Incidents paging by SMS and phone
On-call pages that text each member, call them a few minutes later unless someone acknowledged, read the alert in your org's voice, and take 1 to acknowledge or 2 to escalate.
Venom Incidents pages through the bell, webhooks and Slack. With a Comms provider, a level of an escalation policy can also text and call the people it tells.
Turning it on
- Add a provider in Settings, Comms (Sinch for both SMS and calls; your own gateway for SMS only).
- Each member who should be paged opens Incidents, On-call, enters their number under Your phone for pages
and types back the 6-digit code Comms texts them (over the API:
PUT /incidents/api/v1/oncall/phonewith{"phone", "tz"}, thenPOST /incidents/api/v1/oncall/phone/verifywith{"code"}). The code lasts 10 minutes and allows 5 tries; a new one can be sent once a minute. Only a verified number is paged. - In Incidents, On-call, Escalation policies, tick SMS and/or Phone call on a level. With both, set how
many minutes the call waits after the text (0 to 60, default 5). Over the API a level is
{"targets": [...], "delayMin": 10, "notify": ["sms", "call"], "callAfterMin": 5}.
What a page does
For each member the level tells who has a verified number, Comms starts one run of an urgent profile:
- the text, at once:
[SEV2] checkout errors spiking, reply 1 to ack or open https://…/incidents/…; - the call, after the wait, unless anyone acknowledged the page: it reads the page's title and detail, then says "press 1 to acknowledge, 2 to escalate".
Replying 1 or ack to the text, or pressing 1 on the call, acknowledges the page the same way the button does: a timeline entry, the bell cleared, an audit entry. 2 escalates to the next level. Acknowledging or resolving the page anywhere (the console, Slack, a reply) stops every member's pending call.
Pages are urgent, so they ignore quiet hours (each time is in the audit log). They never override an opt-out: a member who replied STOP gets no text from that number until they reply START.
The voice
By default the call is read by the provider's text-to-speech. To use your own voice, set TTS voice on the Comms
settings page to <model>/<voice> (for example gpt-4o-mini-tts/alloy) and pick the API key it is billed to. The
provider then fetches each spoken step from your cloud, which makes it through /v1/audio/speech like any other
speech: routed to your voice provider, priced and logged.
Without a provider
Nothing changes: the bell, webhooks and Slack work as before, and the SMS and call options do nothing until a provider exists.