fix(gateway): journal task_id auto-injection works from blocked/paused

Smoke-5 root cause. Agents wrote 5 decisions / 8 reflections / 1 struggle
during the run — every single entry persisted with task_id=NULL. The C8
tracing gate then never saw them and PMs spiraled forever on
'missing: journal:decision' while their decisions sat orphaned.

Cause: ContentActions.note/say/dm/notify called
TaskService.get_active_task_for_agent for task_id auto-injection. That
helper filters to _DEV_ACTIVE_STATUSES = {claimed, in_progress,
verifying, awaiting_qa, awaiting_documentation}. BLOCKED, PAUSED, and
NEEDS_REVISION fall outside that set — so the moment an agent gets
stuck (which is exactly when they journal), auto-injection returns None
and the entry persists without task_id.

Fix:
- New TaskService.get_journal_context_task_for_agent — same shape as
  get_active_task_for_agent but the status set
  _JOURNAL_CONTEXT_STATUSES adds BLOCKED, PAUSED, NEEDS_REVISION.
- ContentActions.note/say/dm/notify use the new lookup.
- ContentActions.commit keeps the narrow get_active_task_for_agent —
  can't commit from blocked, so the dev-active set is correct there.

Tests:
- tests/unit/services/test_journal_context_lookup.py — 5 tests pinning
  the two queries: journal-context INCLUDES blocked/paused/needs_revision,
  dev-active EXCLUDES them.
- Existing content-actions tests updated to stub the new method
  alongside the old one.

