Merge branch 'development' of github.com:enso-labs/orchestra into development

Author: ryaneggz

Changelog.md

@@ -19,6 +19,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   - bug/370-anthropic-streaming (2025-09-16)
 
 ### Changed
+  - feat/464-update-input-support-multiple-keys (2025-11-11)
   - feat/490-can-read-deepagents-agent-files (2025-11-10)
   - feat/485-user-env (2025-11-08)
   - feat/482-teams-webhook (2025-11-05)

backend/src/constants/llm.py

@@ -1,58 +1,97 @@
-from enum import Enum
 import os
+from enum import Enum
+from src.services.llm import llm_service
 from src.constants import (
-    OPENAI_API_KEY,
-    ANTHROPIC_API_KEY,
-    OLLAMA_BASE_URL,
-    GROQ_API_KEY,
-    GOOGLE_API_KEY,
-    XAI_API_KEY,
+	OPENAI_API_KEY,
+	ANTHROPIC_API_KEY,
+	OLLAMA_BASE_URL,
+	GROQ_API_KEY,
+	GOOGLE_API_KEY,
+	XAI_API_KEY,
 )
+from src.utils.logger import logger
 
 
 class ChatModels(str, Enum):
-    if OPENAI_API_KEY:
-        OPENAI_REASONING_03 = "openai:o3"
-        OPENAI_REASONING_04_MINI = "openai:o4-mini"
-        OPENAI_GPT_5_NANO = "openai:gpt-5-nano"
-        OPENAI_GPT_5_MINI = "openai:gpt-5-mini"
-        OPENAI_GPT_5 = "openai:gpt-5"
-        # OPENAI_GPT_5_CODEX = "openai:gpt-5-codex"
-    if ANTHROPIC_API_KEY:
-        ANTHROPIC_CLAUDE_3_7_SONNET = "anthropic:claude-3-7-sonnet-latest"
-        ANTHROPIC_CLAUDE_4_SONNET = "anthropic:claude-sonnet-4"
-        ANTHROPIC_CLAUDE_4_OPUS = "anthropic:claude-opus-4-1"
-        ANTHROPIC_CLAUDE_4_5_HAIKU = "anthropic:claude-haiku-4-5"
-        ANTHROPIC_CLAUDE_4_5_SONNET = "anthropic:claude-sonnet-4-5"
-    if XAI_API_KEY:
-        XAI_GROK_4 = "xai:grok-4"
-        XAI_GROK_4_FAST = "xai:grok-4-fast"
-        XAI_GROK_4_FAST_NON_REASONING = "xai:grok-4-fast-non-reasoning"
-        XAI_GROK_CODE_FAST_1 = "xai:grok-code-fast-1"
-    if GOOGLE_API_KEY:
-        GOOGLE_GEMINI_2_5_FLASH_LITE = "google_genai:gemini-2.5-flash-lite"
-        GOOGLE_GEMINI_2_5_FLASH = "google_genai:gemini-2.5-flash"
-        GOOGLE_GEMINI_2_5_PRO = "google_genai:gemini-2.5-pro"
-    if GROQ_API_KEY:
-        GROQ_OPENAI_GPT_OSS_120B = "groq:openai/gpt-oss-120b"
-        GROQ_LLAMA_3_3_70B_VERSATILE = "groq:llama-3.3-70b-versatile"
-    if OLLAMA_BASE_URL:
-        OLLAMA_QWEN3 = "ollama:qwen3"
-        
+	if OPENAI_API_KEY:
+		OPENAI_REASONING_03 = "openai:o3"
+		OPENAI_REASONING_04_MINI = "openai:o4-mini"
+		OPENAI_GPT_5_NANO = "openai:gpt-5-nano"
+		OPENAI_GPT_5_MINI = "openai:gpt-5-mini"
+		OPENAI_GPT_5 = "openai:gpt-5"
+		# OPENAI_GPT_5_CODEX = "openai:gpt-5-codex"
+	if ANTHROPIC_API_KEY:
+		ANTHROPIC_CLAUDE_3_7_SONNET = "anthropic:claude-3-7-sonnet-latest"
+		ANTHROPIC_CLAUDE_4_SONNET = "anthropic:claude-sonnet-4"
+		ANTHROPIC_CLAUDE_4_OPUS = "anthropic:claude-opus-4-1"
+		ANTHROPIC_CLAUDE_4_5_HAIKU = "anthropic:claude-haiku-4-5"
+		ANTHROPIC_CLAUDE_4_5_SONNET = "anthropic:claude-sonnet-4-5"
+	if XAI_API_KEY:
+		XAI_GROK_4 = "xai:grok-4"
+		XAI_GROK_4_FAST = "xai:grok-4-fast"
+		XAI_GROK_4_FAST_NON_REASONING = "xai:grok-4-fast-non-reasoning"
+		XAI_GROK_CODE_FAST_1 = "xai:grok-code-fast-1"
+	if GOOGLE_API_KEY:
+		GOOGLE_GEMINI_2_5_FLASH_LITE = "google_genai:gemini-2.5-flash-lite"
+		GOOGLE_GEMINI_2_5_FLASH = "google_genai:gemini-2.5-flash"
+		GOOGLE_GEMINI_2_5_PRO = "google_genai:gemini-2.5-pro"
+	if GROQ_API_KEY:
+		GROQ_OPENAI_GPT_OSS_120B = "groq:openai/gpt-oss-120b"
+		GROQ_LLAMA_3_3_70B_VERSATILE = "groq:llama-3.3-70b-versatile"
+		
+		
+def get_ollama_models():
+	models = []
+	try:
+		import requests
+		response = requests.get(f"{OLLAMA_BASE_URL.rstrip('/')}/api/tags", timeout=3)
+		if response.ok:
+			data = response.json()
+			tags = data.get("models", []) if isinstance(data, dict) else []
+			for tag in tags:
+				model_name = tag.get("name")
+				if model_name:
+					models.append(f"ollama:{model_name}")
+	except Exception as e:
+		logger.error(f"Error getting Ollama models: {e}")
+		pass
+	return models
+		
+def get_all_models():
+	models = []
+	if OPENAI_API_KEY:
+		models.extend(llm_service.model_by_provider(provider="openai"))
+	if ANTHROPIC_API_KEY:
+		models.extend(llm_service.model_by_provider(provider="anthropic"))
+	if GOOGLE_API_KEY:
+		models.extend(llm_service.model_by_provider(provider="google"))
+	if GROQ_API_KEY:
+		models.extend(llm_service.model_by_provider(provider="groq"))
+	if XAI_API_KEY:
+		models.extend(llm_service.model_by_provider(provider="xai"))
+	if OLLAMA_BASE_URL:
+		models.extend(get_ollama_models())
+	return sorted(models)
+
 def get_free_models():
-    return [
-        ChatModels.ANTHROPIC_CLAUDE_4_5_HAIKU.value,
-        ChatModels.OPENAI_GPT_5_NANO.value,
-        ChatModels.GOOGLE_GEMINI_2_5_FLASH_LITE.value,
-        ChatModels.GROQ_OPENAI_GPT_OSS_120B.value,
-        ChatModels.GROQ_LLAMA_3_3_70B_VERSATILE.value,
-    ]
+	models = []
+	if OPENAI_API_KEY:
+		models.append(ChatModels.OPENAI_GPT_5_NANO.value)
+	if ANTHROPIC_API_KEY:
+		models.append(ChatModels.ANTHROPIC_CLAUDE_4_5_HAIKU.value)
+	if GOOGLE_API_KEY:
+		models.append(ChatModels.GOOGLE_GEMINI_2_5_FLASH_LITE.value)
+	if GROQ_API_KEY:
+		models.append(ChatModels.GROQ_OPENAI_GPT_OSS_120B.value)
+	if OLLAMA_BASE_URL:
+		models.extend(get_ollama_models())
+	return sorted(models)
 
 
 def get_system_prompt():
-    path = "src/static/prompts/md"
-    with open(os.path.join(path, "default.md"), "r") as file:
-        return file.read()
+	path = "src/static/prompts/md"
+	with open(os.path.join(path, "default.md"), "r") as file:
+		return file.read()
 
 
 DEFAULT_SYSTEM_PROMPT = get_system_prompt()

backend/src/flows/__init__.py

@@ -12,14 +12,16 @@
 from deepagents import SubAgent, create_deep_agent
 
 
+from src.schemas.models.auth import ProtectedUser
 from src.services.memory import memory_service
 from src.services.tool import tool_service
 from src.tools.memory import MEMORY_TOOLS
-from src.schemas.entities import LLMRequest, LLMStreamRequest
+from src.schemas.entities import LLMRequest
 from src.utils.logger import logger
 from src.utils.format import init_system_prompt
 from src.schemas.contexts import ContextSchema
 from src.schemas.entities.a2a import A2AServers
+from src.utils.middleware import add_ai_message_metadata, pii_middleware
 
 
 async def add_memories_to_system():
@@ -50,7 +52,7 @@ def graph_builder(
     subagents: list[SubAgent] = [],
     prompt: str = "You are a helpful assistant.",
     model: str = "openai:gpt-5-nano",
-    context_schema: Type[Any] | None = None,
+    context_schema: Type[ContextSchema] | None = None,
     checkpointer: BaseCheckpointSaver | None = None,
     store: BaseStore | None = None,
     graph_id: Literal[
@@ -74,6 +76,7 @@ def graph_builder(
         system_prompt=prompt,
         checkpointer=checkpointer,
         context_schema=context_schema,
+        middleware=[add_ai_message_metadata] + pii_middleware(),
         store=store,
     )
     return deep_agent
@@ -96,7 +99,7 @@ async def init_tools(
     return tools
 
 
-async def init_subagents(params: LLMRequest | LLMStreamRequest) -> list[SubAgent]:
+async def init_subagents(params: LLMRequest) -> list[SubAgent]:
     result = []
     for subagent in params.subagents:
         subagent_dict = {
@@ -120,22 +123,28 @@ async def init_memories(system_prompt: str, tools: list[BaseTool]):
     return tools + MEMORY_TOOLS, prompt
 
 
-def init_config(params: LLMRequest | LLMStreamRequest):
-    if params.metadata:
-        return RunnableConfig(
-            configurable=params.metadata.model_dump(),
-            max_concurrency=10,
-            recursion_limit=100,
-        )
-    else:
-        return None
+def init_config(
+    params: LLMRequest,
+    user: ProtectedUser | None = None,
+    max_concurrency: int = 4,
+    recursion_limit: int = 100,
+) -> RunnableConfig:
+    return RunnableConfig(
+        configurable={
+            "user_id": user.id if user else None,
+            "thread_id": params.metadata.thread_id or str(uuid4()),
+            "assistant_id": params.metadata.assistant_id or None,
+        },
+        max_concurrency=max_concurrency,
+        recursion_limit=recursion_limit,
+        metadata={**params.metadata.model_dump()},
+    )
 
 
 ################################################################################
 ### Construct Agent
 ################################################################################
 async def construct_agent(
-    # params: LLMRequest | LLMStreamRequest,
     system_prompt: str,
     tools: list[BaseTool],
     model: BaseChatModel,
@@ -218,10 +227,4 @@ def astream(
     ) -> AsyncGenerator[BaseMessage, None]:
         return self.graph.astream(
             messages, config=config, stream_mode=stream_mode, context=context
-        )
-
-    async def aget_state(self, config: RunnableConfig = None):
-        # if config is None:
-        #     config = self.config
-        state = await self.graph.aget_state(config)
-        return state
+        )

backend/src/routes/v0/assistant.py

@@ -10,7 +10,6 @@
 from src.utils.auth import verify_credentials
 from src.utils.logger import logger
 from src.services.assistant import (
-    assistant_service,
     AssistantSearch,
     Assistant,
     ASSISTANT_EXAMPLES,