Skip to main content

Call recording upload from the media sidecar (server-to-server)

POST 
/api/webrtc/calls/:callId/recording

NOT a panel endpoint. Our own media sidecar multipart-POSTs the finished recording here when a call ends. No session, no Sanctum, no CSRF — trust is the shared bridge token, and the tenant comes from the opaque context the sidecar echoes back rather than from the call id, so no pre-tenant lookup is needed.

Two audio legs may arrive: the customer's and the agent's. Their offsets are what let the transcript be reassembled as a dialogue instead of two monologues, so a leg without an offset is still ingested but cannot be interleaved.

Request​

Responses​

Recording ingested.