crewswarm 0.8.1-beta

package/scripts/refactor-configs.mjs

@@ -0,0 +1,39 @@
+import fs from 'fs';
+import path from 'path';
+import { execSync } from 'child_process';
+
+console.log('Beginning config path consolidation...');
+
+// 1. Files containing config.json
+const filesRaw = execSync('git grep -l "config.json" || true').toString().split('\n').filter(Boolean);
+let updatedCount = 0;
+
+for (const file of filesRaw) {
+  // skip package.json, lock files, and this very script
+  if (file.includes('package') || file.includes('refactor-configs.mjs') || file.includes('CHANGELOG.md')) continue;
+  
+  const fullPath = path.resolve(process.cwd(), file);
+  if (!fs.existsSync(fullPath)) continue;
+
+  let content = fs.readFileSync(fullPath, 'utf8');
+  const og = content;
+  
+  // Replace direct string paths
+  content = content.replace(/\.crewswarm[\/\\]config\.json/g, '.crewswarm/crewswarm.json');
+  
+  // Replace path.join(..., "config.json") -> path.join(..., "crewswarm.json")
+  // We only replace "config.json" exactly.
+  content = content.replace(/"config\.json"/g, '"crewswarm.json"');
+  content = content.replace(/'config\.json'/g, "'crewswarm.json'");
+  
+  // Also handle `config.json` in markdown text
+  content = content.replace(/`~\/\.crewswarm\/config\.json`/g, '`~/.crewswarm/crewswarm.json`');
+  
+  if (content !== og) {
+    fs.writeFileSync(fullPath, content);
+    console.log(`Updated ${file}`);
+    updatedCount++;
+  }
+}
+
+console.log(`Successfully updated ${updatedCount} files to use crewswarm.json contextually.`);

package/scripts/release-check.sh

@@ -0,0 +1,46 @@
+#!/usr/bin/env bash
+# Lightweight public-release checklist runner.
+#
+# Usage:
+#   bash scripts/release-check.sh
+#   bash scripts/release-check.sh --full
+
+set -euo pipefail
+
+REPO_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+cd "$REPO_DIR"
+
+FULL=0
+for arg in "$@"; do
+  [[ "$arg" == "--full" ]] && FULL=1
+done
+
+echo ""
+echo "━━━ crewswarm release check ━━━"
+echo "repo: $REPO_DIR"
+echo "date: $(date)"
+
+echo ""
+echo "1. Doctor"
+npm run doctor
+
+echo ""
+echo "2. Static smoke"
+bash scripts/smoke.sh --no-build
+
+echo ""
+echo "3. Health"
+if [[ "${CI:-}" == "true" ]]; then
+  npm run health -- --quiet --no-services
+else
+  npm run health -- --quiet
+fi
+
+if [[ "$FULL" -eq 1 ]]; then
+  echo ""
+  echo "4. Live dispatch smoke"
+  npm run smoke
+fi
+
+echo ""
+echo "Release check passed."

package/scripts/resolve-node-bin.sh

@@ -0,0 +1,25 @@
+#!/usr/bin/env bash
+
+if [[ -n "${NODE:-}" && -x "${NODE}" ]]; then
+  printf "%s\n" "$NODE"
+  exit 0
+fi
+
+for candidate in \
+  "/usr/local/opt/node/bin/node" \
+  "/opt/homebrew/opt/node/bin/node" \
+  "/usr/local/bin/node" \
+  "/opt/homebrew/bin/node"
+do
+  if [[ -x "$candidate" ]]; then
+    printf "%s\n" "$candidate"
+    exit 0
+  fi
+done
+
+if command -v node >/dev/null 2>&1; then
+  command -v node
+  exit 0
+fi
+
+exit 1

package/scripts/restart-all-from-repo.sh

@@ -0,0 +1,329 @@
+#!/usr/bin/env bash
+# Restart the full crewswarm stack using only repo scripts (OpenCode, RT daemon, gateways, crew-lead, dashboard, Vibe, watch).
