Where it is raised
POST /api/v1/notification-channelsPOST /api/v1/notification-channels/{id}/testPATCH /api/v1/notification-channels/{id}DELETE /api/v1/notification-channels/{id}
- Keyed MCP endpoint, channel tools
What triggers it
The route sends the type and config to that channel type's own formatter, and whatever reason it returns becomes the message. An unknown type, a config key the type needs, a non-https or private endpoint, or an owner email that cannot be removed all land here. Duplicate names and plan ceilings are lifted out to RU-3009 and RU-2007.
Example response
POST /api/v1/notification-channels
HTTP 400
{
"error": "That notification channel type is not available.",
"code": "RU-3007"
}Common causes
- A `type` outside discord, msteams, opsgenie, incidentio, firehydrant, generic_json (sms, voice and mobile_push are not yet deliverable).
- A `config` missing the key the type needs: webhookUrl (discord, msteams, firehydrant), url (generic_json), apiKey and region (opsgenie), apiKey and alertSourceConfigId (incidentio).
- An endpoint that resolves to a private or internal address.
How to fix it
- Fix the config key the `error` field names for this channel type.
- Compare the config keys against the API docs at /docs/api, "POST /notification-channels".