> ## Documentation Index
> Fetch the complete documentation index at: https://smithers-feat-claude-workflow-mirror.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# monitor-smithers

> Watchdog over Smithers runs that detects stuck, blocked, failed, or over-budget runs and escalates.

`monitor-smithers` is a watchdog over your Smithers fleet. It reads the live run
table, sorts every run into a health bucket, and, only when something is wrong,
produces concrete escalation actions (which gate to clear, which run to triage,
which to cancel). Reach for it on a schedule or whenever you want a quick
"is anything stuck?" sweep of in-flight runs.

```bash theme={null}
bunx smithers-orchestrator workflow run monitor-smithers --input '{"staleMinutes":15}'
```

## Stages

1. **poll**: deterministic (no agent), shells out to `bunx smithers-orchestrator ps --format json`,
   parses it, and normalises each run into `{ runId, status, ageMinutes, lastEvent }`.
   Any failure degrades to an empty list rather than throwing.
2. **classify**: a cheap/fast agent sorts the runs into buckets: `healthy`,
   `stuck`, `blocked`, `failed`, `overBudget`.
3. **triage**: runs only when a non-healthy run exists. A smart agent surfaces
   the pending approval/question for blocked runs (`bunx smithers-orchestrator why`, `bunx smithers-orchestrator approve`),
   recommends the `triage-run` workflow for stuck/failed runs, and returns a digest.

## Inputs

| Input          | Type   | Default                                                      |
| -------------- | ------ | ------------------------------------------------------------ |
| `staleMinutes` | number | `15` (a run idle past this many minutes is treated as stuck) |

## Monitor it

```bash theme={null}
bunx smithers-orchestrator workflow run monitor-smithers -d   # detach
bunx smithers-orchestrator ps                                 # active / paused / recent
bunx smithers-orchestrator logs RUN_ID -f                    # follow events
bunx smithers-orchestrator inspect RUN_ID                    # full run state + outputs
```

A healthy fleet short-circuits before the expensive triage step, so a clean
sweep is cheap. When the triage step does run, its `digest` and `actions` tell
you exactly which runs need a human and what to do first.