+# Run from repo root: ./scripts/restart-all-from-repo.sh
+# Optional: --no-dashboard  skip dashboard
+#           --no-studio    skip Vibe + file watcher
+#           --no-bridges   skip Telegram + WhatsApp bridges
+
+REPO_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+cd "$REPO_DIR"
+export CREWSWARM_DIR="$REPO_DIR"
+export OPENCLAW_DIR="$REPO_DIR"   # backward compat for scripts that only check this
+export NODE="$("$REPO_DIR/scripts/resolve-node-bin.sh")"
+# Work around intermittent Node 24/25 ESM loader crashes (`Unknown system error -11, read`)
+# seen in the RT daemon and Vibe startup paths on this machine.
+export NODE_DISABLE_COMPILE_CACHE="${NODE_DISABLE_COMPILE_CACHE:-1}"
+
+# ── Color helpers ─────────────────────────────────────────────────────────────
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[0;33m'
+BOLD='\033[1m'
+RESET='\033[0m'
+
+# ── Service status tracking ──────────────────────────────────────────────────
+declare -A SVC_STATUS  # "up" or "down"
+
+# ── Graceful kill: SIGTERM first, SIGKILL after 5s ────────────────────────────
+graceful_kill_pattern() {
+  local pattern="$1"
+  local pids
+  pids=$(pgrep -f "$pattern" 2>/dev/null || true)
+  if [[ -z "$pids" ]]; then return 0; fi
+  # Send SIGTERM
+  echo "$pids" | xargs kill -TERM 2>/dev/null || true
+  # Wait up to 5s for processes to exit
+  local waited=0
+  while [[ $waited -lt 50 ]]; do
+    if ! pgrep -f "$pattern" >/dev/null 2>&1; then
+      return 0
+    fi
+    sleep 0.1
+    waited=$((waited + 1))
+  done
+  # Still alive — SIGKILL
+  pgrep -f "$pattern" 2>/dev/null | xargs kill -9 2>/dev/null || true
+}
+
+graceful_kill_port() {
+  local port="$1"
+  local pids
+  pids=$(lsof -ti :"$port" 2>/dev/null || true)
+  if [[ -z "$pids" ]]; then return 0; fi
+  echo "$pids" | xargs kill -TERM 2>/dev/null || true
+  local waited=0
+  while [[ $waited -lt 50 ]]; do
+    if ! lsof -ti :"$port" >/dev/null 2>&1; then
+      return 0
+    fi
+    sleep 0.1
+    waited=$((waited + 1))
+  done
+  lsof -ti :"$port" 2>/dev/null | xargs kill -9 2>/dev/null || true
+}
+
+# ── Health polling: wait for a URL to respond 200 (or any 2xx/3xx) ────────────
+wait_for_health() {
+  local url="$1"
+  local name="$2"
+  local max_attempts="${3:-30}"
+  local attempt=0
+  while [ $attempt -lt $max_attempts ]; do
+    if curl -s --max-time 2 "$url" > /dev/null 2>&1; then
+      echo -e "  ${GREEN}✓${RESET} $name is up"
+      SVC_STATUS[$name]="up"
+      return 0
+    fi
+    attempt=$((attempt + 1))
+    sleep 1
+  done
+  echo -e "  ${RED}✗${RESET} $name failed to start (tried ${max_attempts}s)"
+  SVC_STATUS[$name]="down"
+  return 1
+}
+
+# ── Parse flags ───────────────────────────────────────────────────────────────
+START_DASH=1
+START_STUDIO=1
+START_BRIDGES=1
+for arg in "$@"; do
+  case "$arg" in
+    --no-dashboard) START_DASH=0 ;;
+    --no-studio)    START_STUDIO=0 ;;
+    --no-bridges)   START_BRIDGES=0 ;;
+  esac
+done
+
+# ══════════════════════════════════════════════════════════════════════════════
+# PHASE 1: Stop everything gracefully
+# ══════════════════════════════════════════════════════════════════════════════
+echo ""
+echo -e "${BOLD}Stopping existing crewswarm processes...${RESET}"
+
+# ── Kill by process name (SIGTERM → SIGKILL) ─────────────────────────────────
+# Use pattern without leading "node " so we catch all node binary paths
+graceful_kill_pattern "gateway-bridge.mjs"
+graceful_kill_pattern "opencrew-rt-daemon.mjs"
+graceful_kill_pattern "crew-lead.mjs"
+graceful_kill_pattern "scripts/dashboard.mjs"
+graceful_kill_pattern "scripts/mcp-server.mjs"
+graceful_kill_pattern "scripts/crew-scribe.mjs"
+graceful_kill_pattern "telegram-bridge.mjs"
+graceful_kill_pattern "whatsapp-bridge.mjs"
+graceful_kill_pattern "opencode serve"
+graceful_kill_pattern "pm-loop.mjs"
+graceful_kill_pattern "apps/vibe/server.mjs"
+graceful_kill_pattern "watch-server.mjs"
+graceful_kill_pattern "vite.*vibe"
+
+# ── Kill by port (catches anything that survived — e.g. launchd-managed dashboard)
+for port in 5010 4319 18889 4096 5020 3333 3334; do
+  graceful_kill_port "$port"
+done
+
+# ── Clean stale PID files so start-crew doesn't skip re-spawning
+find /tmp -maxdepth 1 -name "bridge-*.pid" -delete 2>/dev/null; true
+
+sleep 1
+
+# ── Confirm ports are clear ──────────────────────────────────────────────────
+for port in 5010 4319 18889 4096 5020 3333 3334; do
+  HELD=$(lsof -ti :$port 2>/dev/null | wc -l | tr -d ' ')
+  if [ "$HELD" -gt 0 ]; then
+    echo "  WARNING: port $port still held — force killing..."
