fix(voice): treat silent agent turns as successful

[rosetta-livekit-bot]

Summary

• treat an empty agent turn as a successful silent response instead of a run_input error
• port the source end-to-end regression test through RemoteSession, SessionHost, and AgentSession
• add a patch changeset for @livekit/agents

Ports livekit/agents#6779.

Source diff coverage

• livekit-agents/livekit/agents/voice/remote_session.py: Adapted to agents/src/voice/remote_session.ts. Removed the equivalent empty-items error guard and ported the source explanation using TypeScript naming and comment style.
• tests/test_run_input_errors.py: Adapted to agents/src/voice/remote_session.test.ts. Ported the source-added silent-turn test into the target's existing colocated remote-session suite, using its paired in-memory transports, JS FakeLLM, and RemoteSession.sendMessage counterpart. The assertion remains that the response item list is empty and no error is raised.
• No source files were not applicable.

Validation

• pnpm test agents (115 files, 1,589 passed, 5 skipped)
• pnpm --filter @livekit/agents build
• pnpm --filter @livekit/agents lint (passes with existing warnings)
• pnpm exec prettier --check agents/src/voice/remote_session.ts agents/src/voice/remote_session.test.ts .changeset/silent-agent-turns.md
• cue-cli text-mode runtime assertion with an intentionally silent dispatched agent: no error event and no assistant conversation item

---

Ported from livekit/agents#6779

Original PR description

Problem

SessionHost turns a run that produced no chat items into an error:

if not items_list and not error:
    error = "agent produced no response items"

The caller sees it as a raised RuntimeError, so the turn — and, in simulations, the whole job — is scored as an agent failure.

An empty turn is a legitimate outcome. An LLM can return a completion with no text and no tool calls, and every item-add site in agent_activity is guarded on non-empty text (if forwarded_text:), so nothing lands in the chat context or in RunResult.events.

Agents that induce this on purpose hit it reliably. A close-the-call tool whose output tells the model it has already said goodbye and should now say nothing gets exactly that:

OUT : "You've already said goodbye - do NOT give another farewell, sign-off,
       or filler. If the caller just asked a real question, answer it in one
       short sentence; otherwise say nothing."
USER: "Okay, thank you."        <- agent emits nothing, turn fails

Agent-side traces for that turn confirm the model, not the framework, produced the silence:

[agent_turn]  dur=0.28s  lk.interrupted=False
  [llm_node]    lk.response.text=''
    [llm_request] gen_ai.usage.output_tokens=1

One output token, empty response text, no tool call, not interrupted — and the transcript up to that point is complete and correct.

Fix

Report the empty item list and let the caller decide what the silence means. Simulation drivers already model a silent turn explicitly, so they can end the call cleanly instead of failing it.

Test

test_run_silent_turn_is_not_an_error drives a real AgentSession whose LLM returns an empty completion through SessionHost → RemoteSession. Without the fix it fails with the exact production error:

RuntimeError: session request run_input failed: agent produced no response items

[changeset-bot]

🦋 Changeset detected

Latest commit: 04ba245

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 39 packages

Name	Type
@livekit/agents	Patch
@livekit/agents-plugin-anam	Patch
@livekit/agents-plugin-anthropic	Patch
@livekit/agents-plugin-assemblyai	Patch
@livekit/agents-plugin-azure	Patch
@livekit/agents-plugin-baseten	Patch
@livekit/agents-plugin-bey	Patch
@livekit/agents-plugin-cartesia	Patch
@livekit/agents-plugin-cerebras	Patch
@livekit/agents-plugin-deepgram	Patch
@livekit/agents-plugin-did	Patch
@livekit/agents-plugin-elevenlabs	Patch
@livekit/agents-plugin-fishaudio	Patch
@livekit/agents-plugin-google	Patch
@livekit/agents-plugin-hedra	Patch
@livekit/agents-plugin-hume	Patch
@livekit/agents-plugin-inworld	Patch
@livekit/agents-plugin-krisp	Patch
@livekit/agents-plugin-lemonslice	Patch
@livekit/agents-plugin-liveavatar	Patch
@livekit/agents-plugin-livekit	Patch
@livekit/agents-plugin-minimax	Patch
@livekit/agents-plugin-mistral	Patch
@livekit/agents-plugin-mistralai	Patch
@livekit/agents-plugin-neuphonic	Patch
@livekit/agents-plugin-openai	Patch
@livekit/agents-plugin-perplexity	Patch
@livekit/agents-plugin-phonic	Patch
@livekit/agents-plugin-protoface	Patch
@livekit/agents-plugin-resemble	Patch
@livekit/agents-plugin-rime	Patch
@livekit/agents-plugin-runway	Patch
@livekit/agents-plugin-sarvam	Patch
@livekit/agents-plugin-silero	Patch
@livekit/agents-plugin-soniox	Patch
@livekit/agents-plugin-tavus	Patch
@livekit/agents-plugins-test	Patch
@livekit/agents-plugin-trugen	Patch
@livekit/agents-plugin-xai	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no bugs or issues to report.