A broken integration and a quiet one send you the same signal: nothing at all. Catchook watches every webhook you receive and tells you the moment one stops behaving.
Most outages announce themselves — 500s, a queue backing up, a latency graph bending. Inbound webhooks do not. Here is what breaking looks like from inside your system.
What your system sees
nothing
Stripe, GitHub and Shopify all retire webhook subscriptions that keep failing. Your endpoint 500s through a deploy, the retries fail too, and at some point the provider stops trying — for good. The record of that lives in their dashboard, not yours.
What your system sees
nothing
A field moves inside a nested object, or starts arriving as null. Your handler throws, you return 500, the provider retries a few times and gives up. Your error tracker shows a handful of exceptions and then silence, which reads like a resolved blip.
What your system sees
nothing
Somebody rotated a secret and updated one of the two places it lives. Every request now fails verification and is rejected before it ever reaches your handler. From your side this is indistinguishable from no traffic at all.
What your system sees
200 OK…in 8 seconds
Deliveries still succeed, because something downstream degraded rather than failed. Nothing errors. Nothing alerts. Events queue behind the provider's own concurrency limits and the whole integration runs hours late.
Only the last one shows up in a normal APM tool — and only if you were already watching that endpoint.
The part only we can answer
When a provider goes quiet you cannot tell whether your endpoint broke or theirs did. So you read your own logs, wondering about the deploy you shipped an hour ago. Their status page only tells you what they have already decided to publish.
Not your deploy.
Nothing has arrived from this provider for any account receiving from it — every one of them went quiet inside the same minute.
Inbound events · every account on this provider
Your own traffic going quiet tells you nothing on its own. Nine accounts going quiet in the same minute tells you everything.
The same thing every other part of your infrastructure already has. None of these need you to have suspected a problem first.
When this endpoint last delivered, against what is normal for it.
alertstripe silent 3× longer than its quietest hour this month
The share of deliveries failing over a window, not whether one failed.
alertgithub failure rate 14% over 15m, baseline 0.2%
How long your handler takes to answer, trended per endpoint.
alertshopify p95 4.2s, was 210ms an hour ago
The fields you have been receiving, and when a provider changes them.
alertlinear: issue.assignee arrived null on 38% of events, never before
All of it from day one, in the single plan. Nothing held back for a higher tier.
Slack, email or webhook the moment a signal leaves its baseline — including when nothing is arriving at all.
Catchookapp
stripe has been silent for 2h 11m
expected ~40/hr · last event 12:21 UTC
catchook listen puts live provider webhooks on localhost — no tunnel, no public URL, replay on a keypress.
$ catchook listen --forward-to http://localhost:3000
live4 ev · 3 ✓ · 1 ✗
15:05:12stripe2001ms
15:05:11stripe5001ms×2
15:05:10github✗
Full history with headers and the original bytes, re-sendable to your machine or to the real destinations.
Stripe, GitHub, Shopify and Linear in one place, not four dashboards and four log formats.
One inbound webhook to as many destinations as you want, each with its own retry policy and rate limit.
Told when a provider quietly changes a payload, rather than finding out from a stack trace.
Active probing, so an integration idle all week is still known to be working.
Your releases on the same timeline as your incidents, so the question answers itself.
Genuinely useful — once somebody has told you there is a bug. Every question a log answers, Catchook answers too. Two below, it cannot.
| You want to know | Webhook log tools | Catchook |
|---|---|---|
| A delivery failed | In the log, with a replay button | Same — and every attempt kept, not just the last |
| Deliveries have been failing for an hour | In the log, if you happen to open it | You were alerted at minute two |
| Nothing has arrived since 04:00 | No record of an event that never came | Alerted, against what is normal for that endpoint |
| A provider changed the payload | Visible once you diff two events by hand | Alerted when the shape moved |
| Is it you, or is it the provider? | Your logs only ever show your own traffic | Answered from traffic across every account on that provider |
Nothing is held back for an Enterprise tier. If Catchook does it, your plan does it — you pay for how much you put through it, not for which features you are allowed to use.
Rates are set before launch. Waitlist members see them first.
Priced separately, because each one costs us real infrastructure rather than a feature flag:
Catchook is not open for sign-ups yet. Leave an address and you will hear from us when it is — first, and with the launch pricing.
One email when sign-ups open. Nothing else, ever.
