Microsoft Teams Bridge for OpenClaw

Welcome! @komaa/openclaw-msteams-bridge (npm: @komaa/openclaw-msteams-bridge, OpenClaw channel id msteams-voice) adds Microsoft Teams voice and video to an OpenClaw agent.

It is a single OpenClaw plugin. You install it into your OpenClaw gateway, and it runs a small local WebSocket server. The hosted StandIn media bridge joins your Teams call and dials into that server - so your OpenClaw agent can talk, see, and act on a real Teams call without you having to build any Teams media stack yourself. The agent appears in the call as an animated, lip-synced avatar.

What it does

Real-time voice - the caller talks to your agent and hears it reply, with natural barge-in (interrupt mid-sentence). Two engines:
- Realtime - OpenAI or Azure OpenAI speech-to-speech (lowest latency).
- Streaming - STT to agent to TTS, using your OpenClaw-configured providers.
Vision - the agent can look at the caller’s shared screen or camera, both on request (look_at_screen) and ambiently (it stays visually aware during the call), with a per-call spend budget.
Avatar cues - expression hints, viseme lip-sync, and show_to_caller to render an image on the bot’s tile (picture-in-picture or fullscreen).
Meetings - “speak only when addressed” group etiquette with wake phrases and a follow-up window, per-speaker attribution, and an optional end-of-call recap with .docx minutes.
Outbound call-backs - place a call, speak a result or hold a conversation, hang up, with a no-answer/voicemail fallback and cancel-ringing.
DTMF / IVR, bilingual EN/AR, verbal interrupts, echo suppression, and a cutoff goodbye when a StandIn time limit is reached.
Secure transport - a replay-proof HMAC handshake on every connection, a caller allowlist that is closed by default, and a recording-status gate.

How the pieces fit

 Teams call <->  StandIn media bridge  --HMAC WebSocket-->  @komaa/openclaw-msteams-bridge (in the OpenClaw gateway)
   (hosted)       joins the meeting             the local WS server + the call brain
                                                        |
                                                  your OpenClaw agent (real work, tools)

The plugin is the local WebSocket server (default 127.0.0.1:9442; set bindAddress: "0.0.0.0" so the hosted bridge can reach it). StandIn is the client that dials in over an HMAC-authenticated WebSocket using a shared secret both sides hold. You never operate any Teams media yourself - that is StandIn’s job.

Use the sidebar to navigate. Start with Getting Started, or jump to the Configuration Reference or Wire Protocol.

Microsoft Teams Bridge for OpenClaw

What it does

How the pieces fit

Links