Dashboard¶
The Dashboard summarizes the root-cause agent's runs in the project selected in the project switcher: recent runs, how many ran this week, how long they take, and what they cost. Open it with Dashboard in the sidebar. This page explains what each card and chart counts and over what period.
What the numbers are based on
Except for the all-time run total, every number on the Dashboard is computed from the project's 200 most recent runs. Weeks, months and days are calendar periods in UTC. The Dashboard counts root-cause runs only; for failures and triage, use Signatures and the triage queue.
Recent runs¶
The four newest runs, each as a card with the run's short id, its status, its scenario (and how many more it covered), how long ago it started, its cost once known, and the model it used. Select a card to open the run. With no runs yet, the strip offers Start a run.
KPI cards¶
| Card | What it shows |
|---|---|
| Runs this week | Runs submitted since the beginning of the current week (Sunday, 00:00 UTC). The line underneath gives the project's total runs to date. |
| Avg confidence | Not computed yet: shows —. |
| Hallucinations caught | Not computed yet: shows 0, with the number of completed runs underneath. |
| Mean time to root-cause | The average duration of completed runs (the time the agent spent across each run's scenarios), with how many completed runs it averages. It measures the agent's time, not the time from a failure first appearing to someone confirming its root cause. Shows — when no run has completed. |
Latest hypothesis¶
Shows the newest completed run (its short id and scenario) with an Open link. The hypothesis itself isn't shown on the Dashboard; select Open to read it on the run page (see Read the hypothesis).
Run cost¶
A bar chart of what runs cost per day over the last 14 days, by the day each run started. The description above the chart shows the total spent this month (since the 1st, UTC). Runs without a cost yet (still running, or failed before reporting one) add nothing. With no cost in the last 14 days, the card shows "No data yet — kick off runs to populate this view."
Activity breakdown¶
| Chart | What it shows |
|---|---|
| Status mix | A pie chart of runs by status (submitted, running, complete, failed). Labeled "All-time breakdown", it covers the 200 most recent runs. |
| Model usage | A bar chart of how many runs used each model. |
| Pass rate by scenario | A placeholder; it stays empty today. |