Website Change Monitor

Watch any web pages and detect changes between runs. Stores a snapshot and reports what changed.

What it does

• Fetches one or more URLs and extracts their content with Cheerio.
• Compares each page against a snapshot saved from the previous run.
• Reports whether the page changed, with added and removed line counts and a short summary.
• Optionally watches a single region via a CSS selector.
• Optionally ignores numbers to cut noise from counters, view counts, and dates.
• Built to run on a schedule so each run compares to the last one.

Example input

{
  "urls": [
    "https://example.com",
    "https://news.ycombinator.com"
  ],
  "selector": "",
  "mode": "text",
  "ignoreNumbers": false
}

Input

Field	Type	Description
urls	array	One or more page URLs to fetch and monitor. Required.
selector	string	Optional CSS selector to watch only one region (for example #price). Empty watches the whole page.
mode	string	text compares visible text (less noisy), html compares raw markup. Default text.
ignoreNumbers	boolean	Replace digits before comparing to reduce noise. Default false.

Output

One dataset item per URL:

{
  "url": "https://example.com",
  "changed": true,
  "isFirstRun": false,
  "previousHash": "9c1185a5c5e9fc54612808977ee8f548b2258d31",
  "currentHash": "a3f5b2c1d4e6f7089aabbccddeeff00112233445",
  "addedLines": 3,
  "removedLines": 1,
  "changeSummary": "3 line(s) added, 1 removed. + New headline | - Old headline",
  "checkedAt": "2026-07-04T12:00:00.000Z"
}

On the first run for a URL, isFirstRun is true, changed is false, and the baseline snapshot is stored for next time.

Run summary (OUTPUT)

The run's default key-value store record OUTPUT holds an aggregate:

{
  "urlsChecked": 2,
  "changed": 1,
  "new": 1,
  "unchanged": 0,
  "errors": 0,
  "mode": "text",
  "selector": null,
  "ignoreNumbers": false,
  "checkedAt": "2026-07-04T12:00:00.000Z"
}

How it works

The Actor fetches each URL with a plain HTTP request (retrying on 429 and 5xx with backoff), extracts the text or HTML with Cheerio, normalizes it into lines, and computes a content hash. Snapshots are kept in a named key-value store keyed by a hash of each URL, so a scheduled run always compares against the previous run. When the hash differs, it reports the added and removed lines. It needs no API key.