fix(gateway): evidence/QA/doc paths populate files_changed from git

Bug:
    ContentActions.evidence() hard-coded files_changed=[] and diffed
    against HEAD~1 instead of the branch's parent. QA's _build_qa_claim_
    evidence (and doc/_impl mirrors) sourced files_changed from
    work_session.files_modified, which the gateway commit() never
    populates (no add_files_modified plumbing). Result: QA / docs / PM
    reviewers saw an empty change list on real PRs and only the latest
    commit's delta — flagged in smoke-9 when PR #20 showed the README
    change on GitHub but evidence() reported empty.

Fix:
    Added GitService.list_changed_files (git diff --name-only against
    parent branch). evidence(), _build_qa_claim_evidence,
    _claim_doc_evidence, and _build_i_am_done_ok all source files_changed
    from this — git is the authoritative source. evidence() also drops
    the HEAD~1 base so the diff is the full PR.

    Wired EvidenceRepo into ContentActionsDeps so evidence() returns
    journal_highlights too, matching the QA/doc shape.
This commit is contained in:
Renn F
2026-05-15 06:31:38 +02:00
parent d5ff8c7b13
commit 4fdde2b082
13 changed files with 294 additions and 18 deletions
+2 -1
View File
@@ -519,7 +519,7 @@ async def get_choreographer(
async def get_content_actions( async def get_content_actions(
db_session: DbSession, db_session: DbSession,
) -> ContentActions: ) -> ContentActions:
"""Build a ContentActions with all 7 service dependencies wired up.""" """Build a ContentActions with all service dependencies wired up."""
return ContentActions( return ContentActions(
ContentActionsDeps( ContentActionsDeps(
task=TaskService(db_session), task=TaskService(db_session),
@@ -530,6 +530,7 @@ async def get_content_actions(
workspace=WorkspaceService(db_session), workspace=WorkspaceService(db_session),
notifications=NotificationService(), notifications=NotificationService(),
notification_delivery=NotificationDeliveryService(db_session), notification_delivery=NotificationDeliveryService(db_session),
evidence_repo=EvidenceRepo(db_session),
) )
) )
@@ -1556,13 +1556,18 @@ class Choreographer:
async def _build_i_am_done_ok( async def _build_i_am_done_ok(
self, agent_id: UUID, task_id: UUID, t: Any self, agent_id: UUID, task_id: UUID, t: Any
) -> Envelope: ) -> Envelope:
"""Assemble the success envelope for i_am_done / _with_catchup.""" """Assemble the success envelope for i_am_done / _with_catchup.
Task #154: files_changed sourced from git (authoritative) so the
i_am_done envelope shows the same file list QA / docs / PMs will
see — independent of legacy ``add_files_modified`` plumbing.
"""
journal_highlights = await self.evidence_repo.journal_highlights_for_task( journal_highlights = await self.evidence_repo.journal_highlights_for_task(
task_id task_id
) )
files_changed: list[str] = [] files_changed: list[str] = []
if t.work_session_id: if t.branch_name:
files_changed = await self.work_session.files_changed(t.work_session_id) files_changed = await self.git.list_changed_files(branch_name=t.branch_name)
evidence = build_evidence_for_task( evidence = build_evidence_for_task(
t, t,
journal_highlights=journal_highlights, journal_highlights=journal_highlights,
+9 -3
View File
@@ -175,13 +175,19 @@ class DocMixin(_Base):
).with_introspection(task=t, role=role_str) ).with_introspection(task=t, role=role_str)
async def _claim_doc_evidence(self, task: Any, task_id: UUID) -> dict[str, Any]: async def _claim_doc_evidence(self, task: Any, task_id: UUID) -> dict[str, Any]:
"""Build the evidence dict surfaced inline on claim_doc_task ok envelopes.""" """Build the evidence dict surfaced inline on claim_doc_task ok envelopes.
Task #154: files_changed sourced from git (authoritative) instead
of ``work_session.files_modified``, which the gateway commit()
does not populate. The docs writer sees an accurate file list.
"""
files_changed: list[str] = [] files_changed: list[str] = []
if task.work_session_id:
files_changed = await self.work_session.files_changed(task.work_session_id)
diff = "" diff = ""
if task.branch_name: if task.branch_name:
diff = await self.git.diff(branch_name=task.branch_name) diff = await self.git.diff(branch_name=task.branch_name)
files_changed = await self.git.list_changed_files(
branch_name=task.branch_name
)
journal_highlights = await self.evidence_repo.journal_highlights_for_task( journal_highlights = await self.evidence_repo.journal_highlights_for_task(
task_id task_id
) )
+10 -5
View File
@@ -182,16 +182,21 @@ class QAMixin(_Base):
async def _build_qa_claim_evidence(self, t: Any, task_id: UUID) -> Any: async def _build_qa_claim_evidence(self, t: Any, task_id: UUID) -> Any:
"""Assemble the inline evidence payload returned by claim_review. """Assemble the inline evidence payload returned by claim_review.
