Every time an automation is triggered, it performs a run. Show runs lists each run, so you can check that an automation works the way you expect, see why a run was skipped, and fix whatever made one fail. Agent triggers create runs too, and you can check them the same way.
For example, if an automation didn’t create a task you expected, Show runs tells you why.
👉 To learn how automation actions count toward your monthly limit, see Monthly Automation Actions.
Where to Find Runs
Go to Settings > Automations, open the three-dot menu next to an automation, and select Show runs.
📌 For automations linked to a form, you can also open Show runs from the form's Automations tab.
For an Agent's triggers, go to the Agent's Info tab, click the three-dot menu next to the trigger, and select Show runs. The runs open in a Trigger runs window and use the same statuses described below.
Run and Step Statuses
At the top of the page, you'll see how many times the automation completed this month and how many of your monthly actions it used. Below that, each run appears as its own card, showing when it ran and, if a person started it, who triggered it. The list updates on its own when a run finishes.
Each card shows how the run ended. Expand a card to see how each step went:
Icon | On a run | On a step |
Green check | 2 / 2 completed: the run finished without errors. The numbers count the run’s steps, and each action in a Find object step counts once for every object it ran on. | The step succeeded. |
Clock | Running: the run is still in progress or waiting for a result. | The step is waiting for a result. |
Amber minus | Run skipped: condition not met: the Check if conditions weren't met, so no actions ran. This isn't an error, and it doesn't use any actions. | The step had nothing to do, because a condition wasn't met or a Find object step found no objects. |
Red warning | Run failed: [reason]: something went wrong, and the reason tells you what. | The step failed. The run shows why. |
Grey minus, greyed-out name | Not used on runs. | The run never reached this step. |
When a step fails, the steps after it don't run. The exception is an action that couldn't run at all, for example, because Slack isn't connected. In that case, the rest of the run carries on.
📌 Send Slack Message, Send Teams Message, and Send HTTP Request wait for a reply from the other service, so the run shows Running until it arrives. The steps after them don't wait, so they still run even if the reply is a failure. If an HTTP request doesn't get a successful reply, Productive retries it for up to about 12 hours before marking it as failed.
In a Find object step, the badge shows how many objects were found. Each action below it shows how many objects succeeded, failed, or are still waiting. If an action fails for one object, the actions after it don't run for that object, but the next object starts fresh. A yellow minus means the action worked for some objects and failed for others.
📌 Runs from before November 2026 only show as completed or failed. Older runs whose conditions weren't met still show as failed.
Why a Run Failed
When a run fails, it shows Run failed: followed by the reason. Most reasons tell you exactly what to fix, for example, Slack isn't connected or invalid URL.
Here’s what the others mean and how to fix them:
Reason | What to do |
The action couldn't be completed | An action got data it couldn't use, for example, an assignee who isn't on the project, a missing project, task list, or status, or an item that was deleted. This reason also shows when an action inside a Find object step fails. Check the action's fields and the people and items it refers to. |
Too many objects found | Running the Find object step's actions on every object found would go over your monthly action limit, so none of them ran, and neither did the steps after it. Narrow the filters, use fewer actions in the step, wait for your actions to reset next month, or upgrade your plan. |
The organization isn't verified | Send Email and Send HTTP Request can't run until Productive verifies your organization. The rest of the run carries on. Contact support. |
Unsupported sender email | Productive can't send from this address. Use noreply@outbox.productive.io or a verified custom domain. See Customizing Your Sender Email Address. |
No permission to post in the Slack channel | Slack's posting restrictions for that channel or workspace blocked the message. Ask a Slack admin to allow posting there, or choose another channel. |
The Slack account is inactive | The Slack account behind your connection is no longer active, so Productive turns off the Slack integration. Reconnect Slack with an active account. |
👉 To reconnect Slack or invite Productive to a channel, see Organization-Wide Integration for Slack. For Teams, see Organization-Wide Integration for Microsoft Teams.
Why an Agent Run Failed
Agent triggers start the Agent with a Start agent action step. The step waits until the Agent finishes its work, so the run shows Running in the meantime. A green check means the Agent finished the job, not just that it started.
When a run fails, click Show details next to the Start agent action step. It opens a read-only view of the Agent's chat for that run, so you can see what the Agent did and where it stopped.
Reason | What to do |
AI agent could not complete the step | The work started but didn’t finish. The Agent couldn’t do the task, someone stopped it, or the AI model ran into an error. Click Show details to see what happened, then adjust the trigger’s instructions or the Agent’s permissions and connectors. |
AI usage limit reached | Your organization has used all of this month's AI usage. New runs work again once usage resets, or you add more. |
The agent is deactivated | The Agent was deactivated while the run was in progress. Deactivating an Agent turns off all its triggers, and reactivating it doesn’t turn them back on. Reactivate the Agent, then turn its triggers back on. |
Too many agent runs in a short time | The trigger started the Agent far too many times in a short period. That usually means it keeps triggering itself. Productive turned the trigger off and emailed the person responsible. Adjust the trigger or its conditions, then turn it back on. |
The agent couldn't start | The AI service didn’t respond, even after several retries. Try again later with Run now, and contact support if it keeps happening. |
The Slack sender has no access to the agent | The person who posted in Slack can’t use the Agent. Either it isn’t shared with them, or their Slack account isn’t connected to Productive. Share the Agent with them, or ask them to connect their personal Slack integration. |
The agent can't be started this way | In your Sandbox, Agents don’t start from scheduled triggers, so the run fails at this step. Event-based triggers and Run now still start the Agent there, so use Run now to test a scheduled trigger. |



