Webhook Deliveries¶
The Webhook Deliveries page in the self-service portal shows a short-lived, inspectable record of every webhook delivery for your ATS — both the outgoing webhooks CarvOS sends to your endpoint and the incoming webhooks your ATS sends to CarvOS. Use it to diagnose delivery problems without reproducing them.
You only ever see your own ATS's records; one ATS can never see another's.
Opening the page¶
Sign in to the Self-Service Portal and choose Webhook Deliveries from the sidebar. Records are listed newest first.
Filtering¶
Three filters narrow the list:
- Window — the last hour, 6 hours, 24 hours, or the full retention window (the default; the option states how many days that is). Every view covers a time window; there is no unbounded one.
- Direction — outbound (CarvOS → your endpoint) or inbound (your ATS → CarvOS).
- Outcome — success, failure, or error.
Leave Direction or Outcome on All to remove it. Filters carry across pagination, so paging never widens the window you selected.
Counts¶
Above the table is the number of stored records in the selected window — the total, and the split across success, failure and error. The counts cover the whole window, not the page you are looking at, so a page of rows out of a much larger window says so. They ignore the Outcome filter, so narrowing the list to failures still shows how many succeeded.
Timestamps¶
Every timestamp on the page is UTC and is labelled as such, in both the table and the detail dialog — so a delivery time can be quoted in a support request without converting it first.
What the page does and does not show¶
The page is an audit log of individual deliveries, not a traffic chart. It holds one row per verified incoming webhook and per outgoing send. It does not show your REST API calls, or incoming webhooks that failed signature verification — those are rejected before a record is written. Records are deleted after 7 days, and the page never reaches further back than that.
Reloading it and seeing few rows means few records in the selected window; it is not a measurement of your traffic.
Inspecting a delivery¶
Select View on any row to open the detail dialog. It shows the request payload, the response body and status (for outgoing HTTP deliveries), the allowlisted request/response headers, timing, and the correlation id you can copy into a support request.
Replaying a failed delivery¶
Open the detail dialog for a failed or errored outgoing delivery and choose
Replay to have CarvOS re-send it. The replay uses the delivery's stored
payload and your ATS's current webhook URL and signing secret, so it respects
any endpoint or secret changes you have made since the original attempt. The
payload is re-sent unchanged — its reference_id and event_id are preserved, so
your endpoint can dedupe it against the original; only the signature is
regenerated. The dialog then shows the new attempt's outcome.
Each record can be replayed once. If that replay also fails, open it and replay it — retries chain one record to the next, up to a small fixed limit. Replay is not available for successful deliveries, inbound records, or records whose stored payload was truncated.
Best-effort audit¶
Delivery records are a best-effort audit log. Records may be missing during bursts or platform incidents; the webhook delivery itself is unaffected. Do not treat the delivery log as a billing-grade or compliance-grade ledger — the count above the table is a floor, not a total.
Related¶
- Webhook Delivery Log — exactly what is retained, for how long, and who can access it.
- Webhook Testing