+    lsof -ti :$port 2>/dev/null | xargs kill -9 2>/dev/null; true
+    sleep 1
+  fi
+done
+
+echo "  All processes stopped."
+
+# ══════════════════════════════════════════════════════════════════════════════
+# PHASE 2: Start services in order with health polling
+# ══════════════════════════════════════════════════════════════════════════════
+echo ""
+echo -e "${BOLD}Starting services...${RESET}"
+
+# ── 1. OpenCode (optional, non-critical) ─────────────────────────────────────
+echo ""
+echo "Starting optional OpenCode session server (port 4096)..."
+OPENCODE_BIN="$(command -v opencode 2>/dev/null)" || OPENCODE_BIN="/usr/local/bin/opencode"
+if [[ -x "$OPENCODE_BIN" ]]; then
+  nohup "$OPENCODE_BIN" serve --port 4096 --hostname 127.0.0.1 >> /tmp/opencode.log 2>&1 &
+  # Non-critical — don't block on health
+  SVC_STATUS["opencode"]="up"
+  echo "  Started (PID $!)"
+else
+  echo "  (opencode not found; skip)"
+  SVC_STATUS["opencode"]="skip"
+fi
+
+# ── 2. RT daemon (port 18889) ────────────────────────────────────────────────
+echo ""
+echo "Starting RT daemon (port 18889)..."
+nohup "$NODE" scripts/opencrew-rt-daemon.mjs >> /tmp/opencrew-rt-daemon.log 2>&1 &
+
+if ! wait_for_health "http://127.0.0.1:18889/status" "rt-bus" 15; then
+  echo "  Retrying RT daemon once..."
+  graceful_kill_pattern "opencrew-rt-daemon.mjs"
+  graceful_kill_port 18889
+  sleep 1
+  nohup "$NODE" scripts/opencrew-rt-daemon.mjs >> /tmp/opencrew-rt-daemon.log 2>&1 &
+  if ! wait_for_health "http://127.0.0.1:18889/status" "rt-bus" 15; then
+    echo "  Last RT log lines:"
+    tail -n 40 /tmp/opencrew-rt-daemon.log 2>/dev/null || true
+    echo ""
+    echo -e "  ${RED}RT bus failed — continuing with remaining services${RESET}"
+  fi
+fi
+
+# ── 3. Gateway bridges (agents) ──────────────────────────────────────────────
+echo ""
+echo "Starting gateway bridges (crew-main, crew-pm, crew-coder, etc.)..."
+"$NODE" scripts/start-crew.mjs --force
+SVC_STATUS["agents"]="up"
+sleep 1
+
+# ── 4. crew-lead (port 5010) — CRITICAL ─────────────────────────────────────
+echo ""
+echo "Starting crew-lead (port 5010)..."
+# crew-lead is spawned by start-crew.mjs above; this is a safety net only.
+if ! lsof -ti :5010 >/dev/null 2>&1; then
+  nohup "$NODE" crew-lead.mjs >> /tmp/crew-lead.log 2>&1 &
+fi
+wait_for_health "http://127.0.0.1:5010/health" "crew-lead" 30 || true
+
+# ── 5. Dashboard (port 4319) — CRITICAL ─────────────────────────────────────
+if [[ "$START_DASH" -eq 1 ]]; then
+  echo ""
+  echo "Starting dashboard (port 4319)..."
