Listing Watch — Job Alert Monitor for Computrabajo

Turn a one-shot scraper into a subscription. listing-watch schedules computrabajo-jobs-scraper, diffs every run against a stored history of job URLs, and delivers only the listings that are new since your last check. You pay a small per-check fee plus a per-new-listing fee — the premium is for the alert, not for re-reading data you already have.

---

What it does

listing-watch is a wrapper actor: it runs computrabajo-jobs-scraper on your behalf with your search parameters, compares the results against a stateful key-value store of previously seen job URLs, and outputs a compact WatchResult containing only the new listings. Nothing is re-processed and nothing duplicate reaches your dataset.

It works by:

1. Running computrabajo-jobs-scraper with your keyword, country, and optional location/date filters
2. Comparing the returned job URLs against a stored list from your last run (URL is the stable dedup key)
3. Pushing a single WatchResult record with the new listings only
4. Updating its stored state so the next run picks up only what's fresh from there

Run it on an Apify Schedule (daily, hourly, whatever cadence fits) and you get a continuous, deduplicated feed of new Computrabajo postings — a job alert service built entirely from an existing one-shot scraper plus a diff layer.

---

Use cases

• Job alert automation — schedule a daily Computrabajo search for your ideal role and get a compact feed of only new postings, ready to pipe into Slack, email, or your ATS.
• New job listings monitor — track new posting volume by role, location, and country across LatAm without writing your own dedup logic.
• Competitor hiring monitoring — watch what roles a specific employer or industry is hiring for, and get notified the moment a new one goes up.
• HR tech / recruiting pipelines — feed a continuously deduplicated stream of new LatAm job listings into a matching engine, analytics dashboard, or candidate-alert system.
• Labor market trend tracking — measure new-posting velocity by keyword and geography over time, run after run, without manual bookkeeping.

---

Input example

{
  "keyword": "desarrollador python",
  "country": "ar",
  "location": "buenos aires",
  "sinceDate": "2026-07-01",
  "maxNewItems": 20
}

Field	Type	Default	Description
targetActor	enum	computrabajo-jobs-scraper	Which actor to watch. More sources coming in v2.
keyword	string	"desarrollador"	Job search keyword to monitor
country	string	"ar"	Computrabajo country code (ar, mx, co, pe, cl, etc.)
location	string	(none)	Optional city/province filter
sinceDate	string	(none)	ISO date YYYY-MM-DD. Only monitor jobs posted on or after this date.
maxNewItems	integer	20	Maximum new listings to return per run (1–200)
detectRemoved	boolean	false	v2 feature — currently a no-op

Each unique combination of keyword + country + location is one independent "watch" with its own stored history — reuse the same input on a schedule to keep it going.

---

Output example

[
  {
    "watchId": "9849d0d277328a70d1391d105a029ab101c8691646a705a0f7c6709351879832",
    "targetActor": "computrabajo-jobs-scraper",
    "keyword": "desarrollador python",
    "country": "ar",
    "newCount": 3,
    "totalChecked": 60,
    "lastRunAt": "2026-07-22T12:00:00.000Z",
    "checkedAt": "2026-07-23T12:00:00.000Z",
    "previousListingCount": 55,
    "newListings": [
      {
        "url": "https://ar.computrabajo.com/ofertas-de-trabajo/oferta-de-trabajo-de-senior-backend-python-developer-remoto-para-residentes-en-argentina-en-monserrat-F2A4E9D19A683AFB61373E686DCF3405",
        "title": "Senior Backend Python Developer",
        "company": "Kaizen Recursos Humanos",
        "location": "Monserrat, Capital Federal",
        "salary": { "raw": "A convenir", "min": null, "max": null, "currency": null },
        "contractType": "Contrato por tiempo indeterminado",
        "workMode": "remote",
        "postedDate": "2026-07-23",
        "description": "En KaizenRH buscamos Senior Backend Python Developer...",
        "source": "computrabajo",
        "country": "ar"
      }
    ]
  }
]

Each dataset item is a WatchResult containing:

• Watch metadata — watchId, targetActor, keyword, country, checkedAt, lastRunAt
• Counts — newCount, totalChecked, previousListingCount
• New listings — newListings[] — array of Job objects (same schema as computrabajo-jobs-scraper: title, company, location, salary, contract type, work mode, posted date, description, URL)

The very first run of a watch has no history to compare against, so all returned listings count as "new" — that's your baseline. Every run after that reports only what actually changed.