This alone may be 70% of what was killing smoke runs end-to-end.
This commit is contained in:
Renn F
2026-05-14 04:21:07 +02:00
parent 7430c88f63
commit 1bd6eb3372
6 changed files with 154 additions and 6 deletions
+6 -6
View File
@@ -320,8 +320,8 @@ class ContentActions:
"""Gate Set D: refuse content posts on tasks the caller does not own.
Only call this for *explicit* task_id (caller passed it themselves).
Auto-fill from get_active_task_for_agent is implicitly self-owned
and does not need a re-check.
Auto-fill from get_journal_context_task_for_agent is implicitly
self-owned and does not need a re-check.
Allows ``assigned_to=None`` (post-handoff transient state) so QA /
documenter can still inspect tasks between reassignments.
@@ -365,7 +365,7 @@ class ContentActions:
if reject := await self._verify_explicit_task_ownership(agent_id, task_id):
return reject
else:
t = await self.task.get_active_task_for_agent(agent_id)
t = await self.task.get_journal_context_task_for_agent(agent_id)
if t is not None:
task_id = t.id
s = structured or {}
@@ -464,7 +464,7 @@ class ContentActions:
if reject := await self._verify_explicit_task_ownership(agent_id, task_id):
return reject
else:
t = await self.task.get_active_task_for_agent(agent_id)
t = await self.task.get_journal_context_task_for_agent(agent_id)
if t is not None:
task_id = t.id
try:
@@ -516,7 +516,7 @@ class ContentActions:
if reject := await self._verify_explicit_task_ownership(agent_id, task_id):
return reject
else:
t = await self.task.get_active_task_for_agent(agent_id)
t = await self.task.get_journal_context_task_for_agent(agent_id)
if t is not None:
task_id = t.id
if task_id is None:
@@ -592,7 +592,7 @@ class ContentActions:
if reject := await self._verify_explicit_task_ownership(agent_id, task_id):
return reject
else:
t = await self.task.get_active_task_for_agent(agent_id)
t = await self.task.get_journal_context_task_for_agent(agent_id)
if t is not None:
task_id = t.id
await self.notifications.send_ack_notification(
+40
View File
@@ -4489,6 +4489,22 @@ class TaskService(BaseService):
TaskStatus.AWAITING_DOCUMENTATION,
}
# States in which the agent still owns the task for content / journal
# context, even if it isn't progressing. Used by note / say / dm /
# evidence so journal entries written from blocked or paused get the
# task_id auto-attached (otherwise the C8 + tracing gates never see
# the agent's decisions and the agent spirals).
_JOURNAL_CONTEXT_STATUSES: ClassVar[set[TaskStatus]] = {
TaskStatus.CLAIMED,
TaskStatus.IN_PROGRESS,
TaskStatus.VERIFYING,
TaskStatus.AWAITING_QA,
TaskStatus.AWAITING_DOCUMENTATION,
TaskStatus.BLOCKED,
TaskStatus.PAUSED,
TaskStatus.NEEDS_REVISION,
}
# Statuses that count as "still assignable to the agent" for triage.
_AGENT_NON_TERMINAL_STATUSES: ClassVar[set[TaskStatus]] = {
TaskStatus.PENDING,
@@ -4629,6 +4645,30 @@ class TaskService(BaseService):
result = await self.session.execute(query)
return result.scalar_one_or_none()
async def get_journal_context_task_for_agent(
self, agent_id: UUID
) -> TaskTable | None:
"""Most-recently-updated task the agent owns for journal/content context.
Wider than ``get_active_task_for_agent`` — includes BLOCKED, PAUSED,
and NEEDS_REVISION so journal entries written while stuck still
get the task_id auto-attached. Smoke-5 surfaced the bug: PMs
wrote decisions during blocked state, auto-injection returned
None, entries persisted with task_id=NULL, the C8 tracing gate
never saw them, agents spiraled forever.
"""
query = (
select(TaskTable)
.where(
TaskTable.assigned_to == agent_id,
TaskTable.status.in_(self._JOURNAL_CONTEXT_STATUSES),
)
.order_by(TaskTable.updated_at.desc().nullslast())
.limit(1)
)
result = await self.session.execute(query)
return result.scalar_one_or_none()
async def list_pending_for_agent(self, agent_id: UUID) -> list[TaskTable]:
"""Tasks assigned to this agent that are still in PENDING status.
@@ -17,6 +17,7 @@ def _make_deps(**overrides: AsyncMock) -> ContentActionsDeps:
else:
task = AsyncMock()
task.get_active_task_for_agent.return_value = None
task.get_journal_context_task_for_agent.return_value = None
# commit() checks caller role server-side; default-created mocks
# need a default developer role so existing tests pass through.
# Caller-supplied mocks must set agent_for themselves.
@@ -400,6 +401,7 @@ async def test_note_auto_fills_task_id_from_active_task() -> None:
task_obj = MagicMock(id=task_id, status="in_progress")
task_svc = AsyncMock()
task_svc.get_active_task_for_agent.return_value = task_obj
task_svc.get_journal_context_task_for_agent.return_value = task_obj
journal_svc = AsyncMock()
deps = _make_deps(task=task_svc, journal=journal_svc)
@@ -440,6 +442,7 @@ async def test_say_auto_injects_task_id_when_active_task_exists() -> None:
task_obj = MagicMock(id=task_id, status="in_progress")
task_svc = AsyncMock()
task_svc.get_active_task_for_agent.return_value = task_obj
task_svc.get_journal_context_task_for_agent.return_value = task_obj
messaging_svc = AsyncMock()
deps = _make_deps(task=task_svc, messaging=messaging_svc)
@@ -507,6 +510,7 @@ async def test_dm_with_active_task_succeeds() -> None:
task_obj = MagicMock(id=task_id, status="in_progress")
task_svc = AsyncMock()
