DOCS

Discover monitors from a domain

Point HowlOps at a domain and get a curated list of monitor suggestions instead of typing every URL by hand.

All Uptime plans, including Free

Endpoint Discovery scans a domain, works out what is actually worth monitoring, and lets you create monitors for it in a couple of clicks instead of typing every URL by hand. It is aimed at the moment your monitor list is empty: point it at your domain and it hands you a short, curated list instead of an empty form.

The fastest path: domain → Scan → Create

Go to Monitors, click New monitor, then click Discover from a domain in the top right.
Type your domain (for example example.com) and click Scan. This only reads what is publicly published on the domain; nothing is monitored yet.
Review the Recommended list. It comes pre-checked with the handful of things most sites want watched: the homepage, a login/checkout page if one was found, an API endpoint if one exists, and a TLS certificate + DNS check for the domain itself.
Click Create N recommended monitors. You're done.

That's the whole happy path: three clicks, no manual URL entry. Everything described below is optional, progressive disclosure you can ignore if the recommended set is good enough.

What sources are used

A scan looks at what your domain publishes and stops there. It never renders pages or crawls links inside JavaScript-rendered content:

robots.txt — read for Sitemap: lines.
sitemap.xml (and any sitemaps it points to, including a sitemap index and the sitemaps listed inside it) — used as the primary source of page URLs. If robots.txt doesn't name a sitemap, HowlOps tries the conventional /sitemap.xml path as a fallback.
OpenAPI / Swagger — /openapi.json, /openapi.yaml, or /swagger.json, if one of them exists. Discovered operations show up in their own API section.
The infra bundle (always included) — a TLS certificate check and a DNS records check for the domain itself. These are recommended by default because certificate expiry and DNS outages are two of the most common "we didn't notice until it broke" failures.

Every discovered link is checked against the domain you scanned; anything pointing at a different domain (a hostile or misconfigured sitemap referencing another site, for example) is reported as a warning and never fetched or offered as a monitor.

A scan only reads what a page found; every fetch it makes is best-effort, so if robots.txt or one sitemap file is unreachable, the scan just continues with fewer results rather than failing outright. The homepage is always suggested even if nothing else could be read.

How grouping and recommendations work

Large sites can have thousands of URLs. Discovery never dumps all of them on you:

URLs are grouped by their first path segment (/products/wireless-mouse and /products/keyboard both fall under /products).
A group with more than a handful of pages collapses into a single row, for example /products/* (842 pages), with one representative page you can monitor as a stand-in for the whole section. You can still open Review all to see the individual pages if you want one specifically.
Each URL is classified into a category — Homepage, Authentication, API, Checkout, Documentation, Content, Infrastructure, or Other — using its path and hostname. There's no AI involved: it's simple, deterministic keyword matching, so the same site always classifies the same way.
A small set (typically around five to ten items) is pre-checked as Recommended: the homepage, the shortest-looking login/auth page, the shortest-looking API endpoint, the shortest-looking checkout page, and the TLS + DNS infra checks. Everything else starts unchecked.

Reviewing everything

If the recommended set isn't enough, click Review all to see every discovered group and individual page and pick exactly what you want. Suggested monitor type and check interval are pre-filled based on the category (for example, TLS checks default to once a day) but you can still change type, interval, and region after the monitor is created, same as any other monitor.

If your site publishes an OpenAPI or Swagger document, its operations appear in a dedicated API section. GET and HEAD operations behave like any other candidate. POST, PUT, PATCH, and DELETE operations are shown too (so you know they exist) but are never pre-checked, and each one needs you to explicitly tick "I understand this issues a <METHOD> request" before it can be selected — HowlOps never auto-creates a monitor that would write to your API. Creating a write-method monitor also requires the domain to be verified; see below.

Domain verification unlocks the stronger options

Scanning and creating the recommended monitors works on any domain, verified or not. But a few more powerful actions are gated behind proving you actually own the domain, so HowlOps can't be used to quietly hammer someone else's site:

On an unverified domain, every discovered monitor is capped to a minimum 60-second interval (even on plans that normally allow faster checks), restricted to GET/HEAD requests, and a single Create step can produce at most 20 monitors.
Monitors for OpenAPI write operations (POST/PUT/PATCH/DELETE) can only be created once the domain is verified.

If a scan comes back gated, the results page shows a Verify <domain> to unlock panel right there, with both verification methods available inline. See Verify domain ownership for the full walkthrough of both methods and exactly what each one unlocks.

Tracking coverage over time

After creating monitors, click View coverage (or go to Monitors → Coverage and enter a domain you've already scanned) to see how much of what was discovered is actually being watched: a coverage percentage, a breakdown by category, and a New since last scan list of endpoints Discovery has found that you haven't made a decision about yet. From that list you can add a monitor or mark an endpoint as intentionally ignored (so it stops showing up as "new"). Coverage never fetches your site on its own; it reads your most recent scan, so run Scan again on the same domain whenever you want a fresh comparison.

Staying in sync as your site changes (re-sync)

Once you import at least one monitor from a domain, HowlOps starts watching that domain for drift: the gap between what your site actually publishes and what you monitor. The valuable signal is not every new page that appears, it is the opposite, a page you monitor that has left the site's own map. If /product?id=1 is monitored and it quietly disappears from the sitemap while your monitor still points at it, that is worth a second look. It might have been retired on purpose, or your monitor might already be alerting on a real outage.

When any watched domain drifts, the Monitors page shows a small "N domains are out of sync" notice with a link to review. The Coverage page for that domain then lists what changed:

Gone — a monitored endpoint that is no longer in the latest scan. Each one shows the monitor's own live status (up, down, or paused) and the date it left the scan, so you can tell an intentional removal apart from a real outage your monitor is already catching. You choose per endpoint: Keep it as is (the sitemap may simply be incomplete, and this stops the nagging), or Retire it, which pauses the monitor. Retiring never deletes a monitor.
Moved — a Gone endpoint that looks like the same resource as a newly found one (a similar page name or address, for example /product?id=1 becoming /p/1). Here you can Repoint the existing monitor at the new address. Repointing keeps the monitor and all of its history and settings, it just changes where it checks, which is much better than deleting the old one and starting a new one from scratch. If the suggested match is wrong, you can pick a different destination, or fall back to Keep or Retire.
New — endpoints found on the site that you do not monitor yet. These are additive and never make a domain "out of sync" on their own. You can tick several and add them as monitors in one step.

HowlOps never changes a monitor on its own from drift. Every Keep, Retire, and Repoint is your decision, and every one of them is reversible. Watching is automatic once you import from a domain; you do not need to switch anything on.