Skip to main content
The fastest way to understand the bridge is to run its example project - a minimal, working embedding you can copy straight into your own repo. Budget about 10 minutes once the Teams setup is done.

examples/basic-bridge

loadConfig() + startServer() and nothing else - about 20 lines, because the agent’s brain (LLM, tools, conversation logic) lives in your Line deployment on Cartesia, not in the bridge. Env-file config and graceful shutdown are built in.

​
What you need first

  • Node.js >= 20.
  • A Cartesia API key and a deployed Line agent (its agent id) - docs.cartesia.ai/line.
  • A StandIn identity with its shared secret (from pairing or the dashboard). The sandbox works too if you have no Teams bot yet.

​
1. Clone and install

​
2. Configure

Fill in the three required values: Optionally give the call a deterministic opening line (a natural spoken AI disclosure):

​
3. Run and expose

The bridge listens on port 9442. Expose it with a tunnel (Tailscale Funnel, Cloudflare Tunnel, ngrok, dev tunnels - see the installation page) and note the public wss://…/voice/msteams/stream URL.

​
4. Connect to StandIn and call

Set your identity’s Agent calling URL to that wss:// URL in the dashboard (or use the sandbox), then place a Teams call. In the logs you should see:
Speak, and your Line agent answers in its own voice.

​
Where the interesting parts live

There is no vision hook or custom tool to study here - those belong in your Line agent code on Cartesia’s platform. What the bridge forwards to it (caller metadata, call_context and goodbye_request custom events, native DTMF) is documented on the project site:

Your Line Agent - the contract

The start metadata and custom-event shapes your agent code receives, how to detect a Teams call (metadata.from == "msteams"), and the transfer_call limitation.
Before production: set the governor variables (MAX_CALL_MINUTES, CARTESIA_TTS_MODEL + CARTESIA_TTS_VOICE_ID, GOODBYE_TEXT) so time limits end with a spoken goodbye instead of silence.