task_svc.get_active_task_for_agent.return_value = task_obj
task_svc.get_journal_context_task_for_agent.return_value = task_obj
a2a_svc = AsyncMock()
deps = _make_deps(task=task_svc, a2a=a2a_svc)
+2
View File
@@ -22,6 +22,7 @@ def _make_deps(**overrides: AsyncMock) -> ContentActionsDeps:
else:
task = AsyncMock()
task.get_active_task_for_agent.return_value = None
task.get_journal_context_task_for_agent.return_value = None
task.agent_for.return_value = MagicMock(role="cell_pm")
git = overrides.get("git", AsyncMock())
@@ -239,6 +240,7 @@ async def test_notify_auto_fills_task_id_from_active_task() -> None:
task_obj = MagicMock(id=task_id, status="awaiting_pm_review")
task_svc = AsyncMock()
task_svc.get_active_task_for_agent.return_value = task_obj
task_svc.get_journal_context_task_for_agent.return_value = task_obj
task_svc.agent_for.return_value = MagicMock(role="cell_pm")
notif_svc = AsyncMock()
+2
View File
@@ -26,6 +26,7 @@ def _make_deps(**overrides: AsyncMock) -> ContentActionsDeps:
task = AsyncMock()
task.agent_for.return_value = MagicMock(role="developer", slug="be-dev-1")
task.get_active_task_for_agent.return_value = None
task.get_journal_context_task_for_agent.return_value = None
git = overrides.get("git", AsyncMock())
messaging = overrides.get("messaging", AsyncMock())
@@ -68,6 +69,7 @@ async def test_say_posted_status_with_active_task() -> None:
task_svc = AsyncMock()
task_svc.agent_for.return_value = MagicMock(role="developer", slug="be-dev-1")
task_svc.get_active_task_for_agent.return_value = task_obj
task_svc.get_journal_context_task_for_agent.return_value = task_obj
ca = ContentActions(_make_deps(task=task_svc))
env = await ca.say(
@@ -0,0 +1,100 @@
"""Smoke-5: get_journal_context_task_for_agent includes BLOCKED, PAUSED,
NEEDS_REVISION so note/say/dm/notify auto-inject task_id while the agent
is stuck.
Original bug: get_active_task_for_agent filtered to DEV_ACTIVE statuses only.
PMs writing decisions during BLOCKED state got task_id=NULL on their journal
entries. The C8 tracing gate then never found the decisions and the agent
spiraled. Smoke-5 wrote 5 decisions, 8 reflections, 1 struggle all with
task_id=NULL because the agent was stuck when journaling.
"""
from __future__ import annotations
from unittest.mock import AsyncMock, MagicMock
from uuid import uuid4
import pytest
from roboco.models.base import TaskStatus
from roboco.services.task import TaskService
def _service_with(execute_returns: object) -> TaskService:
session = MagicMock()
session.execute = AsyncMock(return_value=execute_returns)
return TaskService(session)
@pytest.mark.asyncio
async def test_journal_context_returns_blocked_task() -> None:
"""A BLOCKED task is returned by the journal-context lookup."""
task = MagicMock(id=uuid4(), status=TaskStatus.BLOCKED)
result = MagicMock()
result.scalar_one_or_none.return_value = task
svc = _service_with(result)
found = await svc.get_journal_context_task_for_agent(uuid4())
assert found is task
@pytest.mark.asyncio
async def test_journal_context_returns_paused_task() -> None:
"""A PAUSED task is returned by the journal-context lookup."""
task = MagicMock(id=uuid4(), status=TaskStatus.PAUSED)
result = MagicMock()
result.scalar_one_or_none.return_value = task
svc = _service_with(result)
found = await svc.get_journal_context_task_for_agent(uuid4())
assert found is task
@pytest.mark.asyncio
async def test_journal_context_returns_needs_revision_task() -> None:
"""A NEEDS_REVISION task is returned so QA-rejected devs still get task_id."""
task = MagicMock(id=uuid4(), status=TaskStatus.NEEDS_REVISION)
result = MagicMock()
result.scalar_one_or_none.return_value = task
svc = _service_with(result)
found = await svc.get_journal_context_task_for_agent(uuid4())
assert found is task
@pytest.mark.asyncio
async def test_journal_context_query_filters_by_journal_statuses() -> None:
"""The query's where clause uses _JOURNAL_CONTEXT_STATUSES, not _DEV_ACTIVE."""
result = MagicMock()
result.scalar_one_or_none.return_value = None
session = MagicMock()
session.execute = AsyncMock(return_value=result)
svc = TaskService(session)
agent_id = uuid4()
await svc.get_journal_context_task_for_agent(agent_id)
assert session.execute.await_count == 1
sent_query = session.execute.await_args.args[0]
rendered = str(sent_query.compile(compile_kwargs={"literal_binds": True}))
for s in ("blocked", "paused", "needs_revision", "in_progress", "claimed"):
assert s in rendered, (
f"Journal-context query missing status {s}. Rendered SQL: {rendered}"
)
@pytest.mark.asyncio
async def test_dev_active_query_still_excludes_blocked() -> None:
"""The narrow get_active_task_for_agent stays narrow — commit() relies on it."""
result = MagicMock()
result.scalar_one_or_none.return_value = None
session = MagicMock()
session.execute = AsyncMock(return_value=result)
svc = TaskService(session)
await svc.get_active_task_for_agent(uuid4())
sent_query = session.execute.await_args.args[0]
rendered = str(sent_query.compile(compile_kwargs={"literal_binds": True}))
assert "'blocked'" not in rendered, (
"Dev-active query must NOT include BLOCKED — commit() would otherwise "
"allow commits from a blocked task."
)
assert "'paused'" not in rendered
assert "'needs_revision'" not in rendered