feat(explorer): send interactivity flag in client

[roaga]

Sends an is_interactive flag in the client. Defaults to False, but sets to True for the UI chat endpoint.

Part of AIML-1689: indicator on state whether on Sentry UI or not

[roaga]

AIML-1689 indicator on state whether on Sentry UI or not

[cursor]

Bug: Missing is_interactive flag in continue_run payload

The continue_run method doesn't include is_interactive in its payload, unlike start_run which sends it on line 166. When users continue a conversation from the UI chat endpoint, the continuation requests won't be marked as interactive, causing inconsistent behavior where only the first message in a conversation is flagged as interactive while follow-ups are not, despite all interactions occurring on the Sentry UI.

src/sentry/seer/explorer/client.py#L197-L230

sentry/src/sentry/seer/explorer/client.py

Lines 197 to 230 in 8e3930e

	def continue_run(
	self,
	run_id: int,
	prompt: str,
	insert_index: int | None = None,
	on_page_context: str | None = None,
	) -> int:
	"""
	Continue an existing Seer Explorer session. This allows you to add follow-up queries to an ongoing conversation.
	Args:
	run_id: The run ID from start_run()
	prompt: The follow-up task/query for the agent
	insert_index: Optional index to insert the message at
	on_page_context: Optional context from the user's screen
	Returns:
	int: The run ID (same as input)
	Raises:
	requests.HTTPError: If the Seer API request fails
	"""
	path = "/v1/automation/explorer/chat"
	payload: dict[str, Any] = {
	"organization_id": self.organization.id,
	"query": prompt,
	"run_id": run_id,
	"insert_index": insert_index,
	"on_page_context": on_page_context,
	}
	body = orjson.dumps(payload, option=orjson.OPT_NON_STR_KEYS)

 

---

[codecov]

❌ 2 Tests Failed:

Tests completed	Failed	Passed	Skipped
30017	2	30015	239

View the top 2 failed test(s) by shortest run time

tests.sentry.seer.endpoints.test_organization_seer_explorer_chat.OrganizationSeerExplorerChatEndpointTest::test_post_new_conversation_calls_client

Stack Traces | 1.41s run time

#x1B[1m#x1B[.../seer/endpoints/test_organization_seer_explorer_chat.py#x1B[0m:72: in test_post_new_conversation_calls_client
    mock_client_class.assert_called_once_with(self.organization, ANY)
#x1B[1m#x1B[.../hostedtoolcache/Python/3.13.1....../x64/lib/python3.13/unittest/mock.py#x1B[0m:989: in assert_called_once_with
    return self.assert_called_with(*args, **kwargs)
#x1B[1m#x1B[.../hostedtoolcache/Python/3.13.1....../x64/lib/python3.13/unittest/mock.py#x1B[0m:977: in assert_called_with
    raise AssertionError(_error_message()) from cause
#x1B[1m#x1B[31mE   AssertionError: expected call not found.#x1B[0m
#x1B[1m#x1B[31mE   Expected: SeerExplorerClient(<Organization at 0x7f3898f447e0: id=4557170394857472, owner_id=None, name='baz', slug='baz'>, <ANY>)#x1B[0m
#x1B[1m#x1B[31mE     Actual: SeerExplorerClient(<Organization at 0x7f3898f44cb0: id=4557170394857472, owner_id=None, name='baz', slug='baz'>, <SimpleLazyObject: RpcUser(id=732, pk=732, name='', email='admin@localhost', username='admin@localhost', actor_id=None, display_name='admin@localhost', label='admin@localhost', is_superuser=True, is_authenticated=True, is_anonymous=False, is_active=True, is_staff=True, is_unclaimed=False, last_active=datetime.datetime(2025, 11, 24, 22, 30, 41, 944401, tzinfo=datetime.timezone.utc), is_sentry_app=False, password_usable=True, is_password_expired=False, session_nonce=None, roles=frozenset(), permissions=frozenset(), avatar=None, emails=frozenset({'admin@localhost'}), useremails=[RpcUserEmail(id=734, email='admin@localhost', is_verified=True)], authenticators=[])>, is_interactive=True)#x1B[0m

tests.sentry.seer.endpoints.test_organization_seer_explorer_chat.OrganizationSeerExplorerChatEndpointTest::test_post_continue_conversation_calls_client