Bundles files_changed (from work_session) + pr_diff_summary (from Bundles files_changed + pr_diff_summary (both from git, the
git) + journal_highlights so the QA agent has the full PR authoritative source) + journal_highlights so the QA agent has
context up-front and can't miss a piece. the full PR context up-front and can't miss a piece.
Task #154: files_changed comes from ``git.list_changed_files``
instead of ``work_session.files_modified``. The legacy
``add_files_modified`` HTTP path that populated files_modified
is not called by the gateway ``commit()``, so the work_session
list was always empty QA saw no files even on real PRs.
""" """
files_changed: list[str] = [] files_changed: list[str] = []
if t.work_session_id:
files_changed = await self.work_session.files_changed(t.work_session_id)
diff_summary = "" diff_summary = ""
if t.branch_name: if t.branch_name:
diff_summary = await self.git.diff(branch_name=t.branch_name) diff_summary = await self.git.diff(branch_name=t.branch_name)
files_changed = await self.git.list_changed_files(branch_name=t.branch_name)
journal_highlights = await self.evidence_repo.journal_highlights_for_task( journal_highlights = await self.evidence_repo.journal_highlights_for_task(
task_id task_id
) )
+27 -4
View File
@@ -217,6 +217,10 @@ class ContentActionsDeps:
# `NotificationDeliveryService`, not `NotificationService`. Keeping # `NotificationDeliveryService`, not `NotificationService`. Keeping
# them separate so the sender vs receiver concerns stay split. # them separate so the sender vs receiver concerns stay split.
notification_delivery: Any = None notification_delivery: Any = None
# Task #154: evidence() returns journal_highlights for QA/reviewer
# context. Matches the choreographer's EvidenceRepo wiring so both
# paths surface the same shape.
evidence_repo: Any = None
_VALID_NOTIFY_PRIORITIES: frozenset[str] = frozenset(p.value for p in _comms.Priority) _VALID_NOTIFY_PRIORITIES: frozenset[str] = frozenset(p.value for p in _comms.Priority)
@@ -254,6 +258,10 @@ class ContentActions:
def notifications(self) -> Any: def notifications(self) -> Any:
return self._deps.notifications return self._deps.notifications
@property
def evidence_repo(self) -> Any:
return self._deps.evidence_repo
async def commit( async def commit(
self, self,
*, *,
@@ -648,6 +656,13 @@ class ContentActions:
Allows inspection when caller is assignee OR task is unassigned Allows inspection when caller is assignee OR task is unassigned
(post-handoff transient state) strict ownership only blocks (post-handoff transient state) strict ownership only blocks
cross-agent inspection of an actively-owned task. cross-agent inspection of an actively-owned task.
Task #154: ``files_changed`` and ``pr_diff_summary`` are pulled
from git (against the branch's parent) — the authoritative source.
Earlier versions hard-coded ``files_changed=[]`` and used
``HEAD~1`` for the diff base, so QA / reviewers saw an empty
change list and only the latest commit's delta even when the PR
on GitHub had a multi-commit change set.
""" """
t = await self.task.get(task_id) t = await self.task.get(task_id)
if t is None: if t is None:
@@ -659,13 +674,21 @@ class ContentActions:
agent_id=agent_id, branch_name=t.branch_name agent_id=agent_id, branch_name=t.branch_name
) )
diff = "" diff = ""
files_changed: list[str] = []
if t.branch_name: if t.branch_name:
base = "HEAD~1" if t.commits else None diff = await self.git.diff(
diff = await self.git.diff(branch_name=t.branch_name, base=base) branch_name=t.branch_name, actor_agent_id=agent_id
)
files_changed = await self.git.list_changed_files(
branch_name=t.branch_name, actor_agent_id=agent_id
)
journal_highlights = await self.evidence_repo.journal_highlights_for_task(
task_id
)
ev = build_evidence_for_task( ev = build_evidence_for_task(
t, t,
journal_highlights=[], journal_highlights=journal_highlights,
files_changed=[], files_changed=files_changed,
pr_diff_summary=diff, pr_diff_summary=diff,
) )
return Envelope.ok( return Envelope.ok(
+30
View File
@@ -2135,6 +2135,36 @@ class GitService(BaseService):
diff_result = await self._run_git(workspace, diff_args, check=False) diff_result = await self._run_git(workspace, diff_args, check=False)
return diff_result.stdout return diff_result.stdout
async def list_changed_files(
self,
*,
branch_name: str,
base: str | None = None,
actor_agent_id: UUID | None = None,
) -> list[str]:
"""Return the file paths changed on `branch_name` relative to `base`.
