fix(api): export endpoint returns 409 when QA gate blocks (FU-6 UX — avoid false success toast)

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

.taskmaster/tasks/tasks.json

@@ -1967,10 +1967,10 @@
         "id": "56",
         "title": "[Retrieval finding] halacha_filters לא מסננים source_kind — דליפה חוצת-קורפוסים",
         "description": "התגלה תוך כדי משימה 53. ב-search_precedent_library_semantic וב-search_precedent_library_lexical (db.py): chunk_filters כוללים cl.source_kind=$sk אבל halacha_filters כוללים רק review_status. תוצאה: search_precedent_library(external) מחזיר גם הלכות internal_committee, ו-search_internal_decisions(internal) מחזיר גם הלכות external. אי-עקביות: chunks מסוננים, halachot לא. כרגע זה דווקא מסייע למציאוּת (לכן לא רגרסיה), אבל לא עקבי. דורש החלטת מדיניות: או לסנן halachot גם לפי source_kind (עקבי, אך 'מסתיר' שכבות), או להשאיר מאוחד במכוון + לתעד. אם משאירים מאוחד — לעדכן docstrings של שני הכלים שזה לא 'corpus נפרד'.",
-        "status": "pending",
+        "status": "cancelled",
         "priority": "low",
         "dependencies": [],
-        "details": "db.py: search_precedent_library_semantic (~שורה הקודמת ל-3311), search_precedent_library_lexical (3346). שתי הפונקציות: halacha_filters=['h.review_status IN ...'] — חסר cl.source_kind. נמצא בעת בדיקת רגרסיה למשימה 53.",
+        "details": "db.py: search_precedent_library_semantic (~שורה הקודמת ל-3311), search_precedent_library_lexical (3346). שתי הפונקציות: halacha_filters=['h.review_status IN ...'] — חסר cl.source_kind. נמצא בעת בדיקת רגרסיה למשימה 53.\n\n[SUPERSEDED 2026-05-30] נבלע ב-FU-4 / תת-משימה 62.1 (GAP-10). נסגר כדי למנוע כפילות-tracker.",
         "testStrategy": "לאחר החלטה: אם מסננים — search_precedent_library('...substantive...', external) לא מחזיר case_law_id internal; אם משאירים — docstring מעודכן + טסט מאשר התנהגות מכוונת.",
         "subtasks": [],
         "updatedAt": "2026-05-30T11:09:30.257989+00:00"
@@ -1984,7 +1984,7 @@
         "dependencies": [
           "55"
         ],
-        "details": "מקור: case_law.full_text קיים. נתיב: chunker.chunk_document(_hierarchical) → embeddings → החלפת precedent_chunks לתיק. למחוק chunks ישנים של התיק לפני הוספה. אחרי הרצה — ניתן להסיר את פילטר ה->=50 query (אופציונלי). תיקים מושפעים: SELECT DISTINCT case_law_id WHERE length(trim(content))<50.",
+        "details": "מקור: case_law.full_text קיים. נתיב: chunker.chunk_document(_hierarchical) → embeddings → החלפת precedent_chunks לתיק. למחוק chunks ישנים של התיק לפני הוספה. אחרי הרצה — ניתן להסיר את פילטר ה->=50 query (אופציונלי). תיקים מושפעים: SELECT DISTINCT case_law_id WHERE length(trim(content))<50.\n\n[קישור 2026-05-30] קשור ל-FU-3 (task 61, GAP-09 re-index). #57 = מיגרציה חד-פעמית של העבר (re-chunk legacy); FU-3 = ה-invariant הקדמי. נשמרים בנפרד במכוון.",
         "testStrategy": "אחרי re-chunk לתיק לדוגמה: 0 chunks<50 לאותו case_law_id; search_internal_decisions עדיין מחזיר את התיק; ספירת chunks סבירה.",
         "subtasks": [],
         "updatedAt": "2026-05-30T11:19:06.142606+00:00"

mcp-server/src/legal_mcp/services/db.py

@@ -1509,6 +1509,37 @@ async def get_decision_by_case(case_id: UUID) -> dict | None:
     return d
 
