node_exporter --web.listen-address=":$NODE_EXPORTER_PORT" &
+    NODE_EXPORTER_PID=$!
+    sleep 2
+
+    if ! kill -0 $NODE_EXPORTER_PID 2>/dev/null; then
+        echo "Failed to start node_exporter"
+        exit 1
+    fi
+    echo "node_exporter running (PID $NODE_EXPORTER_PID)"
+}
+
+# Create the aggregator CGI script
+create_aggregator_script() {
+    cat > /tmp/aggregate.sh << 'SCRIPT'
+#!/bin/bash
+NODE_EXPORTER_PORT="${NODE_EXPORTER_PORT:-9100}"
+TARGETS_FILE="/tmp/metrics_targets.txt"
+
+# Collect all metrics
+{
+    # System metrics
+    echo "# HELP vibestack_up Whether the metrics aggregator is up"
+    echo "# TYPE vibestack_up gauge"
+    echo "vibestack_up 1"
+    echo ""
+
+    # node_exporter metrics
+    curl -s "http://localhost:$NODE_EXPORTER_PORT/metrics" 2>/dev/null || true
+
+    # Skill metrics
+    if [ -f "$TARGETS_FILE" ]; then
+        while IFS=' ' read -r skill_name target || [ -n "$skill_name" ]; do
+            [[ "$skill_name" =~ ^#.*$ ]] && continue
+            [ -z "$skill_name" ] && continue
+            [ -z "$target" ] && continue
+
+            echo ""
+            echo "# Metrics from skill: $skill_name"
+            curl -s "http://${target}/metrics" 2>/dev/null || true
+        done < "$TARGETS_FILE"
+    fi
+}
+SCRIPT
+    chmod +x /tmp/aggregate.sh
+}
+
+# Serve metrics using socat
+serve_metrics() {
+    echo "Starting metrics aggregator on port $METRICS_PORT..."
+
+    if ! command -v socat &>/dev/null; then
+        echo "Installing socat..."
+        apt-get update && apt-get install -y socat
+    fi
+
+    # Use socat to handle HTTP requests
+    exec socat TCP-LISTEN:$METRICS_PORT,reuseaddr,fork EXEC:"/tmp/metrics_handler.sh"
+}
+
+# Create HTTP handler script
+create_handler_script() {
+    cat > /tmp/metrics_handler.sh << 'HANDLER'
+#!/bin/bash
+# Read HTTP request (we ignore it, always serve /metrics)
+read -r request_line
+while read -r header; do
+    [ "$header" = $'\r' ] && break
+done
+
+# Get metrics
+metrics=$(/tmp/aggregate.sh)
+content_length=${#metrics}
+
+# Send HTTP response
+printf "HTTP/1.1 200 OK\r\n"
+printf "Content-Type: text/plain; charset=utf-8\r\n"
+printf "Content-Length: %d\r\n" "$content_length"
+printf "Connection: close\r\n"
+printf "\r\n"
+printf "%s" "$metrics"
+HANDLER
+    chmod +x /tmp/metrics_handler.sh
+}
+
+# Cleanup on exit
+cleanup() {
+    echo "Shutting down..."
+    [ -n "$NODE_EXPORTER_PID" ] && kill $NODE_EXPORTER_PID 2>/dev/null
+    exit 0
+}
+trap cleanup SIGTERM SIGINT
+
+start_node_exporter
+create_aggregator_script
+create_handler_script
+serve_metrics