Accessing Execution Logs:
From the Agents Table:
- Navigate to the Agents page
- Locate your agent in the table
- Click the "Audit trail" button in the Side Panel
- The Execution Logs page opens
Page Layout:
Header:
- Agent name displayed as page title
- Breadcrumb navigation: "Agents / {Agent name} / Audit trail"
Action Buttons:
- Export: Download logs as Excel or CSV
- Customize columns: Show/hide table columns
- Filter: Narrow displayed runs by criteria
Execution Logs Table:
The logs table displays all historical executions with the following columns:
Run Dates:
- Format: "Month nth Day, Year HH:MM AM/PM"
- Example: "September 1, 2025 02:00 PM"
- Most recent runs appear first (descending order)
- Clickable to view document-level details
Status: Status badges indicate run state:
- "Running" (orange/yellow badge): Execution currently running
- "Completed" (green badge): Execution finished successfully
- "Failed" (red badge): Execution failed
Scope:
- Number of document that will be processed
Errors:
-
Number of errors registered during the run
Clicking on a Run:
- Click anywhere on a run row
- Navigate to Run Details page
- View document-by-document extraction results
- See success/failure for each individual contract
Important Notes:
- Execution logs retain indefinitely for audit purposes
- Progress numbers update in real-time during active executions
- Only runs for documents you have access to appear in logs
- Deleting an agent removes all associated execution logs
- Failed extractions don't create separate failed document counts (shown in document-level results instead)
- Each manual run creates a separate log entry
- Automated triggers (future) will also create log entries
Best Practices:
Regular Monitoring:
- Check logs weekly for active agents
- Review progress percentages to spot accuracy issues
- Investigate runs with unusually low success rates
- Compare execution frequency against expectations
Performance Analysis:
- Track success rates over time
- Identify when extraction accuracy degrades
- Correlate execution results with agent configuration changes
- Use exported data for trend analysis
Troubleshooting Agents:
- Identify problematic run in logs table
- Note low progress percentage or failed status
- Review individual document failures
- Investigate common failure patterns
- Adjust agent configuration based on findings
- Re-run manually to validate improvements