Sometimes you do not need another dashboard. You need one answer: did this page change?
That sounds simple, but a useful page monitor has to make four decisions well. Here is the actual one-click setup in Firefox, running against the production service.
1. Fetch outside the browser
A browser extension is a convenient way to add the current page, but it should not be the scheduler. If checks run only inside Firefox, they stop when the browser or laptop closes. Store the URL on a server and schedule the fetch there.
For a small monitor, the flow can stay boring:
- validate that the URL is public HTTP or HTTPS;
- fetch the rendered HTML;
- extract visible text;
- hash the normalized result;
- compare it with the previous hash.
Server-side checks also make the behavior easier to explain: the monitor keeps working even when Firefox is closed.
2. Normalize before comparing
A raw HTML diff is usually too noisy. Timestamps, rotating IDs, analytics attributes and reordered markup can all change without changing what a person sees.
Start with visible text. Collapse repeated whitespace and remove known volatile elements. Keep both the old and new version so an alert can show evidence, not just say that something happened.
This does not eliminate every false positive. It does make the remaining ones debuggable.
3. Treat screenshots as evidence
Text tells you that a value changed. A before-and-after screenshot often tells you why. Keep screenshots with the check result and timestamp them. They are especially useful for prices, availability messages and landing-page copy.
Be explicit about the boundary: public HTML pages are a sensible first version. Logged-in pages, PDFs and highly dynamic apps require different fetching and comparison strategies.
4. Make the alert actionable
A good alert contains:
- the page title and URL;
- the check time;
- the previous and current evidence;
- a direct path to inspect or disable the monitor.
Do not send an email merely because the fetch failed once. Network errors are not page changes. Record the error, retry, and only notify according to a separate failure policy.
A small implementation I am testing
I built Repondea Page Monitor around this shape: a Firefox extension adds the current page in one click, while the server performs the checks and keeps before/after captures. It currently supports public HTML pages only. One daily page is free; the paid plan checks up to ten pages hourly.
You can try the web version without installing the extension: https://repondea.com/en/monitor?source=devto
The part I am still tuning is false-change noise. If you use page monitors today, what kind of change wastes most of your alerts?
Top comments (0)