diff --git a/codesage/governance/orchestrator.py b/codesage/governance/orchestrator.py
new file mode 100644
index 0000000..a0d6aaa
--- /dev/null
+++ b/codesage/governance/orchestrator.py
@@ -0,0 +1,330 @@
+"""治理任务编排引擎
+实现架构设计第 3.2.1 节的完整编排能力
+"""
+import asyncio
+import logging
+from typing import List, Dict, Optional, Set
+from dataclasses import dataclass, field
+from enum import Enum
+import networkx as nx  # 用于依赖图管理
+from queue import PriorityQueue
+
+from codesage.history.models import Issue
+from codesage.governance.patch_manager import Patch, PatchManager
+
+# Configure logging
+logger = logging.getLogger(__name__)
+
+class TaskStatus(Enum):
+    PENDING = "pending"
+    RUNNING = "running"
+    SUCCESS = "success"
+    FAILED = "failed"
+    RETRYING = "retrying"
+    SKIPPED = "skipped"
+    PENDING_REGENERATION = "pending_regeneration"
+
+class FailureReason(Enum):
+    TRANSIENT = "transient"
+    CONTEXTUAL = "contextual"
+    VALIDATION = "validation"
+    CONFLICT = "conflict"
+    UNKNOWN = "unknown"
+
+@dataclass
+class FixTask:
+    """修复任务（增强版）"""
+    id: str
+    issue: Issue
+    patch: Patch
+    priority: float  # 基于风险评分计算
+    dependencies: List[str] = field(default_factory=list)  # 依赖的任务 ID
+    status: TaskStatus = TaskStatus.PENDING
+    retry_count: int = 0
+    max_retries: int = 3
+    validation_config: Dict = field(default_factory=dict)
+
+    @property
+    def file_path(self) -> str:
+        return self.issue.file_path
+
+    def __lt__(self, other):
+        # High priority (higher value) comes first.
+        return self.priority > other.priority
+
+
+class FailureAnalyzer:
+    """失败原因分析器"""
+
+    def analyze_failure(self, task: FixTask, error: str) -> FailureReason:
+        """分析任务失败原因
+
+        分类:
+        - TRANSIENT: 临时性错误（网络超时、资源占用）→ 可重试
+        - CONTEXTUAL: 上下文不匹配（锚点未找到）→ 需 Jules 重新生成
+        - VALIDATION: 验证失败（测试不通过）→ 需人工介入
+        - CONFLICT: 代码冲突（Git merge 失败）→ 需人工解决
+        """
+        error_lower = error.lower()
+        if "timeout" in error_lower or "network" in error_lower:
+            return FailureReason.TRANSIENT
+        elif "anchor not found" in error_lower or "fuzzy match failed" in error_lower:
+            return FailureReason.CONTEXTUAL
+        elif "test failed" in error_lower or "validation failed" in error_lower:
+            return FailureReason.VALIDATION
+        elif "merge conflict" in error_lower:
+            return FailureReason.CONFLICT
+        else:
+            return FailureReason.UNKNOWN
+
+class GovernanceOrchestrator:
+    """治理任务编排器
+
+    核心能力（对齐架构设计）:
+    1. 依赖关系解析（DAG 构建）
+    2. 优先级队列调度
+    3. 失败重试策略
+    4. 并行执行支持
+    """
+
+    def __init__(
+        self,
+        patch_manager: PatchManager,
+        max_parallel: int = 3,
+        retry_strategy: str = "exponential_backoff",
+        failure_analyzer: Optional[FailureAnalyzer] = None
+    ):
+        self.patch_manager = patch_manager
+        self.max_parallel = max_parallel
+        self.retry_strategy = retry_strategy
+        self.failure_analyzer = failure_analyzer or FailureAnalyzer()
+
+        # 任务依赖图（使用 NetworkX）
+        self.task_graph = nx.DiGraph()
+
+        # 优先级队列
+        self.task_queue = PriorityQueue()
+
+        # 任务状态跟踪
+        self.tasks: Dict[str, FixTask] = {}
+
+    def add_task(self, task: FixTask):
+        """添加任务到编排器
+
+        自动分析依赖关系（基于文件依赖）
+        """
+        self.tasks[task.id] = task
+        self.task_graph.add_node(task.id, task=task)
+
+        # 添加依赖边
+        for dep_id in task.dependencies:
+            self.task_graph.add_edge(dep_id, task.id)
+
+        # 检查循环依赖
+        if not nx.is_directed_acyclic_graph(self.task_graph):
+            self.task_graph.remove_node(task.id)
+            del self.tasks[task.id]
+            raise ValueError(f"Circular dependency detected involving task {task.id}")
+
