Webhooks send monitor events to a URL you choose, so Slack, Discord, n8n or your own code can react the moment something changes. They are free on every account.
Set one up
Open Webhooks on your dashboard and choose New webhook. Enter the endpoint URL (it must start with http:// or https://), an optional label, and the events you want. Select none to receive all of them. A webhook belongs to one monitor and only hears about that monitor.
The signing secret is shown once, when you create the webhook. Copy it then; it cannot be shown again. Send test delivers a sample, and the webhook list shows whether the last delivery worked.
Pasting a Discord channel webhook URL or a Slack incoming-webhook URL needs nothing else. Link Robins recognizes them and sends a proper message instead of the JSON below.
The events
incident.opened: a monitor went down
incident.resolved: a monitor came back up
monitor.status: any status change, including pauses
ssl.expiring: a certificate is close to expiring
domain.expiring: a domain registration is close to expiring
status_page.update: a status page posted an update
flarum.issue: a Flarum forum check found something new
monitor.maintenance_started and monitor.maintenance_ended: a scheduled maintenance window began or ended
These come with Link Robins Pro:
monitor.regional: a monitor is down in one region only, or has recovered there
monitor.slow and monitor.slow_resolved: the page stayed slower than your threshold, and is back to normal
forum.spike and forum.quiet: a burst of new members or posts, or a forum that has gone quiet
dns.changed: nameservers, mail servers or the address changed
security.changed: the daily security check found a worse grade or a new serious finding
page.changed: a page looks completely different or blank
forum.start_failed and forum.start_recovered: a forum's JavaScript stopped starting, and started again (see Forum starts (Pro))
login.failed and login.recovered: members can't sign in to a forum, or the check paused itself, and sign-in works again (see Login check (Pro))
lighthouse.score_dropped: a daily Lighthouse score dropped and stayed lower on two reports (see Lighthouse report (Pro))
crawl.new_broken_links: the weekly site crawl found new broken pages or links (see Site crawl (Pro))
What a delivery looks like
Every delivery is a POST with a JSON body. This is incident.opened:
{
"event": "incident.opened",
"sentAt": "2026-11-04T14:03:00+00:00",
"delivery": "0b6c1c5e-7f1e-4c52-9a0e-1d2f3a4b5c6d",
"site": "Link Robins",
"data": {
"monitor": {
"id": 12,
"name": "My forum",
"url": "https://forum.example.com",
"type": "http",
"status": "offline",
"checkedAt": "2026-11-04T14:02:41+00:00"
},
"incidentId": 345,
"cause": "The site answered with a server error (502 Bad Gateway).",
"statusCode": 502,
"startedAt": "2026-11-04T14:02:41+00:00"
}
}
incident.resolved carries the same monitor object, plus incidentIds, resolvedAt and downSeconds. Every event has data.monitor, so you always know which monitor it is about. delivery is unique per delivery, which makes it a good key for ignoring duplicates.
These headers come with it:
X-LR-Event: the event name
X-LR-Delivery: the same id as delivery in the body
X-LR-Timestamp: when it was signed, in Unix seconds
X-LR-Signature: sha256= followed by the signature
Maintenance, Forum starts, Login check, Lighthouse and Site crawl events
Each of these carries data.monitor, like every event. monitor.maintenance_started is sent once per monitor in the window, and monitor.maintenance_ended when the window ends or you end it early:
{
"event": "monitor.maintenance_started",
"data": {
"monitor": { "id": 12, "name": "My forum", "url": "https://forum.example.com", "type": "http", "status": "online", "checkedAt": "2026-11-12T14:00:41+00:00" },
"windowId": 7,
"startsAt": "2026-11-12T15:00:00+00:00",
"endsAt": "2026-11-12T16:00:00+00:00",
"note": "Updating Flarum"
}
}
monitor.maintenance_ended has the same fields plus endedEarly (true if you used End now); its endsAt is when the window really ended. note is null when you didn't write one.
forum.start_failed adds an error object, and forum.start_recovered carries only monitor:
{
"event": "forum.start_failed",
"data": {
"monitor": { "id": 12, "name": "My forum", "url": "https://forum.example.com", "type": "http", "status": "online", "checkedAt": "2026-11-12T14:00:41+00:00" },
"error": {
"message": "Uncaught TypeError: Cannot read properties of undefined",
"file": "https://forum.example.com/assets/extensions/vendor-ext/forum.js",
"extension": "vendor-ext",
"kind": "exception"
}
}
}
file and extension are null when the error doesn't show them. kind is a short label for the type of failure.
login.failed adds an error object and paused. paused is false when the check ran and signing in failed, and true when the check stopped itself (after repeated refused passwords, or because the sign-in asks for a captcha or a second factor). login.recovered carries only monitor:
{
"event": "login.failed",
"data": {
"monitor": { "id": 12, "name": "My forum", "url": "https://forum.example.com", "type": "http", "status": "online", "checkedAt": "2026-11-13T14:00:41+00:00" },
"paused": false,
"error": {
"reason": "session_lost",
"message": "Signing in appeared to work, but the forum did not keep the session.",
"extension": null
}
}
}
reason is a short label for why it failed, message says it in plain English, and extension is null unless the error shows which extension was involved. The test account's password never appears in a webhook. A login failure never opens an incident or changes your status page.
lighthouse.score_dropped is sent once for each score that dropped. formFactor is mobile or desktop, category is performance, accessibility, best_practices or seo, from is the score's recent level (the middle of the seven reports before) and to is today's score:
{
"event": "lighthouse.score_dropped",
"data": {
"monitor": { "id": 12, "name": "My forum", "url": "https://forum.example.com", "type": "http", "status": "online", "checkedAt": "2026-11-13T14:00:41+00:00" },
"formFactor": "mobile",
"category": "performance",
"from": 62,
"to": 48
}
}
crawl.new_broken_links is sent once per crawl that finds new broken pages or links. counts has every finding type from that crawl, newCount is how many broken pages and links are new since the last crawl, and newIssues lists up to 50 of them:
{
"event": "crawl.new_broken_links",
"data": {
"monitor": { "id": 12, "name": "My forum", "url": "https://forum.example.com", "type": "http", "status": "online", "checkedAt": "2026-11-13T14:00:41+00:00" },
"pagesCrawled": 312,
"counts": { "broken_link": 3, "mixed_content": 5, "missing_description": 40 },
"newCount": 2,
"newIssues": [
{ "type": "broken_link", "url": "https://example.org/old-page", "status": 404, "foundOn": ["https://forum.example.com/d/24-some-discussion"] }
]
}
}
Check the signature
The signature is an HMAC-SHA256, using your signing secret, of the timestamp, a dot and the exact body that was sent. Recompute it from the raw body (not from JSON you parsed and re-encoded), compare, and ignore requests whose timestamp is more than a few minutes old.
import crypto from 'node:crypto';
function isFromLinkRobins(secret, timestamp, rawBody, header) {
const mac = crypto.createHmac('sha256', secret).update(timestamp + '.' + rawBody).digest('hex');
const expected = Buffer.from('sha256=' + mac);
const given = Buffer.from(header);
return expected.length === given.length && crypto.timingSafeEqual(expected, given);
}
What counts as delivered
Answer with any 2xx status within 10 seconds. Anything else, including a timeout, is shown as a failed delivery in the webhook list. Redirects are not followed, and the address must be on the public internet; private and internal addresses are refused.