+  if launchctl list com.crewswarm.dashboard >/dev/null 2>&1; then
+    launchctl stop com.crewswarm.dashboard 2>/dev/null; sleep 1
+    launchctl start com.crewswarm.dashboard 2>/dev/null
+    echo "  (via launchd)"
+  else
+    nohup "$NODE" scripts/dashboard.mjs >> /tmp/dashboard.log 2>&1 &
+  fi
+  wait_for_health "http://127.0.0.1:4319/" "dashboard" 30 || true
+else
+  SVC_STATUS["dashboard"]="skip"
+fi
+
+# ── 6. Messaging bridges (Telegram + WhatsApp) ──────────────────────────────
+if [[ "$START_BRIDGES" -eq 1 ]]; then
+  echo ""
+  # Verify RT bus is still up before starting bridges
+  if [[ "${SVC_STATUS[rt-bus]:-down}" == "up" ]]; then
+    echo "Starting Telegram bridge..."
+    nohup "$NODE" "$REPO_DIR/telegram-bridge.mjs" >> /tmp/telegram-bridge.log 2>&1 &
+    SVC_STATUS["telegram"]="up"
+    echo "  PID: $!"
+
+    echo "Starting WhatsApp bridge..."
+    nohup "$NODE" "$REPO_DIR/whatsapp-bridge.mjs" >> /tmp/whatsapp-bridge.log 2>&1 &
+    SVC_STATUS["whatsapp"]="up"
+    echo "  PID: $!"
+  else
+    echo -e "  ${YELLOW}Skipping messaging bridges — RT bus is not up${RESET}"
+    SVC_STATUS["telegram"]="skip"
+    SVC_STATUS["whatsapp"]="skip"
+  fi
+fi
+
+# ── 7. MCP + OpenAI-compat server (port 5020) ───────────────────────────────
+echo ""
+echo "Starting MCP + OpenAI-compat server (port 5020)..."
+if ! lsof -ti :5020 >/dev/null 2>&1; then
+  nohup "$NODE" scripts/mcp-server.mjs >> /tmp/crewswarm-mcp.log 2>&1 &
+  wait_for_health "http://127.0.0.1:5020/health" "mcp-server" 15 || true
+else
+  echo "  (already running on :5020)"
+  SVC_STATUS["mcp-server"]="up"
+fi
+
+# ── 8. Vibe Studio + file watcher (ports 3333, 3334) ────────────────────────
+if [[ "$START_STUDIO" -eq 1 ]]; then
+  echo ""
+  echo "Starting Vibe (Studio) + file watcher (ports 3333, 3334)..."
+  if [[ ! -d "$REPO_DIR/apps/vibe/dist" ]]; then
+    echo "  Building Studio (first run)..."
+    (cd "$REPO_DIR" && npm run studio:build) >/dev/null 2>&1 || true
+  fi
+  sleep 1
+  nohup "$NODE" "$REPO_DIR/apps/vibe/server.mjs" >> /tmp/studio.log 2>&1 &
+  nohup env NODE_DISABLE_COMPILE_CACHE=1 "$NODE" "$REPO_DIR/apps/vibe/watch-server.mjs" >> /tmp/studio-watch.log 2>&1 &
+  wait_for_health "http://127.0.0.1:3333/" "vibe-studio" 25 || true
+  # Watch server doesn't have HTTP health — just check port
+  for i in $(seq 1 10); do
+    if lsof -ti :3334 >/dev/null 2>&1; then
+      echo -e "  ${GREEN}✓${RESET} file watcher is up"
+      SVC_STATUS["file-watcher"]="up"
+      break
+    fi
+    if [[ "$i" -eq 10 ]]; then
+      echo -e "  ${YELLOW}!${RESET} file watcher not detected on :3334"
+      SVC_STATUS["file-watcher"]="down"
+    fi
+    sleep 1
+  done
+else
+  SVC_STATUS["vibe-studio"]="skip"
+  SVC_STATUS["file-watcher"]="skip"
+fi
+
+# ══════════════════════════════════════════════════════════════════════════════
+# PHASE 3: Summary
+# ══════════════════════════════════════════════════════════════════════════════
+echo ""
+echo -e "${BOLD}━━━ Service Summary ━━━${RESET}"
+
+print_status() {
+  local name="$1"
+  local port="$2"
+  local status="${SVC_STATUS[$name]:-unknown}"
+  case "$status" in
+    up)   echo -e "  ${GREEN}✓${RESET} $name  :$port" ;;
