docs(env): Windows commands' descendants survive a terminated daemon

The job object uses silent breakaway, as libuv does, so it only kills each
command's own process.

packages/env/daemon/src/sys/windows.rs

@@ -550,7 +550,8 @@ pub fn after_spawn(child: &Child) {
     }
 }
 
-/// Nothing to do: the daemon cannot catch `TerminateProcess`, and the job object kills its commands with it.
+/// Nothing to do: the daemon cannot catch `TerminateProcess`. When it ends, the job object kills each command's own
+/// process; processes those start leave the job (silent breakaway, as in libuv) and keep running.
 pub fn on_exit_signal(_on_signal: impl FnOnce(i32) + Send + 'static) {}
 
 /// Node's `killProcessTree` on Windows: `taskkill /F /T`, not waited for.

packages/env/docs/protocol.md

@@ -62,10 +62,11 @@ command as a full pipe would.
 ## Liveness
 
 Each side sends `ping` every 5 seconds. The daemon kills every process group it started and exits after 30 seconds
-without receiving any bytes, when stdin ends, or on `SIGTERM`, `SIGINT` or `SIGHUP` (on Windows, a job object kills
-its commands with it). A dropped connection or a stopped daemon therefore stops remote commands. The client
-fails requests of a lost connection with `{ code: "unknown", lost: true }` and starts a new daemon on the next request;
-handles belong to the daemon that opened them, so requests on them fail instead of reaching the new one.
+without receiving any bytes, when stdin ends, or on `SIGTERM`, `SIGINT` or `SIGHUP`. A dropped connection or a
+stopped daemon therefore stops remote commands. On Windows, a terminated daemon cannot clean up: a job object kills
+each command's own process, but not processes that one started. The client fails requests of a lost connection with
+`{ code: "unknown", lost: true }` and starts a new daemon on the next request; handles belong to the daemon that opened
+them, so requests on them fail instead of reaching the new one.
 
 ## Operations