Apify Actor Run Monitor: Empty Output Alerts

A scheduled Actor can finish with status SUCCEEDED and still deliver nothing your pipeline can use. This monitor checks execution status and output health on up to 50 Actors in one pass, so an empty dataset, a schedule that stopped firing, a run of repeated failures or a field that vanished from the output become explicit alerts instead of a report that goes wrong downstream three days later.

What the health monitor checks on every Actor

• whether the latest run exists and is recent enough;
• whether the latest run failed, timed out, or was aborted;
• the failure rate across the most recent 1–20 runs;
• whether the latest successful dataset meets a minimum item count;
• whether one output sample contains every required dotted field path.

Silent failures this catches

• catch a successful schedule that silently produced an empty dataset;
• audit an Actor portfolio after deployments or schema changes;
• watch stale, failed, timed-out, or repeatedly failing automations;
• verify required output fields before downstream workflows consume the data.

Input-Output

The default input checks the public apify/web-scraper Actor. Apify's limited-permission run token cannot read Actor run or dataset metadata, so add a personal API token for actionable health results. Without one, inaccessible targets are returned as unknown rather than failing the run:

{
  "actorIds": ["apify/web-scraper"]
}

Add a personal token and your health contract for public or private portfolios:

{
  "apiToken": "YOUR_APIFY_PERSONAL_API_TOKEN",
  "actorIds": ["apify/web-scraper", "apify/website-content-crawler"],
  "maxAgeHours": 24,
  "recentRuns": 10,
  "failureRateThresholdPercent": 50,
  "minimumItems": 1,
  "requiredFields": ["url"],
  "alertsOnly": false
}

Apify encrypts the secret apiToken at rest and decrypts it only for the run. The token is sent only to api.apify.com in the authorization header and is never included in output or logs. Give it only the permissions needed to read the monitored Actors, runs, and datasets.

Data returned for each Actor

Field	What it holds
actorRef	The Actor inspected, as you passed it
health	ok, warning, unhealthy or unknown
alertReasons	Why it is not ok, in plain language
latestStatus / latestRunAgeHours	State and age of the most recent run
failureRatePercent	Share of failures across the recent runs inspected
lastSuccessfulItemCount	Dataset size of the last successful run
missingFields	Required dotted paths absent from the sampled item
checkedAt	Timestamp of the inspection

The SUMMARY record aggregates healthy, warning, unhealthy, and unknown results.

Pricing

Pay per Actor inspected. The actor-checked event is charged once before each inspection, including healthy Actors omitted by alertsOnly. The run stops before unpaid work when the buyer's maximum charge is reached. Dataset rows themselves are not a second billable event.

Reading health states

• ok: no configured rule was violated.
• warning: the latest run is still in a non-terminal state.
• unhealthy: no run exists, a run appears stuck or failed, recent failures cross the configured threshold, the latest successful dataset is too small, or required fields are absent.
• unknown: the Actor or its storage could not be read; the error is isolated so the rest of the portfolio is still checked.

FAQ

Does this rerun or modify monitored Actors?

No. It only reads run and dataset metadata.

Why is my Actor unknown?

The Actor or its storage could not be read. Add a personal token with the minimum read permissions needed for the target, then check the returned reason.

Does it inspect every dataset row?

No. It reads the item count and at most one sample row per Actor for required-field checks.

Limits and responsible use

• Maximum 50 Actors per run.
• Maximum 20 recent runs inspected per Actor.
• Maximum 50 required field paths, 256 characters and 20 levels per path.
• A run charges once before each inspection and stops when no further actor-checked event fits the buyer's limit.
• At most one dataset item is sampled per Actor for field checks.
• Set minimumItems to 0 for Actors that intentionally produce no dataset rows.
• Field checks inspect one sample and do not prove every row has the same schema.
• Only Actors and storages accessible to the account that launches the run can be inspected.
• No proxy, browser, login scraping, personal-data targeting, automatic reruns, or outbound webhooks.

This is an unofficial independent utility and is not affiliated with Apify. Review alert thresholds against each Actor's expected schedule and output contract.