feat: add deterministic message interaction stages

backend/services/workflow/agent.py

@@ -2,8 +2,6 @@
 
 from __future__ import annotations
 
-from copy import deepcopy
-
 from models import AssistantConfig
 from pipecat.flows import ContextStrategy, ContextStrategyConfig, NodeConfig
 from pipecat.frames.frames import LLMUpdateSettingsFrame
@@ -108,7 +106,6 @@ class WorkflowAgentStage:
         node_id: str,
         *,
         functions: list,
-        greeting_context_message: dict[str, str] | None,
         leading_messages: list[dict[str, str]] | None = None,
     ) -> NodeConfig:
         data = self._engine.data(node_id)
@@ -119,11 +116,6 @@ class WorkflowAgentStage:
             if data.get("contextPolicy") == "fresh"
             else ContextStrategy.APPEND
         )
-        greeting_messages = (
-            [deepcopy(greeting_context_message)]
-            if strategy == ContextStrategy.RESET and greeting_context_message
-            else []
-        )
         fixed_reply_messages = (
             [{"role": "assistant", "content": entry_speech}]
             if entry_mode == "fixed_speech" and entry_speech
@@ -133,7 +125,6 @@ class WorkflowAgentStage:
             "name": node_id,
             "role_message": self.role_message(node_id),
             "task_messages": [
-                *greeting_messages,
                 *(leading_messages or []),
                 *fixed_reply_messages,
             ],

backend/services/workflow/models.py

@@ -15,6 +15,7 @@ class WorkflowStatus(StrEnum):
     ROUTING = "routing"
     RUNNING_AGENT = "running_agent"
     RUNNING_ACTION = "running_action"
+    RUNNING_MESSAGE = "running_message"
     HANDOFF = "handoff"
     ENDED = "ended"
 

backend/services/workflow/output.py

@@ -5,65 +5,15 @@ from __future__ import annotations
 from typing import Any
 from uuid import uuid4
 
-from pipecat.frames.frames import OutputTransportMessageUrgentFrame, TTSSpeakFrame
+from pipecat.frames.frames import OutputTransportMessageUrgentFrame
 from pipecat.utils.time import time_now_iso8601
 
-from services.brains.base import BrainRuntime
-from services.runtime_variables import DynamicVariableStore
+from services.fixed_speech import FixedSpeechOutput
 
 
-class WorkflowOutput:
+class WorkflowOutput(FixedSpeechOutput):
     """Publish debug events and fixed speech without duplicating persistence."""
 
-    def __init__(
-        self,
-        store: DynamicVariableStore,
-        runtime: BrainRuntime,
-    ) -> None:
-        self._store = store
-        self._runtime = runtime
-        self._client_ready = False
-        self._pending_transcripts: list[dict[str, Any]] = []
-
-    async def mark_client_ready(self) -> None:
-        self._client_ready = True
-        pending = self._pending_transcripts
-        self._pending_transcripts = []
-        for message in pending:
-            await self.emit(message)
-
-    async def speak(
-        self,
-        text: str,
-        *,
-        source: str,
-        node_id: str | None = None,
-    ) -> None:
-        """Record, display and synthesize one Workflow-owned utterance."""
-        content = text.strip()
-        if not content:
-            return
-        self._store.record("agent", content)
-        transcript = {
-            "type": "transcript",
-            "role": "assistant",
-            "content": content,
-            "timestamp": time_now_iso8601(),
-            "source": source,
-            **({"nodeId": node_id} if node_id else {}),
-        }
-        if self._client_ready:
-            await self.emit(transcript)
-        else:
-            self._pending_transcripts.append(transcript)
-
-        track_speech = getattr(self._runtime.call_end, "track_speech", None)
-        if callable(track_speech):
-            track_speech()
-        await self._runtime.queue_frame(
-            TTSSpeakFrame(content, append_to_context=False)
-        )
-
     async def emit_node_active(self, node_id: str | None) -> None:
         if node_id:
             await self.emit({"type": "node-active", "nodeId": node_id})