 
+async def get_critical_qa_failures(case_id: UUID) -> list[dict]:
+    """Return critical-severity failures from the case's latest QA run.
+
+    ``qa_results`` is cleared+rewritten per ``validate_decision`` run, so the
+    current rows for a ``case_id`` ARE the latest run. Returns rows where
+    ``severity='critical' AND passed=false``. Callers distinguish "no QA run
+    yet" (no rows at all) via ``qa_run_exists`` below.
+    """
+    pool = await get_pool()
+    async with pool.acquire() as conn:
+        rows = await conn.fetch(
+            """SELECT check_name, severity, passed, errors
+               FROM qa_results
+               WHERE case_id = $1 AND severity = 'critical' AND passed = false
+               ORDER BY check_name""",
+            case_id,
+        )
+    return [dict(r) for r in rows]
+
+
+async def qa_run_exists(case_id: UUID) -> bool:
+    """True if a QA run has ever been recorded for this case (any rows)."""
+    pool = await get_pool()
+    async with pool.acquire() as conn:
+        n = await conn.fetchval(
+            "SELECT count(*) FROM qa_results WHERE case_id = $1",
+            case_id,
+        )
+    return bool(n)
+
+
 async def update_decision(decision_id: UUID, **fields) -> None:
     if not fields:
         return
@@ -3165,7 +3196,10 @@ async def search_precedent_library_semantic(
     of halacha review status.
     """
     pool = await get_pool()
-    halacha_filters = ["h.review_status IN ('approved', 'published')"]
+    halacha_filters = [
+        "h.review_status IN ('approved', 'published')",
+        f"cl.source_kind = '{source_kind}'",
+    ]
     chunk_filters = [f"cl.source_kind = '{source_kind}'"]
     h_params: list = [query_embedding, limit]
     c_params: list = [query_embedding, limit]
@@ -3398,7 +3432,10 @@ async def search_precedent_library_lexical(
         return []
 
     pool = await get_pool()
-    halacha_filters = ["h.review_status IN ('approved', 'published')"]
+    halacha_filters = [
+        "h.review_status IN ('approved', 'published')",
+        f"cl.source_kind = '{source_kind}'",
+    ]
     chunk_filters = [f"cl.source_kind = '{source_kind}'"]
     # $1 = query, $2 = limit. Filters append starting at $3.
     h_params: list = [query, limit]

mcp-server/src/legal_mcp/services/qa_validator.py

@@ -67,7 +67,7 @@ def check_neutral_background(blocks: list[dict]) -> dict:
     """בדיקת ניטרליות בלוק הרקע (ו)."""
     vav = next((b for b in blocks if b["block_id"] == "block-vav"), None)
     if not vav or not vav.get("content"):
-        return {"name": "neutral_background", "passed": True, "errors": [], "severity": "critical"}
+        return {"name": "neutral_background", "passed": True, "errors": [], "severity": "warning"}
 
     errors = []
     lines = vav["content"].split("\n")

mcp-server/src/legal_mcp/tools/drafting.py

@@ -399,6 +399,26 @@ async def export_docx(case_number: str, output_path: str = "") -> str:
 
     case_id = UUID(case["id"])
 
+    # INV-EX3 / INV-QA3: a decision cannot be exported while critical QA gates
+    # fail (or before QA has been run at all). Gate on the STORED qa_results —
+    # cheap SELECT, no LLM re-run.
+    if not await db.qa_run_exists(case_id):
+        return json.dumps({
+            "status": "error",
+            "message": "ייצוא נחסם: בקרת איכות (QA) טרם רצה על התיק. "
+                       "הרץ validate_decision לפני ייצוא.",
+        }, ensure_ascii=False, indent=2)
+
+    critical = await db.get_critical_qa_failures(case_id)
+    if critical:
+        gate_names = ", ".join(r["check_name"] for r in critical)
+        return json.dumps({
+            "status": "error",
+            "message": f"ייצוא נחסם: שערי QA קריטיים נכשלו ({gate_names}). "
+                       f"תקן את הליקויים והרץ validate_decision מחדש לפני ייצוא.",
+            "failed_gates": [r["check_name"] for r in critical],
+        }, ensure_ascii=False, indent=2)
+
     try:
         path = await docx_exporter.export_decision(case_id, output_path or None)
         # Register this export as the new source of truth

