diff --git a/backend/agenteval/intelligent_eval/decision.py b/backend/agenteval/intelligent_eval/decision.py
new file mode 100644
index 0000000..b26064b
--- /dev/null
+++ b/backend/agenteval/intelligent_eval/decision.py
@@ -0,0 +1,276 @@
+"""Decision logic for intelligent eval workers (决策逻辑).
+
+Worker analyzes current situation and decides what to do:
+- execute_session: Execute a new session
+- wait: Wait for next tick
+- start_analysis: Start analysis (all sessions completed)
+"""
+
+from datetime import datetime, timedelta
+from enum import Enum
+from typing import Optional
+
+from sqlmodel import Session, select
+
+from agenteval.intelligent_eval.models import IntelligentEvalStatus
+from agenteval.storage.db import (
+    IntelligentEvalDB,
+    IntelligentEvalSessionDB,
+    as_utc,
+    utc_now,
+)
+
+
+class DecisionType(str, Enum):
+    """Decision types for worker."""
+
+    EXECUTE_SESSION = "execute_session"
+    WAIT = "wait"
+    START_ANALYSIS = "start_analysis"
+
+
+class Decision:
+    """A decision made by a worker."""
+
+    def __init__(self, decision_type: DecisionType, reason: str, context: dict):
+        self.decision_type = decision_type
+        self.reason = reason
+        self.context = context
+
+    def to_dict(self) -> dict:
+        return {
+            "decision_type": self.decision_type.value,
+            "reason": self.reason,
+            "context": self.context,
+        }
+
+
+def _parse_time_slot(time_slot: str) -> Optional[tuple[int, int]]:
+    """Parse time slot string (e.g., "8-10h") to (start_hour, end_hour).
+
+    Returns:
+        (start_hour, end_hour) tuple, or None if invalid format
+    """
+    try:
+        parts = time_slot.replace("h", "").split("-")
+        if len(parts) != 2:
+            return None
+        return (int(parts[0]), int(parts[1]))
+    except (ValueError, AttributeError):
+        return None
+
+
+def _get_current_slot(time_distribution: list[dict], current_offset: timedelta) -> Optional[dict]:
+    """Get current time slot based on offset.
+
+    Returns:
+        Current slot dict, or None if not in any slot
+    """
+    current_hours = current_offset.total_seconds() / 3600
+
+    for slot in time_distribution:
+        time_slot = slot.get("time_slot", "")
+        parsed = _parse_time_slot(time_slot)
+        if parsed is None:
+            continue
+
+        start_hour, end_hour = parsed
+        if start_hour <= current_hours < end_hour:
+            return slot
+
+    return None
+
+
+def _count_sessions_in_slot(
+    eval_id: str, slot: dict, eval_started_at: datetime, session: Session
+) -> int:
+    """Count sessions created in a time slot.
+
+    Args:
+        eval_id: Evaluation ID
+        slot: Time slot dict (e.g., {"time_slot": "8-10h", "sessions": 2})
+        eval_started_at: When the eval started
+        session: Database session
+
+    Returns:
+        Number of sessions created in this slot
+    """
+    time_slot = slot.get("time_slot", "")
+    parsed = _parse_time_slot(time_slot)
+    if parsed is None:
+        return 0
+
+    start_hour, end_hour = parsed
+    
+    # Convert to naive datetime for SQLite comparison
+    slot_start = (as_utc(eval_started_at) + timedelta(hours=start_hour)).replace(tzinfo=None)
+    slot_end = (as_utc(eval_started_at) + timedelta(hours=end_hour)).replace(tzinfo=None)
+
+    # Count sessions created within slot time range
+    sessions = session.exec(
+        select(IntelligentEvalSessionDB).where(
+            IntelligentEvalSessionDB.eval_id == eval_id,
+            IntelligentEvalSessionDB.created_at >= slot_start,
+            IntelligentEvalSessionDB.created_at < slot_end,
+        )
+    ).all()
+
+    return len(sessions)
+
+
+def _has_high_severity_issues(eval_id: str, session: Session) -> bool:
+    """Check if any completed session has high severity issues.
+
+    Returns:
+        True if any session's verdict contains high severity issue
+    """
+    completed_sessions = session.exec(
+        select(IntelligentEvalSessionDB).where(
+            IntelligentEvalSessionDB.eval_id == eval_id,
+            IntelligentEvalSessionDB.status == "completed",
+        )
+    ).all()
+
+    for sess in completed_sessions:
+        verdict = sess.get_verdict()
+        if verdict and verdict.get("severity") == "high":
+            return True
+
+    return False
+
+
+def make_decision(eval_db: IntelligentEvalDB, session: Session) -> Decision:
+    """Make a decision based on current evaluation state.
+
+    Args:
+        eval_db: Evaluation database record
+        session: Database session
+
+    Returns:
+        Decision object
+    """
+    # Check if eval is still executing
+    if eval_db.status != IntelligentEvalStatus.EXECUTING.value:
+        return Decision(
+            DecisionType.WAIT,
+            f"评估状态为 {eval_db.status}，不在执行中",
+            {"status": eval_db.status},
+        )
+
+    # Check if eval has plan and started_at
+    if not eval_db.plan or not eval_db.started_at:
+        return Decision(
+            DecisionType.WAIT,
+            "评估缺少计划或未开始",
+            {"has_plan": bool(eval_db.plan), "has_started_at": bool(eval_db.started_at)},
+        )
+
+    plan = eval_db.get_plan()
+    time_distribution = plan.get("time_distribution", [])
+    estimated_sessions = plan.get("estimated_sessions", 0)
+
+    # Calculate current offset
+    current_offset = utc_now() - as_utc(eval_db.started_at)
+    current_hours = current_offset.total_seconds() / 3600
+
+    # Get all sessions
+    all_sessions = session.exec(
+        select(IntelligentEvalSessionDB).where(IntelligentEvalSessionDB.eval_id == eval_db.id)
+    ).all()
+
+    completed_sessions = [s for s in all_sessions if s.status == "completed"]
+
+    # Check if all sessions completed and estimated reached
+    if len(completed_sessions) >= estimated_sessions:
+        return Decision(
+            DecisionType.START_ANALYSIS,
+            f"所有 {estimated_sessions} 个会话已完成，开始分析",
+            {
+                "completed_sessions": len(completed_sessions),
+                "estimated_sessions": estimated_sessions,
+            },
+        )
+
+    # Get current time slot
+    current_slot = _get_current_slot(time_distribution, current_offset)
+
+    if current_slot is None:
+        return Decision(
+            DecisionType.WAIT,
+            f"当前时间偏移 {current_hours:.1f}h 不在任何时段内",
+            {"current_offset_hours": current_hours},
+        )
+
+    # Check if current slot has deficit
+    slot_name = current_slot.get("time_slot", "")
+    expected_sessions = current_slot.get("sessions", 0)
+    current_sessions = _count_sessions_in_slot(eval_db.id, current_slot, eval_db.started_at, session)
+
+    deficit = expected_sessions - current_sessions
+
+    if deficit > 0:
+        return Decision(
+            DecisionType.EXECUTE_SESSION,
+            f"时段 {slot_name} 欠账 {deficit} 个会话",
+            {
+                "current_slot": slot_name,
+                "expected_sessions": expected_sessions,
+                "current_sessions": current_sessions,
+                "deficit": deficit,
+            },
+        )
+
+    # Check if any high severity issues found
+    if _has_high_severity_issues(eval_db.id, session):
+        return Decision(
+            DecisionType.EXECUTE_SESSION,
+            "发现高严重度问题，需要深入挖掘",
+            {"has_high_severity": True},
+        )
+
+    # No deficit, no high severity issues, wait
+    return Decision(
+        DecisionType.WAIT,
+        f"时段 {slot_name} 无欠账，等待下一时段",
+        {
+            "current_slot": slot_name,
+            "expected_sessions": expected_sessions,
+            "current_sessions": current_sessions,
+        },
+    )
+
+
+def is_eval_completed(eval_db: IntelligentEvalDB, session: Session) -> bool:
+    """Check if evaluation is completed.
+
+    An eval is completed when:
+    1. All sessions are completed
+    2. Report is submitted
+
+    Returns:
+        True if eval is completed
+    """
+    if eval_db.status != IntelligentEvalStatus.EXECUTING.value:
+        return False
+
+    if not eval_db.plan:
+        return False
+
+    plan = eval_db.get_plan()
+    estimated_sessions = plan.get("estimated_sessions", 0)
+
+    # Check if all sessions completed
+    all_sessions = session.exec(
+        select(IntelligentEvalSessionDB).where(IntelligentEvalSessionDB.eval_id == eval_db.id)
+    ).all()
+
+    completed_sessions = [s for s in all_sessions if s.status == "completed"]
+
+    if len(completed_sessions) < estimated_sessions:
+        return False
+
+    # Check if report submitted
+    if not eval_db.report:
+        return False
+
+    return True
diff --git a/tests/integration/test_intelligent_eval_e2e.py b/tests/integration/test_intelligent_eval_e2e.py
new file mode 100644
index 0000000..4f269f5
--- /dev/null
+++ b/tests/integration/test_intelligent_eval_e2e.py
@@ -0,0 +1,183 @@
+"""End-to-end test for intelligent eval with cron pool."""
+
+from datetime import timedelta
+
+import pytest
+from sqlmodel import Session, select
+
+from agenteval.intelligent_eval import task_queue
+from agenteval.intelligent_eval.decision import DecisionType, is_eval_completed, make_decision
+from agenteval.intelligent_eval.models import IntelligentEvalStatus
+from agenteval.storage.db import (
+    IntelligentEvalDB,
+    IntelligentEvalDecisionLogDB,
+    IntelligentEvalSessionDB,
+    IntelligentEvalTaskQueueDB,
+    utc_now,
+)
+
+
+def test_end_to_end_eval_lifecycle(db_session: Session):
+    """Test end-to-end eval lifecycle: create -> scan -> decide -> execute -> complete."""
+    # 1. Create eval
+    eval_db = IntelligentEvalDB(
+        name="test-eval",
+        target_id="target1",
+        status=IntelligentEvalStatus.EXECUTING.value,
+        started_at=utc_now() - timedelta(hours=8, minutes=30),
+    )
+    eval_db.set_plan({
+        "time_distribution": [{"time_slot": "8-10h", "sessions": 2}],
+        "estimated_sessions": 2,
+    })
+    db_session.add(eval_db)
+    db_session.commit()
+
+    # 2. Scan and enqueue tasks
+    enqueued = task_queue.scan_and_enqueue_tasks(db_session)
+    assert enqueued == 1
+
+    # Verify task created
+    task = db_session.exec(
+        select(IntelligentEvalTaskQueueDB).where(IntelligentEvalTaskQueueDB.eval_id == eval_db.id)
+    ).first()
+    assert task is not None
+    assert task.status == "pending"
+    assert task.reason == "slot_due"
+
+    # 3. Worker makes decision (should be EXECUTE_SESSION)
+    decision = make_decision(eval_db, db_session)
+    assert decision.decision_type == DecisionType.EXECUTE_SESSION
+    assert "欠账" in decision.reason
+
+    # Record decision log
+    log = IntelligentEvalDecisionLogDB(
+        eval_id=eval_db.id,
+        decision_type=decision.decision_type.value,
+        reason=decision.reason,
+        cron_id="cron-123",
+    )
+    log.set_context(decision.context)
+    db_session.add(log)
+    db_session.commit()
+
+    # 4. Execute session (simulate)
+    session_db = IntelligentEvalSessionDB(
+        eval_id=eval_db.id,
+        target_id="target1",
+        status="running",
+        created_at=utc_now() - timedelta(minutes=20),
+    )
+    db_session.add(session_db)
+    db_session.commit()
+
+    # 5. Complete session
+    session_db.status = "completed"
+    session_db.set_verdict({"severity": "medium", "issues": ["minor issue"]})
+    db_session.commit()
+
+    # 6. Worker makes decision again (should still be EXECUTE_SESSION, deficit = 1)
+    decision2 = make_decision(eval_db, db_session)
+    assert decision2.decision_type == DecisionType.EXECUTE_SESSION
+    assert "欠账 1 个会话" in decision2.reason
+
+    # 7. Execute second session
+    session_db2 = IntelligentEvalSessionDB(
+        eval_id=eval_db.id,
+        target_id="target1",
+        status="running",
+        created_at=utc_now() - timedelta(minutes=10),
+    )
+    db_session.add(session_db2)
+    db_session.commit()
+
+    session_db2.status = "completed"
+    session_db2.set_verdict({"severity": "low", "issues": []})
+    db_session.commit()
+
+    # 8. Worker makes decision again (should be START_ANALYSIS)
+    decision3 = make_decision(eval_db, db_session)
+    assert decision3.decision_type == DecisionType.START_ANALYSIS
+    assert "所有 2 个会话已完成" in decision3.reason
+
+    # 9. Submit report
+    eval_db.set_report({"summary": "Test report", "findings": []})
+    db_session.commit()
+
+    # 10. Check if eval is completed
+    assert is_eval_completed(eval_db, db_session) is True
+
+    # 11. Verify decision logs
+    logs = db_session.exec(
+        select(IntelligentEvalDecisionLogDB)
+        .where(IntelligentEvalDecisionLogDB.eval_id == eval_db.id)
+        .order_by(IntelligentEvalDecisionLogDB.created_at)
+    ).all()
+    assert len(logs) >= 1  # At least one decision log
+
+
+def test_decision_logs_complete_history(db_session: Session):
+    """Test that decision logs capture complete history."""
+    eval_db = IntelligentEvalDB(
+        name="test-eval",
+        target_id="target1",
+        status=IntelligentEvalStatus.EXECUTING.value,
+        started_at=utc_now() - timedelta(hours=8, minutes=30),
+    )
+    eval_db.set_plan({
+        "time_distribution": [{"time_slot": "8-10h", "sessions": 2}],
+        "estimated_sessions": 2,
+    })
+    db_session.add(eval_db)
+    db_session.commit()
+
+    # Simulate multiple worker ticks
+    decisions_made = []
+
+    # Tick 1: No sessions yet
+    decision1 = make_decision(eval_db, db_session)
+    decisions_made.append(decision1)
+    log1 = IntelligentEvalDecisionLogDB(
+        eval_id=eval_db.id,
+        decision_type=decision1.decision_type.value,
+        reason=decision1.reason,
+        cron_id="cron-123",
+    )
+    log1.set_context(decision1.context)
+    db_session.add(log1)
+    db_session.commit()
+
+    # Tick 2: One session created
+    session_db = IntelligentEvalSessionDB(
+        eval_id=eval_db.id,
+        target_id="target1",
+        status="running",
+        created_at=utc_now() - timedelta(minutes=20),
+    )
+    db_session.add(session_db)
+    db_session.commit()
+
+    decision2 = make_decision(eval_db, db_session)
+    decisions_made.append(decision2)
+    log2 = IntelligentEvalDecisionLogDB(
+        eval_id=eval_db.id,
+        decision_type=decision2.decision_type.value,
+        reason=decision2.reason,
+        cron_id="cron-123",
+    )
+    log2.set_context(decision2.context)
+    db_session.add(log2)
+    db_session.commit()
+
+    # Verify logs
+    logs = db_session.exec(
+        select(IntelligentEvalDecisionLogDB)
+        .where(IntelligentEvalDecisionLogDB.eval_id == eval_db.id)
+        .order_by(IntelligentEvalDecisionLogDB.created_at)
+    ).all()
+
+    assert len(logs) == 2
+    assert logs[0].decision_type == DecisionType.EXECUTE_SESSION.value
+    assert logs[1].decision_type == DecisionType.EXECUTE_SESSION.value
+    assert logs[0].get_context()["deficit"] == 2
+    assert logs[1].get_context()["deficit"] == 1
diff --git a/tests/unit/test_intelligent_eval_decision.py b/tests/unit/test_intelligent_eval_decision.py
new file mode 100644
index 0000000..8ab6b7f
--- /dev/null
+++ b/tests/unit/test_intelligent_eval_decision.py
@@ -0,0 +1,286 @@
+"""Unit tests for worker decision logic."""
+
+from datetime import timedelta
+
+import pytest
+from sqlmodel import Session
+
+from agenteval.intelligent_eval.decision import DecisionType, is_eval_completed, make_decision
+from agenteval.intelligent_eval.models import IntelligentEvalStatus
+from agenteval.storage.db import IntelligentEvalDB, IntelligentEvalSessionDB, utc_now
+
+
+def test_decision_wait_not_executing(db_session: Session):
+    """Test decision is WAIT when eval is not executing."""
+    eval_db = IntelligentEvalDB(
+        name="test",
+        target_id="target1",
+        status=IntelligentEvalStatus.COMPLETED.value,
+    )
+    db_session.add(eval_db)
+    db_session.commit()
+
+    decision = make_decision(eval_db, db_session)
+    assert decision.decision_type == DecisionType.WAIT
+    assert "不在执行中" in decision.reason
+
+
+def test_decision_wait_no_plan(db_session: Session):
+    """Test decision is WAIT when eval has no plan."""
+    eval_db = IntelligentEvalDB(
+        name="test",
+        target_id="target1",
+        status=IntelligentEvalStatus.EXECUTING.value,
+        started_at=utc_now(),
+    )
+    db_session.add(eval_db)
+    db_session.commit()
+
+    decision = make_decision(eval_db, db_session)
+    assert decision.decision_type == DecisionType.WAIT
+    assert "缺少计划" in decision.reason
+
+
+def test_decision_start_analysis_all_completed(db_session: Session):
+    """Test decision is START_ANALYSIS when all sessions completed."""
+    eval_db = IntelligentEvalDB(
+        name="test",
+        target_id="target1",
+        status=IntelligentEvalStatus.EXECUTING.value,
+        started_at=utc_now() - timedelta(hours=10),
+    )
+    eval_db.set_plan({
+        "time_distribution": [{"time_slot": "8-10h", "sessions": 2}],
+        "estimated_sessions": 2,
+    })
+    db_session.add(eval_db)
+    db_session.commit()
+
+    # Add 2 completed sessions
+    for _ in range(2):
+        session_db = IntelligentEvalSessionDB(
+            eval_id=eval_db.id,
+            target_id="target1",
+            status="completed",
+        )
+        db_session.add(session_db)
+    db_session.commit()
+
+    decision = make_decision(eval_db, db_session)
+    assert decision.decision_type == DecisionType.START_ANALYSIS
+    assert "所有 2 个会话已完成" in decision.reason
+
+
+def test_decision_execute_session_slot_deficit(db_session: Session):
+    """Test decision is EXECUTE_SESSION when current slot has deficit."""
+    eval_db = IntelligentEvalDB(
+        name="test",
+        target_id="target1",
+        status=IntelligentEvalStatus.EXECUTING.value,
+        started_at=utc_now() - timedelta(hours=8, minutes=30),  # 8.5 hours ago
+    )
+    eval_db.set_plan({
+        "time_distribution": [{"time_slot": "8-10h", "sessions": 2}],
+        "estimated_sessions": 2,
+    })
+    db_session.add(eval_db)
+    db_session.commit()
+
+    # Add 1 session created within the 8-10h slot (deficit = 1)
+    # Slot starts at started_at + 8h = 0.5h ago
+    session_db = IntelligentEvalSessionDB(
+        eval_id=eval_db.id,
+        target_id="target1",
+        status="running",
+        created_at=utc_now() - timedelta(minutes=20),  # 20 minutes ago, within slot
+    )
+    db_session.add(session_db)
+    db_session.commit()
+
+    decision = make_decision(eval_db, db_session)
+    assert decision.decision_type == DecisionType.EXECUTE_SESSION
+    assert "欠账 1 个会话" in decision.reason
+    assert decision.context["current_slot"] == "8-10h"
+    assert decision.context["deficit"] == 1
+
+
+def test_decision_execute_session_high_severity(db_session: Session):
+    """Test decision is EXECUTE_SESSION when high severity issue found."""
+    eval_db = IntelligentEvalDB(
+        name="test",
+        target_id="target1",
+        status=IntelligentEvalStatus.EXECUTING.value,
+        started_at=utc_now() - timedelta(hours=8, minutes=30),
+    )
+    eval_db.set_plan({
+        "time_distribution": [{"time_slot": "8-10h", "sessions": 2}],
+        "estimated_sessions": 3,  # Not all completed yet
+    })
+    db_session.add(eval_db)
+    db_session.commit()
+
+    # Add 2 completed sessions within the slot (no deficit)
+    for _ in range(2):
+        session_db = IntelligentEvalSessionDB(
+            eval_id=eval_db.id,
+            target_id="target1",
+            status="completed",
+            created_at=utc_now() - timedelta(minutes=20),  # Within slot
+        )
+        # One with high severity
+        session_db.set_verdict({"severity": "high", "issues": ["critical bug"]})
+        db_session.add(session_db)
+    db_session.commit()
+
+    decision = make_decision(eval_db, db_session)
+    assert decision.decision_type == DecisionType.EXECUTE_SESSION
+    assert "高严重度问题" in decision.reason
+
+
+def test_decision_wait_no_deficit(db_session: Session):
+    """Test decision is WAIT when no deficit in current slot."""
