> For the complete documentation index, see [llms.txt](https://docs.autommate.com/userguide/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.autommate.com/userguide/mcp-and-ai-reference/read-platform-summaries-schedules-and-agent-health.md).

# Read platform summaries schedules and agent health

Use summaries, running schedules and agent metrics to choose the next diagnostic read.

Use operational summaries to locate an investigation. Use individual run evidence and external result checks to establish what happened.

### Platform summary

platform\_get\_summary accepts window values 24h, 7d or 30d. Its response example includes:

| Section      | Observed fields                                                                                |
| ------------ | ---------------------------------------------------------------------------------------------- |
| Window       | `window`, `period.from`, `period.to`, `generated_at`                                           |
| `automs`     | `total`, `active_hyper_flows`, `remaining_slots`                                               |
| `operations` | `in_progress`, `queued`, `window_total`, `window_finished`, `window_completed`, `success_rate` |
| `alerts`     | `count`                                                                                        |
| `agents`     | `total`, `connected`, `disconnected`                                                           |

The aggregate operation state `queued` is not a declared `runs_search.status` value; that filter accepts `pending`. Do not silently substitute one vocabulary for another without the product contract.

The example generated\_at uses a UTC suffix, while period.from and period.to have no offset. Preserve timestamps as returned and confirm the time zone before combining them with another report.

`success_rate` describes an operational metric. It does not by itself prove the business outcomes of the included runs.

### Mate Agent health

agents\_get\_health reads recent host metrics. Set live\_check: true to additionally probe reachability; the examples below use false.

An agent record can contain these example fields:

* Identity and freshness: `agent_id`, `name`, `connectivity`, `last_metric_at`, `metrics_age_seconds`.
* Runtime: `version`, `mode`, and `device` fields `os`, `cpu`, `ram`, `timezone`.
* Health: `severity`, `score`, `throttle_active`, `throttle_tier`.
* Service metrics: `name`, `instance_count`, `cpu_percent`, `memory_percent`, `severity`.
* A `live` field, returned as null when no probe result was available in the sampled read.

The example envelope includes items, next\_cursor, total and a summary with total, online, stale and unknown. Check each tool’s scope and freshness before comparing these counts.

Agent version describes that runtime component, not the platform or MCP release. Shared deployments can list agents outside the caller’s tenant; confirm the active connection independently of agent names.

### Running schedules

The schedules\_search example uses items and next\_cursor without total. Schedule records include:

{% code collapsedlinecount="10" %}

```
autom_id, autom_name, description,
next_run_at, last_run_at, timezone,
version.label, version.type,
agent_name, started_by
```

{% endcode %}

This tool lists running schedules. Paused or disabled schedules can be absent. Read the relevant Autom for its schedule summary and running state instead of treating absence as deletion.

After a schedule’s due time, use run history to check whether its execution occurred.

### Choose the next read

| Question                                  | Next read                                                          |
| ----------------------------------------- | ------------------------------------------------------------------ |
| Is work still running?                    | Summary, then `runs_search` with `in_progress`.                    |
| Why did this Autom fail?                  | `runs_get_latest_error`, then the selected run.                    |
| Is the execution host reporting recently? | `agents_get_health` and metric age.                                |
| When should an active schedule run next?  | `schedules_search`.                                                |
| Did the requested external change happen? | Inspect the run output and the authoritative target-system record. |

None of these reads grants permission to restart a run, change a schedule, or modify a connection.