Mirrors ``diff`` but invokes ``git diff --name-only`` so the
gateway evidence path can populate ``files_changed`` from the
authoritative git state independent of whether the agent
ever called the legacy ``add_files_modified`` HTTP endpoint
(which the gateway commit() does not call). Empty paths are
skipped; output preserves git's order.
"""
from roboco.services.gateway.merge_chain import parent_branch_for
workspace = await self._workspace_for_branch(
branch_name, actor_agent_id=actor_agent_id
)
if base is None:
parent = parent_branch_for(branch_name)
await self._run_git(workspace, ["fetch", "origin", parent], check=False)
args = ["diff", "--name-only", f"origin/{parent}...{branch_name}"]
else:
args = ["diff", "--name-only", f"{base}...{branch_name}"]
result = await self._run_git(workspace, args, check=False)
return [line for line in result.stdout.splitlines() if line.strip()]
async def commit( async def commit(
self, self,
*, *,
@@ -119,6 +119,12 @@ class _StubGit:
del branch_name, base, actor_agent_id del branch_name, base, actor_agent_id
return "stub diff" return "stub diff"
async def list_changed_files(
self, *, branch_name: str, base: Any = None, actor_agent_id: Any = None
) -> list[str]:
del branch_name, base, actor_agent_id
return []
async def pr_target(self, pr_number: int, *, actor_agent_id: Any = None) -> str: async def pr_target(self, pr_number: int, *, actor_agent_id: Any = None) -> str:
del pr_number, actor_agent_id del pr_number, actor_agent_id
return "main" return "main"
@@ -121,6 +121,12 @@ class _StubGit:
del branch_name, base, actor_agent_id del branch_name, base, actor_agent_id
return "stub diff" return "stub diff"
async def list_changed_files(
self, *, branch_name: str, base: Any = None, actor_agent_id: Any = None
) -> list[str]:
del branch_name, base, actor_agent_id
return []
async def pr_target(self, pr_number: int, *, actor_agent_id: Any = None) -> str: async def pr_target(self, pr_number: int, *, actor_agent_id: Any = None) -> str:
del pr_number, actor_agent_id del pr_number, actor_agent_id
return "main" return "main"
+1 -1
View File
@@ -70,9 +70,9 @@ async def test_claim_doc_task_returns_evidence() -> None:
task_svc.list_paused_for_agent.return_value = [] task_svc.list_paused_for_agent.return_value = []
task_svc.doc_claim.return_value = after task_svc.doc_claim.return_value = after
work_svc = AsyncMock() work_svc = AsyncMock()
work_svc.files_changed.return_value = ["README.md"]
git_svc = AsyncMock() git_svc = AsyncMock()
git_svc.diff.return_value = "+++ diff" git_svc.diff.return_value = "+++ diff"
git_svc.list_changed_files.return_value = ["README.md"]
deps = _make_deps(task=task_svc, work_session=work_svc, git=git_svc) deps = _make_deps(task=task_svc, work_session=work_svc, git=git_svc)
c = Choreographer(deps) c = Choreographer(deps)
+1 -1
View File
@@ -78,9 +78,9 @@ async def test_claim_review_returns_evidence_inline() -> None:
task_svc.list_paused_for_agent.return_value = [] task_svc.list_paused_for_agent.return_value = []
task_svc.qa_claim.return_value = t_claimed task_svc.qa_claim.return_value = t_claimed
work_svc = AsyncMock() work_svc = AsyncMock()
work_svc.files_changed.return_value = ["README.md"]
git_svc = AsyncMock() git_svc = AsyncMock()
git_svc.diff.return_value = "+++ diff content" git_svc.diff.return_value = "+++ diff content"
git_svc.list_changed_files.return_value = ["README.md"]
deps = _make_deps(task=task_svc, work_session=work_svc, git=git_svc) deps = _make_deps(task=task_svc, work_session=work_svc, git=git_svc)
c = Choreographer(deps) c = Choreographer(deps)
@@ -37,6 +37,11 @@ def _make_deps(**overrides: AsyncMock) -> ContentActionsDeps:
