kestrel Apify actors — documentation

30 scrapers for travel prices, guest reviews, ecommerce and app stores, each one an Apify actor. Every actor bills per delivered row: rows a filter removed, empty runs and failed targets are free. Every date input takes a relative value like 30 days, so a schedule never goes stale, and none of them needs an API key on the site being read. Each page below is the full reference for one actor — every input, every output field, a real row, working code and its honest limits.

Start here

Install the client, set your Apify token, call an actor by name:

pip install apify-client          # or: npm i apify-client
export APIFY_TOKEN=...            # console.apify.com/account/integrations

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("kestrel/agoda-reviews-scraper").call(run_input={
    "hotelNames": ["Park Hyatt Tokyo"], "maxReviewsPerHotel": 200,
    "reviewsSort": "most_recent",
})
for row in client.dataset(run["defaultDatasetId"]).iterate_items():
    if row["type"] == "review":
        print(row["rating"], row["reviewer_country"], row["text"][:80])

The same call in JavaScript and curl, for every actor, is in the examples repository — together with importable n8n workflows.

How billing works

Every actor here is pay-per-event, and the event is a delivered row. Three consequences worth knowing before you plan a job:

• Filters run before billing. A complaints feed that reads 300 reviews and keeps the 8 below your rating ceiling bills 8 rows. The status row still reports how many were filtered.
• Nothing is charged for nothing. Free context rows (the hotel, listing, product or company row), status rows, targets with no data, unknown ids, duplicate inputs and failed targets are all free.
• Charged equals delivered. When a run reaches its spending limit it stops delivering rather than writing rows it could not charge for, so the dataset and the invoice agree.

Per-row prices differ by source and are listed on each actor page, and on the pricing tab of its Apify listing. The cheapest row in the suite is a Trustpilot review; the most expensive is a rate-parity row that compares every booking site for one property.

Hotel & flight prices

Guest & customer reviews

Ecommerce

Apps

Guides: how to scrape it, what it costs, where it breaks

The pages above are the reference for each actor. These 20 guides answer the question first — how a source is reached, which of its behaviours quietly corrupt data, what a job costs, and which actor ends the search. Every fact in them is measured, and the limitations are stated.

Ready-made runs

163 of these actors' inputs are saved and published as ready-made runs — open one on the Apify Store, press Start, and it goes with the fields filled in. Useful both as a first run and as a worked example of what a sensible input looks like.

Which review scraper do I need?

Review scrapers are 17 of the 30, and they differ in what a row carries and in how deep a single property or company goes — from four comments on Despegar to the whole corpus on Trip.com. The review scraper comparison lays out the row contents, the sub-ratings, the reply field, the language coverage, the per-row price and the real ceiling for each one.

FAQ

What am I charged for?

One event per delivered row. Rows a filter removed before billing, free context and status rows, targets with no data, unknown ids and failed targets cost nothing, and a run that delivers nothing costs nothing. Each actor's pay-per-event schema is on its own page.

Do I need an API key for the site being read?

No. The actors read public pages and the JSON endpoints those pages call, with no login and no account on the source site. The only credential is your Apify token.

Will a scheduled run go stale?

No. Date inputs take relative values — checkIn: "30 days", sinceDate: "7 days" — so a schedule keeps meaning the same thing every morning.

Can I export to CSV or Excel?

Yes. Every run writes an Apify dataset, exportable to CSV, Excel, JSON or XML from the Console, or over the API by appending ?format=csv to the dataset items endpoint.

Can I call these from n8n, Make or an AI agent?

Yes. Each actor is callable through Apify's integrations, over the REST API, and from an AI agent through Apify's MCP server. The examples repository carries importable n8n workflows.

What happens when a run hits a spending limit?

The run stops delivering rather than handing you rows it could not charge for, so the row count in your dataset and the count on your invoice always agree.