Skip to main content

Monitoring Agentic Flow Executions in ConvyAI

Overview​

The Monitoring page shows the execution history of all Agentic Flows in the tenant. Each row represents a single run with its outcome and timing metadata. Use this page to analyze process performance, verify past executions and investigate errors.

The Monitoring page is accessible to both agenticorchestrator_admin and agenticorchestrator_user roles.

Run list​

Each row in the execution list shows:

FieldDescription
Flow nameThe name of the Agentic Flow that generated the run.
Execution IDUnique identifier of the run instance.
Final stateThe technical outcome of the run: end, stop, or error.
ResultThe semantic value set by the Result node, if configured.
Start timeTimestamp of when the run began.
End timeTimestamp of when the run concluded.

The list is paginated with 10 results per page by default. An inline search field is available for quick lookup by execution ID or flow name.

Final states​

StateMeaning
endThe run completed and reached a Result node.
stopThe run was manually stopped before completion.
errorThe run terminated due to an unhandled node error.

Note that end indicates technical completion - the semantic outcome (success, failure, or a custom label) is expressed by the Result field set on the Result node.

Filtering​

Use the filter toolbar to narrow the execution list. Filters combine: each one you add narrows the results.

FilterOptions
Filter period byThe date the time range applies to: Start time (default) or End time.
From / ToStart and end of the range, with date and time. The shortcuts fill both fields for you, and you can still edit them.
Execution IDExact match on the run identifier.
Flow nameDrop-down with a search field. It lists all the flows of the tenant, including renamed or deleted flows that still have executions.
StatusDrop-down with the available states. You can select more than one.
Result valueFree text: lists the executions whose result contains the text, ignoring upper and lower case. Executions with no result are excluded.

Click Save filters to apply the filters and Clear filters to remove them and show the full list. The inline search keeps working together with the filters. Filters you have set are visually distinct from empty ones. If From is later than To, an error appears and the filter is not applied.

The Period field starts on Last 7 days. The maximum range you can search is defined by the agenticOrchestrator.monitoring.days_of_report_searchability setting.

Filter period by​

  • Start time applies the range to the date the executions started. The list also includes executions still open.
  • End time applies the range to the date the executions ended. Executions still running or waiting are not listed, so combining End time with the running or waiting status returns an empty list.

Use End time to list the executions counted by the Dashboard, which counts by end date. Clear filters brings the selector back to Start time.

Open Monitoring from the Dashboard​

When you open Monitoring from a Dashboard or Flow Report card, the filters are already set and the list is already filtered. You can edit the filters, and Clear filters removes them.

The list refreshes every 10 seconds and keeps the filters you set. When you open a run and come back to the list, the filters are cleared.

Run detail​

Clicking a row in the list opens the run detail view, which shows the execution graph with the status of each node, timing per step and any error details.

Attaching to a run in live debug​

Rows for runs that have not yet reached a final state expose an action to open the run directly in live debug mode on the Agentic Designer canvas, attached to the execution as it happens. See Debug Executions for what you can see and do once attached.