---

Pricing

This actor uses pay-per-event pricing — no rental, no subscription, no charge for duplicate listings.

Event	Price	What triggers it
Actor start	$0.001	One-time per run (covers orchestration compute)
listing-checked	$1.00 / 1,000 ($0.001/ea)	Charged once per watch run — the cost of running the check itself
listing-new	$5.00 / 1,000 ($0.005/ea)	Charged per genuinely new listing discovered. The premium over raw scraping pays for the dedup + alerting value, not the data

Important — this actor internally runs computrabajo-jobs-scraper on your behalf. That sub-run is billed to you separately, at computrabajo-jobs-scraper's own published rates (currently ~$0.001 start + $0.0015 per job scraped). You'll see it as its own line item in your Apify usage. listing-watch's fees above are only for the orchestration and alerting layer on top.

Real-world cost examples (measured on live platform validation runs, keyword: "desarrollador", country: "ar"):

Scenario	Jobs checked	New listings	listing-watch fees	computrabajo pass-through	Total
First run (seed watch) — measured	15	5	$0.027	$0.024	~$0.05
Second run, nothing new — measured	15	0	$0.002	$0.024	~$0.03
Larger watch (100 checked, 10 new)	100	10	$0.052	$0.151	~$0.20

Most of the steady-state cost is the underlying Computrabajo scrape (unavoidable — the only way to know what's new is to re-check what's there). listing-watch's own fee stays low when nothing changes and scales only with genuinely new results.

---

Use with AI agents (MCP)

This actor is available as a tool for AI assistants via the Apify MCP server. Agents can subscribe to a job search alert or new listings monitor directly — no Apify console needed, just a function call.

MCP setup

{
  "mcpServers": {
    "apify": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.apify.com?tools=mfapitools/listing-watch",
        "--header",
        "Authorization: Bearer <YOUR_API_TOKEN>"
      ]
    }
  }
}

Example agent prompts

• "Check if there are any new Python developer jobs on Computrabajo Argentina since last week"
• "Set up a watch for remote React jobs in Mexico and tell me only about new postings"
• "Has anything new been posted for 'contador' roles in Buenos Aires?"

The actor returns structured JSON with the fields shown above. Agentic workflows can run this on a schedule (via the Apify Scheduler) and only need to look at newListings on each check — no manual dedup logic required on the agent side.

---

Limitations

• v1 — Computrabajo only. More sources (Trustpilot, licitaciones-feed, others) are candidates for v2; targetActor is an enum of one value today.
• Detect removed listings is v2. The detectRemoved input field currently has no effect.
• No webhook/email notifications in v1. The actor writes to its dataset each run; connect it to Slack, email, or a ticketing system via Apify webhooks or the dataset API.
• Single keyword per watch. One watch = one keyword + country + location. For multiple keywords, create multiple schedules with different inputs.
• State stored in a named key-value store (listing-watch-state), one key per watch (hash of targetActor + keyword + country + location). URLs are capped at 1,000 per watch, newest-preserved, to bound storage growth — job postings older than the most recent 1,000 for a given search won't be tracked for dedup.
• Computrabajo's own charges are billed to you separately — see Pricing above. This actor does not mark up or hide that cost; it's a transparent pass-through from the underlying scrape.
• Cross-actor invocation. This actor calls computrabajo-jobs-scraper (ywW0ObCeHcOu3RdpJ) internally via Actor.call(), with an automatic fallback to the Apify REST API if Actor.call() is unavailable.
• Compliance inherits from the wrapped actor. All the same public-data-only guarantees of computrabajo-jobs-scraper apply — this actor adds no new scraping surface of its own.

---

Changelog

See ./CHANGELOG.md.