Skip to content

Deploy and operate

Deploying a hook returns its ingest URL. After that you watch events, replay them after a fix, get an email when a delivery fails for good, and delete hooks you no longer need.

Deploy

Deploying returns your ingest URL. Paste it into the sender's webhook settings and you are live.

terminal
hooksmith deploy stripe-failed-payments
# https://hooksmith.tech/h/<id>

Operate

Watch events from the terminal or the dashboard. In the dashboard you can filter and search events, read the payload and each destination's request and response, replay an event after a fix, and send a test event through a hook.

The hook's page lists each destination's retry delays, or the platform default (30s → 24h) when the YAML has no retries block. Past the monthly allowance, the usage page shows the overage count.

Deleting a hook. Workspace owners delete a hook from its page (Danger zone), or with hooksmith delete <name> --remote. The ingest URL stops accepting events at once, and pending retries are cancelled and not sent. Past events stay in Logs, marked as deleted, until your plan's retention removes them. A deleted hook does not count toward your plan's hook limit. Deploying a hook with the same name again brings it back at the same URL. Members cannot delete hooks.

terminal
hooksmith logs                      # recent events
hooksmith events stripe-failed-payments
hooksmith replay <event-id>
hooksmith open stripe-failed-payments  # dashboard URL

Failure alerts

Hooksmith emails the workspace when a delivery ends in failed, after its retries are used up. Intermediate failed attempts do not send anything.

  • At most one failure email per hook in each cooldown window. The window is 60 minutes by default. Self-hosters can change it with HOOKSMITH_ALERT_COOLDOWN_MINUTES. Failures inside the window are counted, and the next email says how many.
  • One recovery email when a delivery succeeds after a failure email. It is sent once per incident.
  • Recipients are the workspace owners by default. Owners are looked up when an email is sent, so a new owner is covered without changing anything. Or list up to 5 addresses, for example an on-call alias.
  • Alerts are on until an owner turns them off. When they are off, nothing is sent.

Each email names the hook, the destination and the last error, such as HTTP 500 or a timeout, with any URL cut down to its origin. It links to the event and the hook in the dashboard. It never includes a destination URL, because those can carry secrets in the path or query.

Owners change the settings and send a test alert from the Settings page. Members can read the settings. A test alert goes out even when alerts are turned off.