Bug

monitorMSTeamsProvider() in extensions/msteams/src/monitor.ts returns immediately after starting the Express server. The channel manager in src/gateway/server-channels.ts treats promise resolution as "provider stopped" and triggers auto-restart with exponential backoff, resulting in an infinite restart loop:

[msteams] [default] starting provider (port 3978)
[msteams] [default] auto-restart attempt 1/10 in 5s
[msteams] [default] starting provider (port 3978)
[msteams] [default] auto-restart attempt 2/10 in 10s
...

This cycles through all 10 attempts, then the health-monitor restarts the whole sequence indefinitely.

Root cause

The Slack provider (src/slack/monitor/provider.ts:355-359) correctly blocks until the abort signal fires:

await new Promise<void>((resolve) => {
  opts.abortSignal?.addEventListener("abort", () => resolve(), { once: true });
});

The msteams provider is missing this — it just returns { app, shutdown } after expressApp.listen().

Fix

Add the same blocking pattern before the return in extensions/msteams/src/monitor.ts:

// Block until abort signal fires — the channel manager treats promise
// resolution as "provider stopped" and triggers auto-restart.
await new Promise<void>((resolve) => {
  if (opts.abortSignal?.aborted) {
    resolve();
    return;
  }
  opts.abortSignal?.addEventListener("abort", () => resolve(), { once: true });
});

Also add { once: true } to the existing abort listener to prevent leaks:

if (opts.abortSignal) {
  opts.abortSignal.addEventListener("abort", () => {
    void shutdown();
  }, { once: true });
}