+    def build_execution_plan(self) -> List[List[str]]:
+        """构建执行计划（拓扑排序 + 并行分层）
+
+        Returns:
+            [
+                ["task1", "task2"],  # 第一批（可并行）
+                ["task3"],           # 第二批（依赖 task1/task2）
+                ["task4", "task5"]   # 第三批
+            ]
+        """
+        # 拓扑排序 (Check if graph is empty first)
+        if self.task_graph.number_of_nodes() == 0:
+            return []
+
+        if not nx.is_directed_acyclic_graph(self.task_graph):
+             raise ValueError("Cannot create execution plan: graph has cycles")
+
+        # 按层级分组（支持并行执行）
+        execution_plan = []
+
+        # Copy graph to destructively process it (or just track in-degrees)
+        remaining = set(self.tasks.keys())
+
+        while remaining:
+            # 找出当前可执行的任务（无未完成的依赖）
+            ready = [
+                tid for tid in remaining
+                if all(
+                    dep not in remaining
+                    for dep in self.task_graph.predecessors(tid)
+                )
+            ]
+
+            if not ready:
+                raise ValueError("Deadlock detected in task dependencies")
+
+            ready.sort(key=lambda tid: self.tasks[tid])
+
+            execution_plan.append(ready)
+            remaining -= set(ready)
+
+        return execution_plan
+
+    async def execute(self) -> Dict[str, TaskStatus]:
+        """执行所有任务（异步并行）
+
+        Returns:
+            任务 ID → 最终状态的映射
+        """
+        execution_plan = self.build_execution_plan()
+        results = {}
+
+        for batch in execution_plan:
+            # 并行执行当前批次（限制并发数）
+            batch_results = await self._execute_batch(batch)
+            results.update(batch_results)
+
+            # 检查失败任务
+            failed = [tid for tid, status in batch_results.items() if status == TaskStatus.FAILED]
+            if failed:
+                # 跳过依赖失败任务的后续任务
+                self._skip_dependent_tasks(failed)
+
+        return results
+
+    async def _execute_batch(self, task_ids: List[str]) -> Dict[str, TaskStatus]:
+        """并行执行一批任务"""
+
+        # 限制并发数
+        semaphore = asyncio.Semaphore(self.max_parallel)
+
+        async def execute_with_limit(tid):
+            async with semaphore:
+                if self.tasks[tid].status == TaskStatus.SKIPPED:
+                    return tid, TaskStatus.SKIPPED
+                return tid, await self._execute_single_task(tid)
+
+        # 并发执行
+        if not task_ids:
+            return {}
+
+        results_list = await asyncio.gather(*[
+            execute_with_limit(tid) for tid in task_ids
+        ])
+
+        return dict(results_list)
+
+    async def _execute_single_task(self, task_id: str) -> TaskStatus:
+        """执行单个任务（带重试）"""
+        task = self.tasks[task_id]
+        task.status = TaskStatus.RUNNING
+
+        while task.retry_count <= task.max_retries:
+            try:
+                # 应用补丁
+                loop = asyncio.get_running_loop()
+                result = await loop.run_in_executor(None, self.patch_manager.apply_patch_safe, task)
+
+                if result.success:
+                    task.status = TaskStatus.SUCCESS
+                    logger.info(f"Task {task.id} succeeded")
+                    return TaskStatus.SUCCESS
+                else:
+                    # 失败 → 分析原因
+                    error_msg = result.error or "Unknown error"
+                    reason = self.failure_analyzer.analyze_failure(task, error_msg)
+                    logger.warning(f"Task {task.id} failed (Attempt {task.retry_count + 1}/{task.max_retries + 1}). Reason: {reason}. Error: {error_msg}")
+
+                    if reason == FailureReason.TRANSIENT:
+                         # 临时错误 → 简单重试
+                        task.retry_count += 1
+                        if task.retry_count <= task.max_retries:
+                            task.status = TaskStatus.RETRYING