+    down) echo -e "  ${RED}✗${RESET} $name  :$port" ;;
+    skip) echo -e "  ${YELLOW}-${RESET} $name  :$port  (skipped)" ;;
+    *)    echo -e "  ${YELLOW}?${RESET} $name  :$port  (unknown)" ;;
+  esac
+}
+
+print_status "rt-bus"       18889
+print_status "crew-lead"    5010
+print_status "dashboard"    4319
+print_status "agents"       "--"
+print_status "mcp-server"   5020
+print_status "vibe-studio"  3333
+print_status "file-watcher" 3334
+print_status "opencode"     4096
+if [[ "$START_BRIDGES" -eq 1 ]]; then
+  print_status "telegram"   "--"
+  print_status "whatsapp"   "--"
+fi
+
+echo ""
+echo "Logs: /tmp/opencode.log /tmp/opencrew-rt-daemon.log /tmp/crew-lead.log /tmp/dashboard.log /tmp/crewswarm-mcp.log /tmp/studio.log /tmp/studio-watch.log"
+
+# ── Run health-check if available ────────────────────────────────────────────
+if [[ -f "$REPO_DIR/scripts/health-check.mjs" ]]; then
+  echo ""
+  echo -e "${BOLD}Running health check...${RESET}"
+  "$NODE" "$REPO_DIR/scripts/health-check.mjs" --quiet 2>/dev/null || true
+fi
+
+# ── Exit code: 0 if critical services are up, 1 otherwise ────────────────────
+CREW_LEAD_OK="${SVC_STATUS[crew-lead]:-down}"
+DASHBOARD_OK="${SVC_STATUS[dashboard]:-down}"
+
+if [[ "$CREW_LEAD_OK" == "up" ]] && { [[ "$DASHBOARD_OK" == "up" ]] || [[ "$START_DASH" -eq 0 ]]; }; then
+  echo ""
+  echo -e "${GREEN}${BOLD}All critical services are up.${RESET}"
+  exit 0
+else
+  echo ""
+  echo -e "${RED}${BOLD}One or more critical services failed to start.${RESET}"
+  [[ "$CREW_LEAD_OK" != "up" ]] && echo -e "  ${RED}crew-lead is not responding — check /tmp/crew-lead.log${RESET}"
+  [[ "$DASHBOARD_OK" != "up" && "$START_DASH" -eq 1 ]] && echo -e "  ${RED}dashboard is not responding — check /tmp/dashboard.log${RESET}"
+  exit 1
+fi

package/scripts/restart-crew-lead.sh

@@ -0,0 +1,98 @@
+#!/usr/bin/env bash
+# Restart crew-lead using PID file for reliable process management
+# This prevents accidentally killing the dashboard when restarting crew-lead
+
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+REPO_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
+CREWSWARM_DIR="${CREWSWARM_DIR:-${OPENCLAW_DIR:-$REPO_ROOT}}"
+CREW_LEAD_SCRIPT="$CREWSWARM_DIR/crew-lead.mjs"
+LOG_FILE="/tmp/crew-lead.log"
+PID_FILE="$HOME/.crewswarm/logs/crew-lead.pid"
+PATH="/usr/local/bin:/opt/homebrew/bin:/usr/bin:/bin:${PATH:-}"
+export PATH
+LAUNCH_LABEL="com.crewswarm.crew-lead"
+LAUNCH_PLIST="$HOME/Library/LaunchAgents/${LAUNCH_LABEL}.plist"
+RESOLVE_NODE_BIN="$CREWSWARM_DIR/scripts/resolve-node-bin.sh"
+NODE_BIN="${NODE:-}"
+if [[ -z "$NODE_BIN" ]]; then
+  if [[ -x "$RESOLVE_NODE_BIN" ]]; then
+    NODE_BIN="$("$RESOLVE_NODE_BIN")"
+  elif command -v node >/dev/null 2>&1; then
+    NODE_BIN="$(command -v node)"
+  elif [[ -x /usr/local/bin/node ]]; then
+    NODE_BIN="/usr/local/bin/node"
+  elif [[ -x /opt/homebrew/bin/node ]]; then
+    NODE_BIN="/opt/homebrew/bin/node"
+  else
+    echo "❌ node not found in PATH"
+    exit 1
+  fi
+fi
+
+# Stop every way crew-lead might be running and free :5010 (launchd alone does not
+# kill a nohup/manual node crew-lead.mjs, which causes "Port 5010 in use" on kickstart).
