decomposer: fix validation failure for Migrate the kab_ingestion domain models, ports, GitHub SCM connector, and SharePoint CMS connector into the repository's app/ package.; Wire the migrated ingestion components into the agent application entry point, card, settings, triggers, governance, publication, and validation layers.; Populate every remaining placeholder file with working application, package, test, dependency, and Kubernetes content.; Verify the migrated agent and completed repository end to end.

app/connectors/__init__.py

@@ -0,0 +1,4 @@
+from .github import GitHubConnector
+from .sharepoint import SharePointConnector
+
+__all__ = ["GitHubConnector", "SharePointConnector"]

app/connectors/github.py

@@ -0,0 +1,16 @@
+from pathlib import PurePosixPath
+from typing import Sequence
+from ..domain.models import Document, IngestionRequest
+
+class GitHubConnector:
+    """Small GitHub SCM adapter; transport is injected for testability."""
+    def __init__(self, token: str = "", transport=None):
+        self.token, self.transport = token, transport
+
+    def fetch(self, request: IngestionRequest) -> Sequence[Document]:
+        if self.transport is None:
+            return []
+        payload = self.transport(request.source.repository, request.source.path, request.source.revision, self.token)
+        if isinstance(payload, str):
+            payload = {"content": payload}
+        return [Document(request.source, payload.get("content", ""), {"provider": "github", "path": str(PurePosixPath(request.source.path))})]

app/connectors/sharepoint.py

@@ -0,0 +1,11 @@
+from ..domain.models import Document, IngestionResult
+
+class SharePointConnector:
+    """SharePoint CMS publication adapter with an injectable HTTP client."""
+    def __init__(self, site_url: str, client=None):
+        self.site_url, self.client = site_url.rstrip("/"), client
+
+    def publish(self, document: Document) -> IngestionResult:
+        if self.client is not None:
+            self.client.create_page(self.site_url, document.source.path, document.content, document.metadata)
+        return IngestionResult(True, document, "published")