collect mcp server names

Author: nithinjamespallicka

src/datu/app_config.py

@@ -57,6 +57,13 @@ class DatuConfig(BaseSettings):
         schema_sample_limit (int): The maximum number of rows to sample from the schema.
         schema_categorical_threshold (int): The threshold for categorical columns in the schema.
         enable_schema_rag (bool): Enable RAG for schema extraction.
+        enable_anonymized_telemetry (bool): Enable anonymized telemetry. Default is True.
+        app_environment (str): The application environment (e.g., "dev", "test", "prod"). Default is "dev".
+        telemetry (TelemetryConfig | None): Configuration settings for telemetry.
+        enable_mcp (bool): Whether to enable MCP integration. Default is False.
+        mcp (MCPConfig | None): Configuration settings for MCP integration.
+        enable_schema_rag (bool): Enable RAG for schema extraction.
+        schema_rag (SchemaRAGConfig | None): Configuration settings for schema RAG.
 
     Attributes:
         host (str): The host address for the application.
@@ -78,7 +85,8 @@ class DatuConfig(BaseSettings):
         mcp (MCPConfig | None): Configuration settings for MCP integration.
         enable_schema_rag (bool): Enable RAG for schema extraction.
         schema_rag (SchemaRAGConfig | None): Configuration settings for schema RAG.
-
+        enable_anonymized_telemetry (bool): Enable anonymized telemetry.
+        telemetry (TelemetryConfig | None): Configuration settings for telemetry.
 
     """
 

src/datu/factory/llm_client_factory.py

@@ -6,6 +6,8 @@
 from typing import Literal
 
 from datu.llm_clients.openai_client import OpenAIClient
+from datu.telemetry.product.events import MCPClientEvent
+from datu.telemetry.product.posthog import get_posthog_client
 
 
 def get_llm_client(provider: Literal["openai"] | None = None) -> OpenAIClient | None:
@@ -20,6 +22,15 @@ def get_llm_client(provider: Literal["openai"] | None = None) -> OpenAIClient |
         ValueError: If the specified provider is not supported.
     """
     if provider == "openai":
-        return OpenAIClient()
+        openai_client = OpenAIClient()
+        if openai_client.agent:
+            posthog_client = get_posthog_client()
+            if openai_client.mcp_client and getattr(openai_client.mcp_client, "config", None):
+                servers = openai_client.mcp_client.config.get("mcpServers", {})
+                server_names = list(servers.keys()) if servers else []
+            else:
+                server_names = []
+            posthog_client.capture(MCPClientEvent(server_names=server_names))
+        return openai_client
     else:
         raise ValueError("Invalid LLM provider specified in configuration.")

src/datu/main.py

@@ -13,7 +13,7 @@
 from datu.app_config import get_logger, settings
 from datu.routers import chat, metadata, transformations
 from datu.schema_extractor.schema_cache import load_schema_cache
-from datu.telemetry.product.events import OpenAIEvents
+from datu.telemetry.product.events import OpenAIEvent
 from datu.telemetry.product.posthog import get_posthog_client
 
 logger = get_logger(__name__)
@@ -59,7 +59,7 @@ def start_app() -> None:
     """
     logger.info("Starting the FastAPI application...")
     posthog_client = get_posthog_client()
-    posthog_client.capture(OpenAIEvents())
+    posthog_client.capture(OpenAIEvent())
     uvicorn.run(app, host=settings.host, port=settings.port)
 
 

src/datu/telemetry/product/events.py

@@ -1,6 +1,6 @@
 from typing import Any, ClassVar, Dict
 
-from datu.app_config import get_app_settings, get_logger
+from datu.app_config import get_app_settings
 
 app_settings = get_app_settings()
 
@@ -34,15 +34,15 @@ def batch(self, other: "ProductTelemetryEvent") -> "ProductTelemetryEvent":
         return self
 
 
-class MCPClientStartEvent(ProductTelemetryEvent):
+class MCPClientEvent(ProductTelemetryEvent):
     """Event for when the MCP client starts."""
 
-    def __init__(self, **kwargs):
-        super().__init__(**kwargs)
-        self._props["mcp_client_enabled"] = app_settings.mcp_client_enabled
+    def __init__(self, server_names: list[str]):
+        super().__init__()
+        self._props["mcp_server_names"] = server_names
 
 
-class OpenAIEvents(ProductTelemetryEvent):
+class OpenAIEvent(ProductTelemetryEvent):
     """Event for OpenAI-related telemetry."""
 
     def __init__(self, **kwargs):

src/datu/telemetry/product/posthog.py

@@ -9,7 +9,7 @@
 
 import posthog
 
-from datu.app_config import get_app_settings, get_logger
+from datu.app_config import Environment, get_app_settings, get_logger
 from datu.telemetry.config import TelemetryConfig as TelemetrySettings
 from datu.telemetry.product.events import ProductTelemetryEvent
 
@@ -30,8 +30,13 @@ def __init__(self, settings: Optional[TelemetrySettings]) -> None:
         self._batched_events: Dict[str, ProductTelemetryEvent] = {}
         self._user_id: str = ""
         self._user_id_path: Path = self.USER_ID_PATH
+        self.session_id = str(uuid.uuid4())
 
-        if not app_settings.enable_anonymized_telemetry or "pytest" in sys.modules:
+        if (
+            not app_settings.enable_anonymized_telemetry
+            or "pytest" in sys.modules
+            or app_settings.app_environment in [Environment.TEST.value]
+        ):
             posthog.disabled = True
         else:
             logger.info("Enabled anonymized telemetry. See https://docs.datu.fi for more information.")
@@ -63,11 +68,21 @@ def _base_context(self) -> Dict[str, Any]:
         except importlib.metadata.PackageNotFoundError:
             pkg_version = "unknown"
 
+        extras_installed: Dict[str, bool] = {}
+        try:
+            dist = importlib.metadata.distribution(self.settings.package_name)
+            extras = dist.metadata.get_all("Provides-Extra") or []
+            for extra in extras:
+                extras_installed[extra] = True
+        except importlib.metadata.PackageNotFoundError:
+            extras_installed = {}
+
         return {
             "python_version": sys.version.split()[0],
             "os": platform.system(),
             "os_version": platform.release(),
             "package_version": pkg_version,
+            "extras_installed": extras_installed,
         }
 
     def capture(self, event: ProductTelemetryEvent) -> None: