feat(seer): Report a missing SCM integration on repo lookups

src/sentry/seer/agent/tools.py

@@ -25,6 +25,7 @@
 from sentry.api.serializers.models.groupactionlogentry import get_serialized_activity_items
 from sentry.api.utils import MAX_STATS_PERIOD, default_start_end_dates, get_date_range_from_params
 from sentry.constants import ALL_ACCESS_PROJECT_ID, ObjectStatus
+from sentry.integrations.services.integration import integration_service
 from sentry.issues.action_log.types import GroupActionType, GroupActorType
 from sentry.issues.formatting.formatter import Format
 from sentry.issues.formatting.limits import LIMITS_DEFAULT, LIMITS_LOW
@@ -106,6 +107,7 @@
     ProfileFlamegraphSuccessResponse,
     ProjectMembersResponse,
     ReplayMetadataResponse,
+    RepositoryDefinitionFailure,
     RepositoryDefinitionResponse,
     TeamMembersResponse,
     TraceItemEventsResponse,
@@ -949,12 +951,29 @@ def _find_repository(
     return None
 
 
+def _repository_lookup_failure(
+    organization_id: int, supported_providers: list[str]
+) -> RepositoryDefinitionFailure | None:
+    """Why a lookup found nothing, when it's known."""
+    integrations = integration_service.get_integrations(
+        organization_id=organization_id,
+        providers=supported_providers,
+        status=ObjectStatus.ACTIVE,
+        org_integration_status=ObjectStatus.ACTIVE,
+        limit=1,
+    )
+    if not integrations:
+        return RepositoryDefinitionFailure(failure_reason="no_scm_integration")
+    return None
+
+
 def get_repository_definition(
     *,
     organization_id: int,
     repo_full_name: str,
     external_id: str | None = None,
-) -> RepositoryDefinitionResponse | None:
+    report_failure_reason: bool = False,
+) -> RepositoryDefinitionResponse | RepositoryDefinitionFailure | None:
     """
     Look up a repository that the org has access to.
     Returns full RepoDefinition if found and accessible via code mappings, None otherwise.
@@ -968,6 +987,9 @@ def get_repository_definition(
         repo_full_name: Full repository name in format "owner/repo-name" (e.g., "getsentry/seer")
         external_id: Optional external repository ID (e.g., GitHub repo ID). Stable across renames.
                      If provided, this is used for lookup instead of name.
+        report_failure_reason: When nothing is found, say why in the response when it's known,
+                     instead of returning None. Opt-in because callers that don't ask would read
+                     that response as a repository.
 
     Returns:
         dict with RepoDefinition fields if found, None otherwise. Includes external_id
@@ -996,9 +1018,8 @@ def get_repository_definition(
                 "seer.rpc.invalid_repo_name_format",
                 extra={"repo_full_name": repo_full_name},
             )
-            return None
-
-        repo = _find_repository(organization_id, repo_full_name, supported_providers)
+        else:
+            repo = _find_repository(organization_id, repo_full_name, supported_providers)
 
     if not repo:
         logger.info(
@@ -1009,6 +1030,8 @@ def get_repository_definition(
                 "external_id": external_id,
             },
         )
+        if report_failure_reason:
+            return _repository_lookup_failure(organization_id, supported_providers)
         return None
 
     # Use the actual repo name from the database, not the requested name.

src/sentry/seer/sentry_data_models.py

@@ -275,6 +275,10 @@ class RepositoryDefinitionResponse(BaseModel):
     external_id: str | None
 
 
+class RepositoryDefinitionFailure(BaseModel):
+    failure_reason: Literal["no_scm_integration"]
+
+
 class EventFilterKeyEntry(BaseModel):
     type: str
 

tests/sentry/seer/agent/test_tools.py

@@ -57,6 +57,7 @@
     ExecuteTimeseriesQueryErrorResponse,
     ExecuteTimeseriesQuerySuccessResponse,
     IssueDetailsResponse,
+    RepositoryDefinitionFailure,
     RepositoryDefinitionResponse,
 )
 from sentry.services.eventstore.models import Event, GroupEvent
@@ -3065,7 +3066,7 @@ def test_get_repository_definition_success(self) -> None:
             repo_full_name="getsentry/seer",
         )
 
-        assert result is not None
+        assert isinstance(result, RepositoryDefinitionResponse)
         assert result.repository_id == repo.id
         assert result.organization_id == self.organization.id
         assert result.integration_id == "123"
@@ -3136,6 +3137,64 @@ def test_does_not_find_another_organizations_or_an_inactive_repo(self) -> None:
 
         assert not self._finds(self.organization, "my-group/sub/app")
 
