To get a notification when a Mac process exits or uses too many resources, add it to ProcXray Watch and enable the alerts you need. Choose a single process session, an app, or an executable, then configure lifecycle notifications or sustained CPU, memory, and GPU thresholds.
This guide covers the Watch controls in ProcXray 1.3.5. You need macOS 14 or later and ProcXray running. If you have not installed it yet, start with the setup guide or download ProcXray.
Choose what you want to watch
Find the process in ProcXray, check its name and executable path, then right-click and select Add to Watch…. In Watch Target, choose the matching rule that fits your task:
| Match | What it follows | Use it for |
|---|---|---|
| PID — this session | This process instance, identified by PID and start time. Removed when ProcXray quits. | Watching one running job until it exits. |
| App — persistent | The app’s main process, matched by bundle identity across relaunches. | Following an app you quit and reopen. |
| Executable — persistent | Instances launched from the same executable path. | Following a command-line tool or service across restarts. |
App and executable choices depend on whether ProcXray can identify that target. An App watch does not add up the app and all its helper processes. An Executable watch evaluates each matching instance separately; it is not a combined resource budget.
Persistent means the configuration is saved across ProcXray restarts. Monitoring stops when you quit ProcXray.
Get notified when a process exits or starts
- Under Lifecycle Notifications, enable Notify when the process exits.
- For an App or Executable target, optionally enable Notify when the process starts. Start notifications are unavailable for a PID watch because that session already exists.
- Click Add to Watch and allow macOS notifications when prompted.
- Keep ProcXray running. Check the Watch view for the target and its events.
An exit event says the process ended; it does not by itself distinguish a crash from a normal exit or prove a job succeeded. Check the application’s own output when you need that answer.
Try a simple lifecycle check
Open TextEdit and add its main process as App — persistent, with both start and exit notifications enabled. After saving any open documents, quit TextEdit normally. Leave ProcXray running and let the Watch view observe the exit before reopening TextEdit. On reopening, wait for it to observe the new process.
You should see the exit and subsequent start in Watch; macOS displays notifications when permitted by your settings. The first observation establishes a baseline, so adding an already running app does not itself produce a start notification. This checks lifecycle alerts without creating artificial CPU load.
Set a CPU, memory, or GPU alert
Under Threshold Notifications, enable the metric you want to monitor and enter a limit. For example, use a 100% CPU threshold, 10 seconds for Must remain above threshold for, and 10 minutes for Notification cooldown. These are starting values, not a recommendation for every workload; compare them with the process’s usual readings.
The same panel supports a memory limit in GB and a GPU limit in %. Start with one metric so you can see which condition generated an event. GPU readings depend on what is available for the target.
Click Add to Watch, or Save when editing an existing target. To revise a saved rule, open the target’s menu in Watch and choose Edit Notifications….
Duration filters brief spikes
The metric must remain at or above the configured limit across observations for the duration you set. A brief spike is not enough. Notification timing depends on sampling; a ten-second duration is not a guarantee that a banner appears exactly ten seconds after the underlying workload changes.
Cooldown limits repeat alerts
Cooldown is not a reminder interval. If a process stays above the threshold, ProcXray does not send a new alert every ten minutes. The condition must recover below 90% of the threshold before it can trigger again, and the cooldown must also have elapsed. Recovery is recorded as an event without a separate system notification.
For a 100% CPU threshold, that means falling below 90% before another sustained high-CPU episode can generate a fresh alert. This prevents repeated alerts as readings fluctuate close to the limit.
Check what happened after an alert
Open the Watch view to inspect the target and recorded events. Persistent targets can show Waiting when no matching process is running; a completed session watch can show Exited. Select the process to inspect its details and recorded resource history while it is available.
The menu bar dashboard also shows up to three recent Watch events from the last 24 hours. Choose Show All to open the Watch list. For a memory investigation, continue with the memory leak guide; a threshold crossing alone does not establish a leak.
If an expected notification does not appear
- Check the target. An App watch follows the main process, not every helper. An Executable watch matches its path, so a different installed copy may not match.
- Check the rule. A PID watch cannot report a future launch. A resource spike may be shorter than the duration, or a previous alert may still be active or in cooldown.
- Check macOS permissions. Open System Settings → Notifications, find ProcXray, and allow notifications. Display and lock-screen settings can affect where they appear. See Apple’s Notifications settings guide.
- Check that ProcXray is running. Hiding the main window is supported; quitting the app stops monitoring. Background process checks use a reduced cadence.
- Allow for sampling. A process that starts and exits between samples can be missed. Watch is not a complete process-execution trace.
Is this different from Activity Monitor?
Activity Monitor is useful for an immediate resource check. ProcXray Watch adds configurable target matching, lifecycle alerts, and sustained resource thresholds. The Activity Monitor comparison explains the other differences, including what the built-in tool already does well.