Stack Traces | 1.64s run time

#x1B[1m#x1B[.../seer/endpoints/test_organization_seer_explorer_chat.py#x1B[0m:93: in test_post_continue_conversation_calls_client
    mock_client_class.assert_called_once_with(self.organization, ANY)
#x1B[1m#x1B[.../hostedtoolcache/Python/3.13.1....../x64/lib/python3.13/unittest/mock.py#x1B[0m:989: in assert_called_once_with
    return self.assert_called_with(*args, **kwargs)
#x1B[1m#x1B[.../hostedtoolcache/Python/3.13.1....../x64/lib/python3.13/unittest/mock.py#x1B[0m:977: in assert_called_with
    raise AssertionError(_error_message()) from cause
#x1B[1m#x1B[31mE   AssertionError: expected call not found.#x1B[0m
#x1B[1m#x1B[31mE   Expected: SeerExplorerClient(<Organization at 0x7f3cc0341130: id=4557170400231424, owner_id=None, name='baz', slug='baz'>, <ANY>)#x1B[0m
#x1B[1m#x1B[31mE     Actual: SeerExplorerClient(<Organization at 0x7f3cb8f96170: id=4557170400231424, owner_id=None, name='baz', slug='baz'>, <SimpleLazyObject: RpcUser(id=746, pk=746, name='', email='admin@localhost', username='admin@localhost', actor_id=None, display_name='admin@localhost', label='admin@localhost', is_superuser=True, is_authenticated=True, is_anonymous=False, is_active=True, is_staff=True, is_unclaimed=False, last_active=datetime.datetime(2025, 11, 24, 22, 32, 3, 964261, tzinfo=datetime.timezone.utc), is_sentry_app=False, password_usable=True, is_password_expired=False, session_nonce=None, roles=frozenset(), permissions=frozenset(), avatar=None, emails=frozenset({'admin@localhost'}), useremails=[RpcUserEmail(id=758, email='admin@localhost', is_verified=True)], authenticators=[])>, is_interactive=True)#x1B[0m

To view more test analytics, go to the Test Analytics Dashboard
_{📋 Got 3 mins? Take this short survey to help us improve Test Analytics.}

[cursor]

Bug: `is_interactive` flag missing from `continue_run` payload

The is_interactive flag is added to the start_run payload but is missing from the continue_run payload. When a conversation is continued through the UI chat endpoint (which sets is_interactive=True on the client), the continue_run method won't send this flag to the Seer API. The stored self.is_interactive value needs to be included in the continue_run payload for consistent behavior between new and continued conversations.

src/sentry/seer/explorer/client.py#L221-L228

sentry/src/sentry/seer/explorer/client.py

Lines 221 to 228 in 1cee59d

	payload: dict[str, Any] = {
	"organization_id": self.organization.id,
	"query": prompt,
	"run_id": run_id,
	"insert_index": insert_index,
	"on_page_context": on_page_context,
	}

 

---

[cursor]

Bug: Missing `is_interactive` flag in continue_run payload

The is_interactive flag is added to the start_run payload but not to the continue_run payload. When the UI chat endpoint continues an existing conversation via continue_run, the is_interactive=True setting stored in self.is_interactive is never sent to the Seer backend. This means follow-up messages in interactive conversations won't receive the full interactive treatment intended by this feature.

src/sentry/seer/explorer/client.py#L221-L228

sentry/src/sentry/seer/explorer/client.py

Lines 221 to 228 in 3d048f5

	payload: dict[str, Any] = {
	"organization_id": self.organization.id,
	"query": prompt,
	"run_id": run_id,
	"insert_index": insert_index,
	"on_page_context": on_page_context,
	}

src/sentry/seer/explorer/client.py#L165-L166

sentry/src/sentry/seer/explorer/client.py

Lines 165 to 166 in 3d048f5

	"intelligence_level": self.intelligence_level,
	"is_interactive": self.is_interactive,

 

---