+stop_crew_lead_processes() {
+  echo "  → Stopping existing crew-lead processes..."
+  if [ -f "$PID_FILE" ]; then
+    PID=$(cat "$PID_FILE" 2>/dev/null || echo "")
+    if [ -n "$PID" ] && kill -0 "$PID" 2>/dev/null; then
+      echo "  → Found crew-lead PID $PID from PID file"
+      kill -9 "$PID" 2>/dev/null || true
+    fi
+    echo "" > "$PID_FILE" 2>/dev/null || true
+  fi
+  pkill -9 -f "crew-lead.mjs" 2>/dev/null || true
+  if lsof -i :5010 -sTCP:LISTEN -t &>/dev/null; then
+    echo "  → Releasing port 5010 (listener still present)..."
+    lsof -ti :5010 -sTCP:LISTEN 2>/dev/null | xargs kill -9 2>/dev/null || true
+  fi
+  local n=0
+  while lsof -i :5010 -sTCP:LISTEN -t &>/dev/null && [ "$n" -lt 60 ]; do
+    sleep 0.25
+    n=$((n + 1))
+  done
+}
+
+echo "🔄 Restarting crew-lead..."
+
+if launchctl list "$LAUNCH_LABEL" >/dev/null 2>&1 || [[ -f "$LAUNCH_PLIST" ]]; then
+  echo "  → Using launchd agent: $LAUNCH_LABEL"
+  launchctl bootout "gui/$(id -u)" "$LAUNCH_PLIST" 2>/dev/null || true
+  stop_crew_lead_processes
+  launchctl bootstrap "gui/$(id -u)" "$LAUNCH_PLIST"
+  launchctl kickstart -k "gui/$(id -u)/$LAUNCH_LABEL"
+  sleep 2
+  if lsof -i :5010 -sTCP:LISTEN -t &>/dev/null; then
+    echo "✅ crew-lead is running at http://127.0.0.1:5010"
+    echo "   Log: $LOG_FILE"
+    echo "   LaunchAgent: $LAUNCH_PLIST"
+    exit 0
+  fi
+  echo "❌ crew-lead launchd start failed. Check log:"
+  echo "   tail -20 $LOG_FILE"
+  exit 1
+fi
+
+stop_crew_lead_processes
+
+# Step 4: Start fresh crew-lead
+echo "  → Starting crew-lead at $CREW_LEAD_SCRIPT..."
+cd "$CREWSWARM_DIR"
+NODE_DISABLE_COMPILE_CACHE=1 nohup "$NODE_BIN" "$CREW_LEAD_SCRIPT" >> "$LOG_FILE" 2>&1 &
+
+# Step 5: Wait for startup and verify
+echo "  → Verifying startup..."
