WebSocket Proxy

Bridge Webex sessions to external voice AI backends without custom protocol glue.

from webex_byova.media import BYOVAMediaServer, WebSocketProxyConnector, DefaultProxyAdapter

server = BYOVAMediaServer()
server.use_proxy(WebSocketProxyConnector(
    url="wss://your-backend.example/ws",
    adapter=DefaultProxyAdapter(),
))

Default JSON protocol

SDK → backend: { "type": "session_start|audio|dtmf|...", "conversation_id", "turn_id", "payload" }

Backend → SDK: { "type": "prompt|end_session", "payload": { "text", "audio" (base64) } }

Custom adapters

Implement ProxyAdapter with to_backend() and from_backend() methods. See contracts/websocket-proxy.md in the feature spec.

Buffer overflow

Default policy disconnects the session when the proxy buffer exceeds proxy_buffer_limit. Set proxy_overflow_policy="drop_oldest" to drop oldest chunks instead.