simulation: --simulation flag disables the worker load limit; text sims skip STT/TTS

livekit-agents/livekit/agents/__main__.py

@@ -72,6 +72,7 @@ def _dispatch(server: AgentServer, args: argparse.Namespace) -> None:
                 reload_addr=args.reload_addr,
                 log_format=args.log_format,
                 dev=args.dev,
+                simulation=args.simulation,
             ),
         )
 
@@ -111,6 +112,9 @@ def main(argv: list[str] | None = None) -> int:
     start_p.add_argument("--api-secret")
     start_p.add_argument("--dev", action="store_true", default=False)
     start_p.add_argument("--reload-addr")
+    # set by `lk simulate`: disables the worker load limit so simulation runs
+    # can saturate the agent
+    start_p.add_argument("--simulation", action="store_true", default=False)
 
     console_p = sub.add_parser("console")
     console_p.add_argument("entrypoint", nargs="?")

livekit-agents/livekit/agents/cli/_legacy.py

@@ -1709,6 +1709,14 @@ def start(
                 help="Time in seconds to wait for jobs to finish before shutting down.",
             ),
         ] = None,
+        simulation: Annotated[
+            bool,
+            typer.Option(
+                hidden=True,
+                help="Run under an agent simulation: the worker load limit is disabled "
+                "so runs can saturate the agent. Set by `lk simulate`.",
+            ),
+        ] = False,
     ) -> None:
         if drain_timeout is not None:
             server.update_options(drain_timeout=drain_timeout)
@@ -1720,6 +1728,7 @@ def start(
                 url=url,
                 api_key=api_key,
                 api_secret=api_secret,
+                simulation=simulation,
             ),
         )
 

livekit-agents/livekit/agents/cli/cli.py

@@ -285,6 +285,9 @@ def _run_worker(server: AgentServer, args: proto.CliArgs) -> None:
     if kwargs:
         server.update_options(**kwargs)
 
+    if args.simulation:
+        server._simulation = True
+
     if args.reload_addr and not args.dev:
         raise ValueError("--reload-addr requires --dev")
 

livekit-agents/livekit/agents/cli/proto.py

@@ -17,6 +17,9 @@ class CliArgs:
     reload_addr: str | None = None
     log_format: str = "json"
     dev: bool = False
+    # set by `lk simulate` when launching the agent under test; disables the
+    # worker load limit so simulation runs can saturate the agent
+    simulation: bool = False
 
 
 def running_job_to_proto(info: RunningJobInfo) -> agent_dev.RunningAgentJobInfo:

livekit-agents/livekit/agents/job.py

@@ -467,7 +467,10 @@ def simulation_context(self) -> SimulationContext | None:
         from .simulation import SimulationContext
 
         try:
-            dispatch = json_format.Parse(metadata, sim_pb.SimulationDispatch())
+            # ignore unknown fields so dispatches from newer servers still parse
+            dispatch = json_format.Parse(
+                metadata, sim_pb.SimulationDispatch(), ignore_unknown_fields=True
+            )
         except json_format.ParseError:
             return None
 

livekit-agents/livekit/agents/simulation.py

@@ -14,6 +14,7 @@
 ScenarioGroup = proto.ScenarioGroup
 SimulationRun = proto.SimulationRun
 SimulationDispatch = proto.SimulationDispatch
+SimulationMode = proto.SimulationMode
 
 # Decoded form of a Scenario's `userdata` (arbitrary JSON). On the wire it is a
 # JSON-encoded string; in a scenarios.yaml it is written as a nested mapping.
@@ -24,6 +25,7 @@
     "ScenarioGroup",
     "SimulationRun",
     "SimulationDispatch",
+    "SimulationMode",
     "ScenarioUserdata",
     "SimulationVerdict",
     "SimulationContext",
@@ -66,11 +68,20 @@ def scenario(self) -> proto.Scenario:
         return self._scenario
 
     @property
-    def run(self) -> proto.SimulationRun | None:
+    def simulation_mode(self) -> int:
+        """How the simulated user interacts with the agent (text chat or audio).
+        Unspecified is treated as text, since simulations predating the field
+        were all text-only."""
+        if self._dispatch.mode == proto.SimulationMode.SIMULATION_MODE_UNSPECIFIED:
+            return proto.SimulationMode.SIMULATION_MODE_TEXT
+        return self._dispatch.mode
+
+    @property
+    def simulation_run(self) -> proto.SimulationRun | None:
         return self._run
 
     @property
-    def job(self) -> proto.SimulationRun.Job | None:
+    def simulation_job(self) -> proto.SimulationRun.Job | None:
         return self._job
 
     @property

livekit-agents/livekit/agents/voice/agent_activity.py

@@ -3893,8 +3893,15 @@ def _init_metrics_from_end_of_turn(self, info: _EndOfTurnInfo) -> llm.MetricsRep
         return metrics_report
 