+    def _lookup(self, organization: Organization, repo_full_name: str) -> Any:
+        return get_repository_definition(
+            organization_id=organization.id,
+            repo_full_name=repo_full_name,
+            report_failure_reason=True,
+        )
+
+    def _connect(self, organization: Organization, provider: str) -> None:
+        self.create_integration(
+            organization=organization,
+            provider=provider,
+            external_id=f"{provider}:{organization.id}",
+            name=provider,
+        )
+
+    @with_feature("organizations:seer-cursor-origin-support")
+    def test_failure_when_no_integrations(self) -> None:
+        no_integration = RepositoryDefinitionFailure(failure_reason="no_scm_integration")
+        assert self._lookup(self.organization, "acme-labs/app") == no_integration
+        assert self._lookup(self.organization, "invalid-format") == no_integration
+
+        # A provider Seer can't read code from doesn't count.
+        self._connect(self.organization, "bitbucket")
+        assert self._lookup(self.organization, "acme-labs/app") == no_integration
+
+    def test_failure_when_the_only_integrations_are_inactive(self) -> None:
+        self.create_integration(
+            organization=self.organization,
+            provider="github",
+            external_id="github:disabled-for-org",
+            oi_params={"status": ObjectStatus.DISABLED},
+        )
+        self.create_integration(
+            organization=self.organization,
+            provider="gitlab",
+            external_id="gitlab:pending-deletion",
+            status=ObjectStatus.PENDING_DELETION,
+        )
+
+        assert self._lookup(self.organization, "acme-labs/app") == RepositoryDefinitionFailure(
+            failure_reason="no_scm_integration"
+        )
+
+    def test_cursor_origin_does_not_count_without_seer_support(self) -> None:
+        self._connect(self.organization, "cursor_origin")
+
+        assert self._lookup(self.organization, "acme-labs/app") == RepositoryDefinitionFailure(
+            failure_reason="no_scm_integration"
+        )
+
+    @with_feature("organizations:seer-cursor-origin-support")
+    def test_no_failure_reason_when_source_control_is_connected(self) -> None:
+        for provider in ("github", "github_enterprise", "gitlab", "cursor_origin"):
+            organization = self.create_organization()
+            self._connect(organization, provider)
+            with self.subTest(provider=provider):
+                assert self._lookup(organization, "acme-labs/app") is None
+
     def test_get_repository_definition_not_found(self) -> None:
         """Test that nonexistent repository returns None"""
         result = get_repository_definition(
@@ -3198,7 +3257,7 @@ def test_get_repository_definition_no_integration_id(self) -> None:
             repo_full_name="getsentry/seer",
         )
 
-        assert result is not None
+        assert isinstance(result, RepositoryDefinitionResponse)
         assert result.integration_id is None
 
     def test_get_repository_definition_unsupported_provider(self) -> None:
@@ -3235,7 +3294,7 @@ def test_get_repository_definition_github_enterprise(self) -> None:
             repo_full_name="getsentry/seer",
         )
 
-        assert result is not None
+        assert isinstance(result, RepositoryDefinitionResponse)
         assert result.provider == "integrations:github_enterprise"
 
     def test_get_repository_definition_multiple_providers(self) -> None:
@@ -3264,7 +3323,7 @@ def test_get_repository_definition_multiple_providers(self) -> None:
             repo_full_name="getsentry/seer",
         )
 
-        assert result is not None
+        assert isinstance(result, RepositoryDefinitionResponse)
         # Should return the first matching repo (in this case, GitHub)
         assert result.provider in [
             "integrations:github",
@@ -3299,7 +3358,7 @@ def test_get_repository_definition_filters_unsupported_with_supported(self) -> N
             repo_full_name="getsentry/seer",
         )
 
-        assert result is not None
+        assert isinstance(result, RepositoryDefinitionResponse)
         assert result.provider == "integrations:github"
         assert result.external_id == "12345678"
 
@@ -3319,7 +3378,7 @@ def test_get_repository_definition_multipart_name(self) -> None:
             repo_full_name="getsentry/project/seer",
         )
 
-        assert result is not None
+        assert isinstance(result, RepositoryDefinitionResponse)
         assert result.owner == "getsentry"
         assert result.name == "project/seer"
 
@@ -3341,7 +3400,7 @@ def test_get_repository_definition_by_external_id(self) -> None:
             external_id="12345678",
         )
 
-        assert result is not None
+        assert isinstance(result, RepositoryDefinitionResponse)
         # Should return the CURRENT name from the database
         assert result.owner == "getsentry"
         assert result.name == "new-name"