fix: improve --town mode robustness from code review

- resolveBeadsRedirect now verifies target exists before returning
- Added failure tracking to runTownActivityFollow (warns on rig disconnect)
- Created fetchTownMutationsWithStatus for tracking daemon availability
- Shows reconnection message when rigs come back online

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

cmd/bd/activity.go

@@ -495,7 +495,8 @@ func discoverRigDaemons() []rigDaemon {
 	return daemons
 }
 
-// resolveBeadsRedirect follows a redirect file if present
+// resolveBeadsRedirect follows a redirect file if present.
+// Similar to routing.resolveRedirect but simplified for activity use.
 func resolveBeadsRedirect(beadsDir string) string {
 	redirectFile := filepath.Join(beadsDir, "redirect")
 	data, err := os.ReadFile(redirectFile)
@@ -513,12 +514,26 @@ func resolveBeadsRedirect(beadsDir string) string {
 		redirectPath = filepath.Join(beadsDir, redirectPath)
 	}
 
-	return filepath.Clean(redirectPath)
+	redirectPath = filepath.Clean(redirectPath)
+
+	// Verify target exists before returning
+	if info, err := os.Stat(redirectPath); err == nil && info.IsDir() {
+		return redirectPath
+	}
+
+	return beadsDir // Fallback to original if redirect target invalid
 }
 
 // fetchTownMutations retrieves mutations from all rig daemons
 func fetchTownMutations(daemons []rigDaemon, since time.Time) []rpc.MutationEvent {
+	events, _ := fetchTownMutationsWithStatus(daemons, since)
+	return events
+}
+
+// fetchTownMutationsWithStatus retrieves mutations and returns count of responding daemons
+func fetchTownMutationsWithStatus(daemons []rigDaemon, since time.Time) ([]rpc.MutationEvent, int) {
 	var allEvents []rpc.MutationEvent
+	activeCount := 0
 
 	var sinceMillis int64
 	if !since.IsZero() {
@@ -535,6 +550,8 @@ func fetchTownMutations(daemons []rigDaemon, since time.Time) []rpc.MutationEven
 			continue
 		}
 
+		activeCount++
+
 		var mutations []rpc.MutationEvent
 		if err := json.Unmarshal(resp.Data, &mutations); err != nil {
 			continue
@@ -548,7 +565,7 @@ func fetchTownMutations(daemons []rigDaemon, since time.Time) []rpc.MutationEven
 		return allEvents[i].Timestamp.Before(allEvents[j].Timestamp)
 	})
 
-	return allEvents
+	return allEvents, activeCount
 }
 
 // runTownActivityOnce fetches and displays events from all rigs once
@@ -649,12 +666,43 @@ func runTownActivityFollow(sinceTime time.Time) {
 	ticker := time.NewTicker(activityInterval)
 	defer ticker.Stop()
 
+	// Track failures for warning messages
+	consecutiveFailures := 0
+	const failureWarningThreshold = 5
+	lastWarningTime := time.Time{}
+	lastActiveCount := activeCount
+
 	for {
 		select {
 		case <-rootCtx.Done():
 			return
 		case <-ticker.C:
-			newEvents := fetchTownMutations(daemons, lastPoll)
+			newEvents, currentActive := fetchTownMutationsWithStatus(daemons, lastPoll)
+
+			// Track daemon availability changes
+			if currentActive < lastActiveCount {
+				consecutiveFailures++
+				if consecutiveFailures >= failureWarningThreshold {
+					if time.Since(lastWarningTime) >= 30*time.Second {
+						if !jsonOutput {
+							timestamp := time.Now().Format("15:04:05")
+							fmt.Fprintf(os.Stderr, "[%s] %s some rigs unreachable (%d/%d active)\n",
+								timestamp, ui.RenderWarn("!"), currentActive, len(daemons))
+						}
+						lastWarningTime = time.Now()
+					}
+				}
+			} else if currentActive > lastActiveCount {
+				// Daemon came back
+				if !jsonOutput {
+					timestamp := time.Now().Format("15:04:05")
+					fmt.Fprintf(os.Stderr, "[%s] %s rig reconnected (%d/%d active)\n",
+						timestamp, ui.RenderPass("✓"), currentActive, len(daemons))
+				}
+				consecutiveFailures = 0
+			}
+			lastActiveCount = currentActive
+
 			newEvents = filterEvents(newEvents)
 
 			for _, e := range newEvents {