     # move them to the end to avoid shadowing the same named modules for mypy
+    @property
+    def _text_only(self) -> bool:
+        # text simulations run without audio: no STT/TTS/VAD
+        return self._session._text_only
+
     @property
     def vad(self) -> vad.VAD | None:
+        if self._text_only:
+            return None
         return self._agent.vad if is_given(self._agent.vad) else self._session.vad
 
     def _resolve_interruption_detection(self) -> inference.AdaptiveInterruptionDetector | None:
@@ -3951,6 +3958,8 @@ def _resolve_interruption_detection(self) -> inference.AdaptiveInterruptionDetec
 
     @property
     def stt(self) -> stt.STT | None:
+        if self._text_only:
+            return None
         return self._agent.stt if is_given(self._agent.stt) else self._session.stt
 
     @property
@@ -3959,4 +3968,6 @@ def llm(self) -> llm.LLM | llm.RealtimeModel | None:
 
     @property
     def tts(self) -> tts.TTS | None:
+        if self._text_only:
+            return None
         return self._agent.tts if is_given(self._agent.tts) else self._session.tts

livekit-agents/livekit/agents/voice/agent_session.py

@@ -688,6 +688,12 @@ async def start(
 
                 job_ctx.init_recording(self._recording_options)
 
+            # Under a text simulation the simulated user interacts over text
+            # streams only: disable audio I/O here, and STT/TTS/VAD via
+            # AgentActivity (both consult _text_only).
+            if self._text_only:
+                logger.info("text simulation: disabling STT/TTS/VAD and audio I/O")
+
             self._session_span = current_span = tracer.start_span("agent_session")
             # we detach here to avoid context issues since tokens need to be detached
             # in the same context as it was created
@@ -738,6 +744,10 @@ async def start(
                 )
                 room_options = copy.copy(room_options)  # shadow copy is enough
 
+                if self._text_only:
+                    room_options.audio_input = False
+                    room_options.audio_output = False
+
                 if self.input.audio is not None:
                     if room_options.audio_input:
                         logger.warning(
@@ -1722,6 +1732,20 @@ def _config_update_added(self, item: llm.AgentConfigUpdate) -> None:
         self._chat_ctx.insert(item)
 
     # move them to the end to avoid shadowing the same named modules for mypy
+    @property
+    def _text_only(self) -> bool:
+        """True when running under a text simulation: the session uses no audio
+        I/O and no audio models (STT/TTS/VAD)."""
+        from ..job import get_job_context
+
+        job_ctx = get_job_context(required=False)
+        if job_ctx is None or (sim_ctx := job_ctx.simulation_context()) is None:
+            return False
+
+        from ..simulation import SimulationMode
+
+        return sim_ctx.simulation_mode == SimulationMode.SIMULATION_MODE_TEXT
+
     @property
     def stt(self) -> stt.STT | None:
         return self._stt

livekit-agents/livekit/agents/worker.py

@@ -359,6 +359,9 @@ def __init__(
 
         self._http_proxy = http_proxy
         self._log_level = _validate_and_normalize_log_level(log_level)
+        # Set by the CLI (--simulation) when the worker runs under an agent
+        # simulation: load shedding is disabled so runs can saturate the agent.
+        self._simulation = False
         self._agent_name = ""
         self._server_type = ServerType.ROOM
         self._id = "unregistered"
@@ -584,6 +587,9 @@ async def run(self, *, devmode: bool = False, unregistered: bool = False) -> Non
                     )
                     self._load_threshold = _default_load_threshold
 
+            if self._simulation:
+                logger.info("simulation mode enabled: worker load limit disabled")
+
             self._loop = asyncio.get_event_loop()
             self._devmode = devmode
             self._job_lifecycle_tasks = set[asyncio.Task[Any]]()
@@ -1017,7 +1023,7 @@ async def aclose(self) -> None:
                 await self._prometheus_server.aclose()
 
             if self._api is not None:
-                await self._api.aclose()  # type: ignore[no-untyped-call]
+                await self._api.aclose()  # type: ignore[no-untyped-call, unused-ignore]
 
             # await asyncio.sleep(0.25)  # see https://github.com/aio-libs/aiohttp/issues/1925
             self._msg_chan.close()
@@ -1296,6 +1302,9 @@ def _is_available(self) -> bool:
         if self._draining:
             return False
 
+        if self._simulation:
+            return True
+
         load_threshold = ServerEnvOption.getvalue(self._load_threshold, self._devmode)
         if math.isinf(load_threshold):
             return True
@@ -1455,7 +1464,7 @@ async def _update_worker_status(self) -> None:
 
         load_threshold = ServerEnvOption.getvalue(self._load_threshold, self._devmode)
         effective_load = self._get_effective_load()
-        is_full = effective_load >= load_threshold
+        is_full = not self._simulation and effective_load >= load_threshold
         currently_available = not is_full and not self._draining
 
         status = (

livekit-agents/pyproject.toml

@@ -29,7 +29,7 @@ dependencies = [
     "certifi>=2025.6.15",
     "livekit==1.1.8",
     "livekit-api>=1.0.7,<2",
-    "livekit-protocol>=1.1.14,<2",
+    "livekit-protocol>=1.1.15,<2",
     "livekit-blingfire~=1.1,<2",
     "protobuf>=3",
     "pyjwt>=2.0",