+sleep 3
+
+if lsof -i :5010 -sTCP:LISTEN -t &>/dev/null; then
+  echo "✅ crew-lead is running at http://127.0.0.1:5010"
+  echo "   Log: $LOG_FILE"
+  echo "   PID file: $PID_FILE"
+  exit 0
+else
+  echo "❌ crew-lead failed to start. Check log:"
+  echo "   tail -20 $LOG_FILE"
+  exit 1
+fi

package/scripts/restart-dashboard.sh

@@ -0,0 +1,104 @@
+#!/usr/bin/env bash
+# Direct dashboard restart script — for agents and CLI use
+# Bypasses the dashboard's API self-restart prohibition
+#
+# Usage: bash scripts/restart-dashboard.sh
+#        npm run restart-dashboard
+
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+REPO_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
+CREWSWARM_DIR="${CREWSWARM_DIR:-${OPENCLAW_DIR:-$REPO_ROOT}}"
+DASHBOARD_SCRIPT="$CREWSWARM_DIR/scripts/dashboard.mjs"
+LOG_FILE="${CREWSWARM_DASH_LOG:-/tmp/dashboard.log}"
+HEALTH_URL="${CREWSWARM_DASH_HEALTH_URL:-http://127.0.0.1:4319/api/health}"
+NODE_BIN="${NODE:-$("$CREWSWARM_DIR/scripts/resolve-node-bin.sh")}"
+LAUNCH_LABEL="com.crewswarm.dashboard"
+LAUNCH_PLIST="$HOME/Library/LaunchAgents/${LAUNCH_LABEL}.plist"
+
+wait_for_dashboard() {
+  local pid="$1"
+  local attempts="${2:-15}"
+
+  for ((i=1; i<=attempts; i++)); do
+    if ! kill -0 "$pid" 2>/dev/null; then
+      echo "  → Dashboard process exited before becoming healthy"
+      return 1
+    fi
+
+    if curl -sSf -m 2 "$HEALTH_URL" >/dev/null 2>&1; then
+      return 0
+    fi
+
+    sleep 1
+  done
+
+  echo "  → Dashboard did not pass health check at $HEALTH_URL"
+  return 1
+}
+
+start_dashboard() {
+  export NODE_DISABLE_COMPILE_CACHE=1
+  if command -v setsid >/dev/null 2>&1; then
+    setsid "$NODE_BIN" "$DASHBOARD_SCRIPT" >> "$LOG_FILE" 2>&1 < /dev/null &
+  else
+    nohup "$NODE_BIN" "$DASHBOARD_SCRIPT" >> "$LOG_FILE" 2>&1 < /dev/null &
+  fi
+  echo $!
+}
+
+echo "🔄 Restarting dashboard..."
+
+if launchctl list "$LAUNCH_LABEL" >/dev/null 2>&1 || [[ -f "$LAUNCH_PLIST" ]]; then
+  echo "  → Using launchd agent: $LAUNCH_LABEL"
+  launchctl bootout "gui/$(id -u)" "$LAUNCH_PLIST" 2>/dev/null || true
+  # Kill only the actual node dashboard process, not shell wrappers that mention it
+pgrep -f 'node.*scripts/dashboard\.mjs' | xargs kill -9 2>/dev/null || true
+  lsof -ti :4319 2>/dev/null | xargs kill -9 2>/dev/null || true
+  sleep 2
+  launchctl bootstrap "gui/$(id -u)" "$LAUNCH_PLIST"
+  launchctl kickstart -k "gui/$(id -u)/$LAUNCH_LABEL"
+  for _ in $(seq 1 20); do
+    if curl -sSf -m 2 "$HEALTH_URL" >/dev/null 2>&1; then
+      echo "✅ Dashboard is running at http://127.0.0.1:4319"
+      echo "   Log: $LOG_FILE"
+      echo "   LaunchAgent: $LAUNCH_PLIST"
+      exit 0
+    fi
+    sleep 1
+  done
+  echo "❌ Dashboard launchd start failed. Check log:"
+  tail -20 "$LOG_FILE" || true
+  exit 1
+fi
+
+echo "  → Stopping existing dashboard processes..."
+# Kill only the actual node dashboard process, not shell wrappers that mention it
+pgrep -f 'node.*scripts/dashboard\.mjs' | xargs kill -9 2>/dev/null || true
+
+echo "  → Waiting for port 4319 to release..."
+sleep 2
+
+# Step 3: Start fresh dashboard
+echo "  → Starting dashboard at $DASHBOARD_SCRIPT..."
+DASH_PID="$(start_dashboard)"
+
+# Step 4: Wait for startup and verify
+echo "  → Verifying startup..."
+if wait_for_dashboard "$DASH_PID"; then
+  sleep 3
+  if ! kill -0 "$DASH_PID" 2>/dev/null; then
+    echo "❌ Dashboard exited shortly after startup. Check log:"
+    tail -20 "$LOG_FILE" || true
+    exit 1
+  fi
+  echo "✅ Dashboard is running at http://127.0.0.1:4319"
+  echo "   PID: $DASH_PID"
+  echo "   Log: $LOG_FILE"
+  exit 0
+else
+  echo "❌ Dashboard failed to start. Check log:"
+  tail -20 "$LOG_FILE" || true
+  exit 1
+fi