mcp-server/tests/test_export_qa_gate.py

@@ -0,0 +1,151 @@
+"""Regression tests for FU-6.
+
+GAP-16 (INV-QA consistency): ``check_neutral_background`` must NOT return a
+``severity='critical'`` result while ``passed=True``. The empty/missing
+block-ו fallback now reports ``severity='warning'`` (consistent with passed).
+
+GAP-15 (INV-EX3 / INV-QA3): ``export_docx`` must refuse to export while
+critical QA gates fail OR before any QA run exists. It gates on the STORED
+``qa_results`` (cheap SELECT via ``db.get_critical_qa_failures`` /
+``db.qa_run_exists``) — it does NOT re-run the LLM validator.
+
+All tests run fully OFFLINE — the pool / db helpers / exporter / git are
+monkeypatched. No live Postgres needed.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import json
+
+import pytest
+
+from legal_mcp.services import db
+from legal_mcp.services import qa_validator
+from legal_mcp.tools import drafting
+
+
+# ── GAP-16 ────────────────────────────────────────────────────────
+
+def test_neutral_background_empty_block_is_warning_not_critical() -> None:
+    """Empty/missing block-ו → passed=True, so severity must be 'warning'."""
+    res = qa_validator.check_neutral_background([])  # no block-vav present
+    assert res["passed"] is True
+    assert res["severity"] == "warning", (
+        "a passed result must not carry severity='critical' (GAP-16)"
+    )
+
+
+def test_neutral_background_dirty_block_still_critical_path_untouched() -> None:
+    """A block-ו with judgment words still fails — fix didn't soften real checks."""
+    bad_word = qa_validator.VALUE_WORDS[0]
+    res = qa_validator.check_neutral_background(
+        [{"block_id": "block-vav", "content": f"הרקע: {bad_word} מאוד"}]
+    )
+    assert res["passed"] is False
+    assert res["errors"], "judgment-word violation should be reported"
+
+
+# ── GAP-15 ────────────────────────────────────────────────────────
+
+@pytest.fixture()
+def patched_export(monkeypatch: pytest.MonkeyPatch) -> dict:
+    """Monkeypatch case lookup, exporter, draft-path setter, and git so that
+    ``export_docx`` is isolated to the QA-gate decision. Returns a dict of
+    call-tracking flags.
+    """
+    calls = {"exported": False, "set_draft": False, "committed": False}
+
+    async def _get_case_by_number(case_number: str) -> dict:
+        return {"id": "00000000-0000-0000-0000-000000000001"}
+
+    async def _export_decision(case_id, output_path=None) -> str:
+        calls["exported"] = True
+        return "/tmp/decision.docx"
+
+    async def _set_active_draft_path(case_id, path) -> None:
+        calls["set_draft"] = True
+
+    def _commit_and_push(case_dir, msg) -> None:
+        calls["committed"] = True
+
+    # find_case_dir is called only on the success path; make it a no-op dir
+    class _FakeDir:
+        def exists(self) -> bool:
+            return False
+
+    monkeypatch.setattr(db, "get_case_by_number", _get_case_by_number)
+    monkeypatch.setattr(drafting.config, "find_case_dir", lambda cn: _FakeDir())
+    monkeypatch.setattr(drafting.git_sync, "commit_and_push", _commit_and_push)
+    # docx_exporter / set_active_draft_path are looked up dynamically; patch both
+    import legal_mcp.services.docx_exporter as docx_exporter
+    monkeypatch.setattr(docx_exporter, "export_decision", _export_decision)
+    monkeypatch.setattr(db, "set_active_draft_path", _set_active_draft_path)
+    return calls
+
+
+def _run(coro):
+    return asyncio.run(coro)
+
+
+def test_export_blocked_when_no_qa_run(
+    patched_export: dict, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    async def _qa_run_exists(case_id) -> bool:
+        return False
+
+    async def _get_critical(case_id) -> list:
+        return []
+
+    monkeypatch.setattr(db, "qa_run_exists", _qa_run_exists)
+    monkeypatch.setattr(db, "get_critical_qa_failures", _get_critical)
+
+    out = json.loads(_run(drafting.export_docx("8001-24")))
+    assert out["status"] == "error"
+    assert "QA" in out["message"] or "validate_decision" in out["message"]
+    assert patched_export["exported"] is False, "must not call the exporter"
+    assert patched_export["committed"] is False, "must not git-commit"
+
+
+def test_export_blocked_when_critical_failures(
+    patched_export: dict, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    async def _qa_run_exists(case_id) -> bool:
+        return True
+
+    async def _get_critical(case_id) -> list:
+        return [
+            {"check_name": "claims_coverage", "severity": "critical",
+             "passed": False, "errors": []},
+            {"check_name": "structural_integrity", "severity": "critical",
+             "passed": False, "errors": []},
+        ]
+
+    monkeypatch.setattr(db, "qa_run_exists", _qa_run_exists)
+    monkeypatch.setattr(db, "get_critical_qa_failures", _get_critical)
+
+    out = json.loads(_run(drafting.export_docx("8001-24")))
+    assert out["status"] == "error"
+    assert out["failed_gates"] == ["claims_coverage", "structural_integrity"]
+    assert "claims_coverage" in out["message"]
+    assert patched_export["exported"] is False, "must not call the exporter"
+    assert patched_export["committed"] is False, "must not git-commit"
+
+
+def test_export_proceeds_when_clean(
+    patched_export: dict, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    async def _qa_run_exists(case_id) -> bool:
+        return True
+
+    async def _get_critical(case_id) -> list:
+        return []
+
+    monkeypatch.setattr(db, "qa_run_exists", _qa_run_exists)
+    monkeypatch.setattr(db, "get_critical_qa_failures", _get_critical)
+
+    out = json.loads(_run(drafting.export_docx("8001-24")))
+    assert out["status"] == "completed", out
+    assert out["path"] == "/tmp/decision.docx"
+    assert patched_export["exported"] is True, "clean QA must allow export"
+    assert patched_export["set_draft"] is True, "active_draft_path must be set"