workspace = overrides.get("workspace", AsyncMock()) workspace = overrides.get("workspace", AsyncMock())
notifications = overrides.get("notifications", AsyncMock()) notifications = overrides.get("notifications", AsyncMock())
notification_delivery = overrides.get("notification_delivery", AsyncMock()) notification_delivery = overrides.get("notification_delivery", AsyncMock())
if "evidence_repo" in overrides:
evidence_repo = overrides["evidence_repo"]
else:
evidence_repo = AsyncMock()
evidence_repo.journal_highlights_for_task.return_value = []
return ContentActionsDeps( return ContentActionsDeps(
task=task, task=task,
git=git, git=git,
@@ -46,6 +51,7 @@ def _make_deps(**overrides: AsyncMock) -> ContentActionsDeps:
workspace=workspace, workspace=workspace,
notifications=notifications, notifications=notifications,
notification_delivery=notification_delivery, notification_delivery=notification_delivery,
evidence_repo=evidence_repo,
) )
@@ -44,6 +44,11 @@ def _make_deps(**overrides: AsyncMock) -> ContentActionsDeps:
journal = overrides.get("journal", AsyncMock()) journal = overrides.get("journal", AsyncMock())
workspace = overrides.get("workspace", AsyncMock()) workspace = overrides.get("workspace", AsyncMock())
notifications = overrides.get("notifications", AsyncMock()) notifications = overrides.get("notifications", AsyncMock())
if "evidence_repo" in overrides:
evidence_repo = overrides["evidence_repo"]
else:
evidence_repo = AsyncMock()
evidence_repo.journal_highlights_for_task.return_value = []
return ContentActionsDeps( return ContentActionsDeps(
task=task, task=task,
git=git, git=git,
@@ -52,6 +57,7 @@ def _make_deps(**overrides: AsyncMock) -> ContentActionsDeps:
journal=journal, journal=journal,
workspace=workspace, workspace=workspace,
notifications=notifications, notifications=notifications,
evidence_repo=evidence_repo,
) )
@@ -0,0 +1,182 @@
"""Task #154: evidence() must populate files_changed + use full PR diff.
Bug:
ContentActions.evidence() hard-coded ``files_changed=[]`` and called
``git.diff(branch_name=..., base="HEAD~1")``. Result: QA / reviewers
inspecting a real PR saw an empty change list and only the latest
commit's delta, even when GitHub showed a multi-commit change set.
Fix:
Pull files via ``git.list_changed_files(branch_name=...)`` (no base
full diff vs parent branch). Pull diff with ``base=None`` so the
full PR diff comes through. Both use git as the authoritative source
instead of the legacy ``work_session.files_modified`` field, which
the gateway ``commit()`` does not populate.
"""
from __future__ import annotations
from unittest.mock import AsyncMock, MagicMock
from uuid import uuid4
import pytest
from roboco.services.gateway.content_actions import ContentActions, ContentActionsDeps
def _deps_for_evidence(
task_svc: AsyncMock,
git_svc: AsyncMock,
workspace_svc: AsyncMock,
evidence_repo: AsyncMock,
) -> ContentActionsDeps:
return ContentActionsDeps(
task=task_svc,
git=git_svc,
messaging=AsyncMock(),
a2a=AsyncMock(),
journal=AsyncMock(),
workspace=workspace_svc,
notifications=AsyncMock(),
notification_delivery=AsyncMock(),
evidence_repo=evidence_repo,
)
def _task_with_pr(task_id: object, *, commits: list[str]) -> MagicMock:
return MagicMock(
id=task_id,
status="awaiting_qa",
assigned_to=None,
branch_name="feature/backend/abc12345--def67890",
work_session_id=uuid4(),
commits=commits,
pr_number=20,
pr_url="https://github.com/org/repo/pull/20",
dev_notes="see PR description",
acceptance_criteria_status=[],
)
@pytest.mark.asyncio
async def test_evidence_populates_files_changed_from_git() -> None:
"""The smoke-9 regression: PR #20 has README change on GitHub but
evidence() reports files_changed=[]. The fix queries git directly."""
