Concepts
Leaks
A leak is a detected pattern of sustained memory growth in one of your processes.
How leaks are created
When the agent detects that heap usage has grown more than 15% across its rolling window, it flags a leak. The leak includes:
- Severity (low, medium, high, critical)
- Confidence percentage
- Growth rate in MB per minute
- Source location (file, function, line)
- Code snippet around the leak line
- AI-generated analysis and fix recommendation
Leak states
- Active — the leak has been detected and needs attention
- Resolved — you've confirmed the leak is fixed
- Ignored — you've marked this as a false positive or acceptable behavior
Duplicate prevention
If the same leak is detected multiple times in a row, Rytena updates the existing active leak instead of creating a new one. This prevents alert spam.
AI analysis
Every leak with source code triggers automatic AI analysis. The AI reviews the code, identifies why memory is being retained, and generates a specific fix.
AI analysis is triggered automatically for high and critical leaks. For lower severity, you can manually trigger it from the leak detail page.