Actor Reliability Watchdog – Uptime & Schema Drift Monitor

This is a buyer-side monitor, not a builder-side gap finder.

Every "opportunity finder" or "gap finder" actor on the Apify Store targets builders deciding what to build next. This one is for the other side of that relationship: you already depend on 20+ actors in production — your own and third-party — and when one of them silently breaks (a target site changes its DOM, an API adds a required field, a login wall goes up), you find out when downstream data looks wrong, not when it actually broke. This actor closes that gap by actually running the actors you depend on and checking what comes back.

What it does

For each actor you're watching:

1. Triggers a real smoke-test run with a known-good input you supply
2. Compares the returned dataset's fields against the fields you expect
3. Pulls the actor's own recent run history to compute a success rate over your check window
4. Classifies the result ok / warning / critical and pushes one report row per watched actor

Input schema

{
  "watchedActors": [
    {
      "actorId": "conceivable_extension/uk-business-leads",
      "expectedFields": ["companyName", "companyNumber", "sicCode"],
      "minSuccessRate": 0.9,
      "testInput": { "...": "a known-good sample input for a smoke-test run" }
    }
  ],
  "checkFrequencyHours": 24
}

checkFrequencyHours sets the look-back window for the success-rate calculation — match it to how often your schedule triggers this actor.

Output schema

{
  "actorId": "string",
  "checkedAt": "ISO timestamp",
  "runSucceeded": "boolean",
  "successRate": "number",
  "missingFields": ["array of expected fields not found in output"],
  "schemaDrift": "boolean",
  "alertLevel": "ok | warning | critical",
  "notes": "string"
}

Severity thresholds

Pricing

This is an internal-tool-first actor — its main value is protecting your own portfolio. Priced modestly if published, not as a primary revenue line.

Setup note

No proxy or browser needed — this is a plain Node actor (apify/actor-node:20 base image, no Playwright/Chrome) that calls the Apify API directly via Actor.newClient(). It needs your Apify API token available at runtime (automatic when run on the Apify platform) with permission to call the watched actors and read their run history.

Each smoke-test run costs whatever the watched actor itself charges (PPE or compute units) — this actor is triggering real runs, not reading cached stats, so budget for that on actors with non-trivial per-run cost.

n8n integration

• Workflow A (trigger): scheduled run every 24h (or your checkFrequencyHours) against all actors you depend on — your own 20+ and any third-party actors your pipelines call.
• Workflow B (processing): on critical or warning alertLevel, push a Slack/email notification with the specific actor, its missing fields, and its last-known-good run for comparison.