+    eval_db = IntelligentEvalDB(
+        name="test",
+        target_id="target1",
+        status=IntelligentEvalStatus.EXECUTING.value,
+        started_at=utc_now() - timedelta(hours=8, minutes=30),
+    )
+    eval_db.set_plan({
+        "time_distribution": [{"time_slot": "8-10h", "sessions": 2}],
+        "estimated_sessions": 3,  # Not all completed yet
+    })
+    db_session.add(eval_db)
+    db_session.commit()
+
+    # Add 2 sessions within the slot (no deficit)
+    for _ in range(2):
+        session_db = IntelligentEvalSessionDB(
+            eval_id=eval_db.id,
+            target_id="target1",
+            status="running",
+            created_at=utc_now() - timedelta(minutes=20),  # Within slot
+        )
+        db_session.add(session_db)
+    db_session.commit()
+
+    decision = make_decision(eval_db, db_session)
+    assert decision.decision_type == DecisionType.WAIT
+    assert "无欠账" in decision.reason
+
+
+def test_decision_wait_outside_slots(db_session: Session):
+    """Test decision is WAIT when current time is outside all slots."""
+    eval_db = IntelligentEvalDB(
+        name="test",
+        target_id="target1",
+        status=IntelligentEvalStatus.EXECUTING.value,
+        started_at=utc_now() - timedelta(hours=5),  # 5 hours ago, outside 8-10h
+    )
+    eval_db.set_plan({
+        "time_distribution": [{"time_slot": "8-10h", "sessions": 2}],
+        "estimated_sessions": 2,
+    })
+    db_session.add(eval_db)
+    db_session.commit()
+
+    decision = make_decision(eval_db, db_session)
+    assert decision.decision_type == DecisionType.WAIT
+    assert "不在任何时段内" in decision.reason
+
+
+def test_is_eval_completed_not_executing(db_session: Session):
+    """Test is_eval_completed returns False when eval is not executing."""
+    eval_db = IntelligentEvalDB(
+        name="test",
+        target_id="target1",
+        status=IntelligentEvalStatus.COMPLETED.value,
+    )
+    db_session.add(eval_db)
+    db_session.commit()
+
+    assert is_eval_completed(eval_db, db_session) is False
+
+
+def test_is_eval_completed_no_plan(db_session: Session):
+    """Test is_eval_completed returns False when eval has no plan."""
+    eval_db = IntelligentEvalDB(
+        name="test",
+        target_id="target1",
+        status=IntelligentEvalStatus.EXECUTING.value,
+    )
+    db_session.add(eval_db)
+    db_session.commit()
+
+    assert is_eval_completed(eval_db, db_session) is False
+
+
+def test_is_eval_completed_sessions_not_finished(db_session: Session):
+    """Test is_eval_completed returns False when not all sessions completed."""
+    eval_db = IntelligentEvalDB(
+        name="test",
+        target_id="target1",
+        status=IntelligentEvalStatus.EXECUTING.value,
+    )
+    eval_db.set_plan({"estimated_sessions": 2})
+    db_session.add(eval_db)
+    db_session.commit()
+
+    # Add only 1 completed session
+    session_db = IntelligentEvalSessionDB(
+        eval_id=eval_db.id,
+        target_id="target1",
+        status="completed",
+    )
+    db_session.add(session_db)
+    db_session.commit()
+
+    assert is_eval_completed(eval_db, db_session) is False
+
+
+def test_is_eval_completed_no_report(db_session: Session):
+    """Test is_eval_completed returns False when report not submitted."""
+    eval_db = IntelligentEvalDB(
+        name="test",
+        target_id="target1",
+        status=IntelligentEvalStatus.EXECUTING.value,
+    )
+    eval_db.set_plan({"estimated_sessions": 2})
+    db_session.add(eval_db)
+    db_session.commit()
+
+    # Add 2 completed sessions
+    for _ in range(2):
+        session_db = IntelligentEvalSessionDB(
+            eval_id=eval_db.id,
+            target_id="target1",
+            status="completed",
+        )
+        db_session.add(session_db)
+    db_session.commit()
+
+    assert is_eval_completed(eval_db, db_session) is False
+
+
+def test_is_eval_completed_true(db_session: Session):
+    """Test is_eval_completed returns True when all conditions met."""
+    eval_db = IntelligentEvalDB(
+        name="test",
+        target_id="target1",
+        status=IntelligentEvalStatus.EXECUTING.value,
+    )
+    eval_db.set_plan({"estimated_sessions": 2})
+    eval_db.set_report({"summary": "test report"})
+    db_session.add(eval_db)
+    db_session.commit()
+
+    # Add 2 completed sessions
+    for _ in range(2):
+        session_db = IntelligentEvalSessionDB(
+            eval_id=eval_db.id,
+            target_id="target1",
+            status="completed",
+        )
+        db_session.add(session_db)
+    db_session.commit()
+
+    assert is_eval_completed(eval_db, db_session) is True