agent_id = uuid4()
task_id = uuid4()
task_svc = AsyncMock()
task_svc.get.return_value = _task_with_pr(task_id, commits=["abc", "def"])
git_svc = AsyncMock()
git_svc.diff.return_value = "diff --git a/README.md b/README.md\n+added line\n"
git_svc.list_changed_files.return_value = ["README.md", "docs/guide.md"]
workspace_svc = AsyncMock()
evidence_repo = AsyncMock()
evidence_repo.journal_highlights_for_task.return_value = []
ca = ContentActions(
_deps_for_evidence(task_svc, git_svc, workspace_svc, evidence_repo)
)
env = await ca.evidence(agent_id=agent_id, task_id=task_id)
body = env.as_dict()
assert body["error"] is None
assert body["evidence"]["files_changed"] == ["README.md", "docs/guide.md"]
assert "diff --git" in body["evidence"]["pr_diff_summary"]
git_svc.list_changed_files.assert_awaited_once()
@pytest.mark.asyncio
async def test_evidence_uses_full_pr_diff_not_head_minus_one() -> None:
"""git.diff must be called with base=None (full PR diff vs parent),
not base='HEAD~1' (only the last commit)."""
agent_id = uuid4()
task_id = uuid4()
task_svc = AsyncMock()
# Multi-commit branch — the pre-fix code passed base='HEAD~1' when
# task.commits was non-empty, masking earlier commits' changes.
task_svc.get.return_value = _task_with_pr(task_id, commits=["sha1", "sha2", "sha3"])
git_svc = AsyncMock()
git_svc.diff.return_value = "full diff"
git_svc.list_changed_files.return_value = []
workspace_svc = AsyncMock()
evidence_repo = AsyncMock()
evidence_repo.journal_highlights_for_task.return_value = []
ca = ContentActions(
_deps_for_evidence(task_svc, git_svc, workspace_svc, evidence_repo)
)
await ca.evidence(agent_id=agent_id, task_id=task_id)
git_svc.diff.assert_awaited_once()
call_kwargs = git_svc.diff.await_args.kwargs
# Pre-fix bug: kwargs['base'] would be 'HEAD~1' for any multi-commit
# branch. Post-fix: base is omitted (or explicitly None).
base = call_kwargs.get("base")
assert base in (None, ""), (
f"git.diff must use full-PR diff (base=None), got base={base!r}"
)
assert call_kwargs.get("branch_name") == "feature/backend/abc12345--def67890"
@pytest.mark.asyncio
async def test_evidence_populates_journal_highlights() -> None:
"""evidence() must return journal_highlights so QA gets the dev's
decision/reflection context same as qa.py's claim_review evidence."""
agent_id = uuid4()
task_id = uuid4()
task_svc = AsyncMock()
task_svc.get.return_value = _task_with_pr(task_id, commits=["abc"])
git_svc = AsyncMock()
git_svc.diff.return_value = ""
git_svc.list_changed_files.return_value = []
workspace_svc = AsyncMock()
evidence_repo = AsyncMock()
highlights = [
{"scope": "decision", "title": "Use README format X", "content": "..."},
{"scope": "reflect", "title": "Lesson learned", "content": "..."},
]
evidence_repo.journal_highlights_for_task.return_value = highlights
ca = ContentActions(
_deps_for_evidence(task_svc, git_svc, workspace_svc, evidence_repo)
)
env = await ca.evidence(agent_id=agent_id, task_id=task_id)
body = env.as_dict()
assert body["evidence"]["journal_highlights"] == highlights
evidence_repo.journal_highlights_for_task.assert_awaited_once_with(task_id)
@pytest.mark.asyncio
async def test_evidence_no_branch_skips_git_calls() -> None:
"""A task without a branch_name has no PR yet — skip git entirely,
still return a valid envelope with empty files_changed."""
agent_id = uuid4()
task_id = uuid4()
task_svc = AsyncMock()
no_branch = MagicMock(
id=task_id,
status="claimed",
assigned_to=agent_id,
branch_name=None,
work_session_id=None,
commits=[],
pr_number=None,
pr_url=None,
dev_notes=None,
acceptance_criteria_status=[],
)
task_svc.get.return_value = no_branch
git_svc = AsyncMock()
workspace_svc = AsyncMock()
evidence_repo = AsyncMock()
evidence_repo.journal_highlights_for_task.return_value = []
ca = ContentActions(
_deps_for_evidence(task_svc, git_svc, workspace_svc, evidence_repo)
)
env = await ca.evidence(agent_id=agent_id, task_id=task_id)
body = env.as_dict()
assert body["error"] is None
assert body["evidence"]["files_changed"] == []
assert body["evidence"]["pr_diff_summary"] == ""
git_svc.diff.assert_not_awaited()
git_svc.list_changed_files.assert_not_awaited()