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

# Runs List

> A single view of every Run across all your agents — monitor active work, spot failures, review evaluation results, and drill into any run.

The Runs List gives you a single view of every Run across all your agents. Use it to monitor active work, spot failures, review evaluation results, and drill into any run without navigating agent by agent.

## Accessing the Runs List

<Steps>
  <Step title="Open from the sidebar" icon="menu">
    Click **Past Runs** in the left sidebar to open the Runs List.
  </Step>

  <Step title="Or open from an agent card" icon="ellipsis">
    You can also reach it from any agent card by clicking **Show past runs** in the card's action menu.
  </Step>
</Steps>

## What the List Shows

Each row represents one Run and displays:

* **Agent** — the agent that started the run; hover to see the full name if it's truncated
* **Evaluation** — for completed runs, an evaluation badge showing how many criteria passed or failed; click the badge to open the full evaluation breakdown
* **Run** — the run number linked directly to the run detail view; if you sent follow-up messages during the run, a small indicator next to the run number shows how many follow-ups were submitted
* **Trigger** — how the run was started (manual, scheduled, event-triggered, API, etc.)
* **Created** — when the run started; dates in the current year omit the year to reduce visual noise
* **Updated** — when the run last changed, shown only if it differs from the created time

Admins and Managers also see a **Created by** column showing which team member started each run.

## Run Statuses

Every Run moves through a lifecycle. The Runs List shows each run's current status, and you can filter by status to focus on the runs that need you.

```mermaid theme={"dark"}
stateDiagram-v2
    [*] --> Running
    Running --> NeedsInput: agent requests input
    NeedsInput --> Running: input provided
    Running --> Completed: finishes successfully
    Running --> Failed: error
    Running --> Stopped: stopped manually
    NeedsInput --> Stopped: stopped manually
    Completed --> [*]
    Failed --> [*]
    Stopped --> [*]
```

## Filtering Runs

Use the filter bar at the top to narrow the list:

* **Needs attention** — quick filter for runs with Failed or Needs Input status
* **Issues** — quick filter for runs that have evaluation issues flagged
* **Agent** — filter by a specific agent; agents are grouped by folder if your team uses folders
* **Status** — filter by Running, Needs Input, Completed, Failed, or Stopped
* **Created by** — filter by team member (visible to Admins and Managers)
* **Trigger** — filter by how runs were started (manual, scheduled, API, etc.)

Active filters are highlighted and each has an individual clear button. Use **Clear all** to reset all filters at once.

## Understanding Evaluations

Duvo automatically scores completed and failed runs against a set of criteria derived from the agent's AOP. Evaluation results surface in two places:

* **In the Runs List**: a badge on each completed run row showing the pass/fail breakdown at a glance
* **In the Agent Builder**: the Evaluations tab shows aggregated scores for the last 24 hours or last 7 days, plus any recurring issue alerts

Click any evaluation badge in the Runs List to open the full evaluation breakdown for that run.

### Evaluation Severity

Each failed criterion is assigned a severity level — **Critical**, **Medium**, or **Low** — based on how significantly it affects the run's outcome.

The badge in the Runs List reflects the highest severity found:

| Badge color | Meaning                                          |
| ----------- | ------------------------------------------------ |
| Red         | At least one Critical issue                      |
| Yellow      | Highest issue is Medium (no Critical issues)     |
| Grey        | All criteria passed, or only Low-severity issues |
| Dashed grey | Run was stopped before evaluation completed      |

When you open the full evaluation breakdown, failed criteria are grouped and sorted by severity so the most important issues appear first. Click any row to expand a short explanation of what the agent did wrong and what it should have done instead. Critical issues expand automatically; Medium and Low issues start collapsed so you can scan and drill in as needed.

Older runs from before severity tracking was introduced show a generic issue count without severity breakdown.

## Pagination

The list shows up to 40 runs per page. If your team has more, use the pagination controls at the bottom to move through pages.