> ## Documentation Index
> Fetch the complete documentation index at: https://squad.so/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# See what your agents did: Runs, Health, and Live Feed

> Find a run and read its detail, check today's run problems on Health, and follow your squad's activity in the Live Feed.

A run is one execution by one agent: a scheduled task, a chat reply, an email it handled, a research call, or an action in a connected app. Three screens show what your agents did:

| Screen        | Use it to                                                 | How to open it                                                                 |
| ------------- | --------------------------------------------------------- | ------------------------------------------------------------------------------ |
| **Runs**      | Find any run and read its full record.                    | **Settings** > **Runs**, or press R.                                           |
| **Health**    | See today's failed runs and provider limits in one place. | Click a link that ends with **details in Health**.                             |
| **Live Feed** | Follow what your squad does as it happens.                | Press F, or click **Open live feed** on **Home**, or **View all** in the bell. |

## Find a run

<Steps>
  <Step title="Open Runs">
    Open **Runs**.
  </Step>

  <Step title="Pick a view">
    Pick a view: **Day** or **Schedule**. **Schedule** is the default. It shows **UPCOMING** scheduled runs and the runs of each day.
  </Step>

  <Step title="Set the filters">
    In the **FILTERS** panel, choose what to show:

    * **AGENTS**: turn agents on or off.
    * **ACTIVITY**: **Scheduled runs**, **Manual runs**, **Chat**, **Emails**, **Research**, **Integrations**, **Tool calls**.
    * **SCHEDULES**: **All schedules** or one schedule.
  </Step>

  <Step title="Move through the days">
    Use **Previous**, **Next**, and **Today** to move through the days. The list shows three days at a time. Click **Load older runs** to go back further.
  </Step>
</Steps>

Runs shows the last 30 days. The header shows the number of runs, how many are active, and how many had errors. **SUMMARY** shows **Total runs**, **Success rate**, and **Avg duration**.

<Warning>The **Week** and **Month** views show fixed sample dates today and do not match the real calendar. Use **Day** or **Schedule**.</Warning>

Each row shows the time, the agent, the trigger, the duration, the skills used, the model, and a status: **Completed**, **Error**, **Timeout**, or **Running**. Research rows also show the credits used. See [What Squad costs](/docs/costs).

## Read a run's detail

Click a row. The detail panel shows:

* **SUMMARY** and, for research runs, **PARAMETERS** and **RESULT PREVIEW**
* **WHAT IT TOUCHED**, and **REQUEST** and **RESPONSE** where they apply
* **USAGE**: **Input tokens**, **Output tokens**, **Cache read tokens**, and **Cache write tokens**
* **SKILLS USED**
* **View related mission** when the run belongs to a mission
* **Technical details**, with the **Model**, **Provider**, **Reasoning** level, and **Channel**.

When a run did not complete, the panel shows one of these labels:

| Label                | What it means                                                                                                         | What to do                                                                                           |
| -------------------- | --------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------- |
| **Interrupted**      | A restart or routine maintenance stopped the run. Squad treats this as routine and does not count it as an error.     | Check that the work finished. If it did not, send the request again.                                 |
| **Presumed stopped** | The run stopped sending events and no end signal arrived. If the run is still working, new activity corrects the row. | Wait. If nothing changes, send the request again.                                                    |
| **Superseded**       | A newer run started in the same conversation and replaced this one.                                                   | Nothing. The newer run carries on.                                                                   |
| **Timed out**        | The run went past its time limit.                                                                                     | Click **Raise the timeout →** to open **Agent Configuration**, and raise the timeout.                |
| **Run failed**       | The run ended with an error. The error text shows below the label.                                                    | Read the error and **Technical details**. See [Fix an agent that is stuck or failing](/docs/agent-stuck). |

## Check today's problems on Health

The **Health** screen has no menu entry. You open it from these one-line links:

| Where              | Link text                                                                          | What it means                                    |
| ------------------ | ---------------------------------------------------------------------------------- | ------------------------------------------------ |
| **Home**           | The number of runs, then **today may need a look**, then **details in Health**     | Runs failed today.                               |
| **Missions** board | The agent's name, then **hit a snag today**, then **details in Health**            | One or more agents had failed runs today.        |
| Agent profile      | The number of runs, then **today may need a look**, then **details in Health**     | This agent had failed runs today.                |
| Agent profile      | **Routine interruptions today, resumed automatically**, then **details in Health** | This agent had only routine interruptions today. |

The **Health** screen shows today's runs only. Its sections are:

* **All clear**: "No failed runs today, and no provider limits in effect."
* **PROVIDER STATUS**: an AI provider is rate-limiting your agents. See [Choose models and manage AI accounts](/docs/ai-accounts#troubleshooting).
* **NEEDS A LOOK**: each failed run, with the agent, the trigger, the time, the error, **Technical details**, and **View in Runs →**.
* **ROUTINE INTERRUPTIONS**: runs that maintenance interrupted. Each row is tagged **resumed**.

The **Health** screen says that interrupted runs pick back up automatically. Check the related mission or chat to confirm that the work finished.

**Back** at the top of **Health** returns you to **Home**.

## Know when Squad resumes a failed scheduled run

When a scheduled run fails because the AI provider had a temporary problem, Squad wakes the agent once more, two minutes later. The agent is told to check where it left off, finish the remaining work, and not redo steps that already succeeded.

| Squad resumes the run when                                                  | Squad does not resume the run when                                                                       |
| --------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------- |
| The provider reported a server error, an overload, or a dropped connection. | The error is a sign-in problem, an invalid key, a usage limit, a quota, billing, or "All models failed". |
| The run was a scheduled run.                                                | The run was a chat reply or another kind of run.                                                         |
|                                                                             | You paused the agent, or you paused that scheduled task after the run started.                           |
|                                                                             | The agent started a newer run, or you sent the agent a message after the run started.                    |

A resumed run shows **Interrupted by maintenance · resumed automatically** in **Runs**. Its detail says "A model-provider hiccup interrupted this run; resumed automatically." It is listed under **ROUTINE INTERRUPTIONS** on **Health**, and it does not lower the agent's **Success** rate.

Squad does not resume a run that failed with a rate-limit error. See [Choose models and manage AI accounts](/docs/ai-accounts#troubleshooting).

## Follow the Live Feed

The **Live Feed** lists what your squad does, newest first. It shows today's items. If there is nothing today, it shows older items.

* **While you were away** sums up the events since your last visit, by agent, with the number of blocked items.
* **PREVIOUSLY SEEN** marks where your last visit ended.
* Filter by agent with **All agents** or an agent's name, and by type with **All activity**, **Missions**, **Needs you**, or **Updates**.
* Click an item to open its mission. An item without a mission opens its activity page, where you can read the detail and add a comment.
* Click **Load older activity →** to see more.

## Check that it works

* A run that you just started shows as **Running** in **Runs**, then **Completed**.
* **Health** shows **All clear** when nothing failed today.
* The **Live Feed** shows your agents' latest actions at the top.

## Troubleshooting

**A run from last month is missing.** Cause: **Runs** shows the last 30 days only. Fix: none in the dashboard. For older work, check the related mission or doc.

**"No runs match your filters".** Cause: the filters hide every run. Fix: turn agents or activity types back on in the **FILTERS** panel.

**An activity link shows "This activity isn’t available anymore."** Cause: the activity is older than the loaded part of the feed. Fix: open the **Live Feed** and click **Load older activity →**.