mcp-server/tests/test_precedent_corpus_isolation.py

@@ -0,0 +1,97 @@
+"""Regression test for GAP-10 / INV-RET1: corpus separation enforced on
+EVERY precedent-library query path — including the halacha sub-query.
+
+Bug: ``search_precedent_library_semantic`` and
+``search_precedent_library_lexical`` filtered the *chunk* sub-query by
+``cl.source_kind`` but NOT the *halacha* sub-query. So an external
+(``source_kind='external_upload'``) search leaked internal-committee
+halachot, and an internal search leaked external-ruling halachot — a
+cross-corpus contamination of the rule-level results.
+
+Fix: the same ``cl.source_kind = '<kind>'`` predicate that gates the
+chunk query now also gates the halacha query, in BOTH functions.
+
+This test runs fully OFFLINE — it monkeypatches ``db.get_pool`` with a
+fake pool that captures every SQL string passed to ``fetch`` instead of
+hitting Postgres. It asserts the captured halacha SQL carries the
+source_kind predicate identical to the chunk SQL.
+"""
+
+from __future__ import annotations
+
+import asyncio
+
+import pytest
+
+from legal_mcp.services import db
+
+
+class _FakePool:
+    """Captures SQL passed to ``fetch``; returns no rows."""
+
+    def __init__(self) -> None:
+        self.queries: list[str] = []
+
+    async def fetch(self, sql: str, *args) -> list:  # noqa: ANN002
+        self.queries.append(sql)
+        return []
+
+
+def _classify(queries: list[str]) -> tuple[str, str]:
+    """Return (halacha_sql, chunk_sql) from the captured queries."""
+    halacha = next(q for q in queries if "FROM halachot h" in q)
+    chunk = next(q for q in queries if "FROM precedent_chunks pc" in q)
+    return halacha, chunk
+
+
+@pytest.fixture()
+def fake_pool(monkeypatch: pytest.MonkeyPatch) -> _FakePool:
+    pool = _FakePool()
+
+    async def _get_pool() -> _FakePool:
+        return pool
+
+    monkeypatch.setattr(db, "get_pool", _get_pool)
+    return pool
+
+
+@pytest.mark.parametrize("source_kind", ["external_upload", "internal_committee"])
+def test_semantic_halacha_query_is_source_kind_scoped(
+    fake_pool: _FakePool, source_kind: str
+) -> None:
+    asyncio.run(
+        db.search_precedent_library_semantic(
+            query_embedding=[0.0] * 8,
+            source_kind=source_kind,
+            include_halachot=True,
+            limit=5,
+        )
+    )
+    halacha_sql, chunk_sql = _classify(fake_pool.queries)
+    predicate = f"cl.source_kind = '{source_kind}'"
+    assert predicate in chunk_sql, "chunk query must be source_kind-scoped (precondition)"
+    assert predicate in halacha_sql, (
+        "halacha query MUST carry the same source_kind predicate as the "
+        "chunk query — otherwise cross-corpus halacha leakage (GAP-10)"
+    )
+
+
+@pytest.mark.parametrize("source_kind", ["external_upload", "internal_committee"])
+def test_lexical_halacha_query_is_source_kind_scoped(
+    fake_pool: _FakePool, source_kind: str
+) -> None:
+    asyncio.run(
+        db.search_precedent_library_lexical(
+            query="zoning setback",
+            source_kind=source_kind,
+            include_halachot=True,
+            limit=5,
+        )
+    )
+    halacha_sql, chunk_sql = _classify(fake_pool.queries)
+    predicate = f"cl.source_kind = '{source_kind}'"
+    assert predicate in chunk_sql, "chunk query must be source_kind-scoped (precondition)"
+    assert predicate in halacha_sql, (
+        "halacha query MUST carry the same source_kind predicate as the "
+        "chunk query — otherwise cross-corpus halacha leakage (GAP-10)"
+    )

web/app.py

@@ -3170,10 +3170,25 @@ async def api_export_docx(case_number: str, background_tasks: BackgroundTasks):
     (markdown body + download link) to the linked issue.
     """
     result = await drafting_tools.export_docx(case_number)
-    try:
+    if isinstance(result, dict):
-        data = json.loads(result)
+        data = result
-    except json.JSONDecodeError:
+    else:
-        raise HTTPException(500, result)
+        try:
+            data = json.loads(result)
+        except (json.JSONDecodeError, TypeError):
+            # export_docx can also return a plain (non-JSON) string, e.g.
+            # "תיק ... לא נמצא." — surface it as a 500 with the raw text.
+            raise HTTPException(500, str(result))
+
+    # FU-6: a QA gate (or another error) can block the export. export_docx
+    # signals this with status == "error". Returning the existing 200 here
+    # would let the UI show a false "exported successfully" toast, so we map
+    # a block to 409 Conflict carrying the Hebrew message + failed_gates.
+    if isinstance(data, dict) and data.get("status") == "error":
+        detail = {"message": data.get("message", "ייצוא נחסם.")}
+        if data.get("failed_gates"):
+            detail["failed_gates"] = data["failed_gates"]
+        raise HTTPException(409, detail)
 
     # Notify the Paperclip plugin to attach the final-decision document.
     docx_filename = (