+                            await self._apply_retry_backoff(task)
+                        else:
+                            task.status = TaskStatus.FAILED
+                            return TaskStatus.FAILED
+
+                    elif reason == FailureReason.CONTEXTUAL:
+                        # 上下文问题 → 请求 Jules 重新生成
+                        logger.warning(f"Task {task_id} needs Jules re-generation")
+                        task.status = TaskStatus.PENDING_REGENERATION
+                        return TaskStatus.PENDING_REGENERATION
+
+                    elif reason == FailureReason.VALIDATION:
+                        # 验证失败 → 人工介入
+                        task.status = TaskStatus.FAILED
+                        self._create_manual_review_ticket(task, error_msg)
+                        return TaskStatus.FAILED
+
+                    elif reason == FailureReason.CONFLICT:
+                        # 冲突 -> Fail
+                        task.status = TaskStatus.FAILED
+                        return TaskStatus.FAILED
+
+                    else:
+                        # 其他/未知 -> 重试
+                        task.retry_count += 1
+                        if task.retry_count <= task.max_retries:
+                            task.status = TaskStatus.RETRYING
+                            await self._apply_retry_backoff(task)
+                        else:
+                            task.status = TaskStatus.FAILED
+                            return TaskStatus.FAILED
+
+            except Exception as e:
+                logger.error(f"Task {task_id} failed with exception: {e}")
+                task.retry_count += 1
+                if task.retry_count > task.max_retries:
+                    task.status = TaskStatus.FAILED
+                    return TaskStatus.FAILED
+                await self._apply_retry_backoff(task)
+
+        task.status = TaskStatus.FAILED
+        return TaskStatus.FAILED
+
+    async def _apply_retry_backoff(self, task: FixTask):
+        """应用重试退避策略"""
+
+        if self.retry_strategy == "exponential_backoff":
+            delay = 2 ** task.retry_count  # 2s, 4s, 8s
+        elif self.retry_strategy == "fixed":
+            delay = 5  # 固定 5 秒
+        else:
+            delay = 1
+
+        logger.info(f"Retrying task {task.id} after {delay}s (attempt {task.retry_count})")
+        await asyncio.sleep(delay)
+
+    def _skip_dependent_tasks(self, failed_task_ids: List[str]):
+        """跳过依赖失败任务的所有后续任务"""
+        for failed_id in failed_task_ids:
+            if failed_id not in self.task_graph:
+                continue
+            # 找出所有依赖该任务的后续任务
+            descendants = nx.descendants(self.task_graph, failed_id)
+            for desc_id in descendants:
+                if self.tasks[desc_id].status == TaskStatus.PENDING:
+                    self.tasks[desc_id].status = TaskStatus.SKIPPED
+                    logger.warning(f"Skipping task {desc_id} due to failed dependency {failed_id}")
+
+    def _create_manual_review_ticket(self, task: FixTask, error: str):
+        """创建人工审查工单（可集成 JIRA/GitHub Issues）"""
+
+        # Issue model has 'description', not 'message'
+        issue_desc = getattr(task.issue, 'description', 'No description')
+        if not issue_desc:
+            # Fallback if description is empty or None
+             issue_desc = 'No description'
+
+        ticket = {
+            "title": f"CodeSnapAI: Manual review needed for {task.id}",
+            "description": f"""
+                Task ID: {task.id}
+                Issue: {issue_desc}
+                Error: {error}
+                File: {task.issue.file_path}:{task.issue.line_number}
+            """,
+            "labels": ["codesnapai", "manual-review", "governance"]
+        }
+        # TODO: 集成 GitHub Issues API
+        logger.info(f"Created manual review ticket: {ticket}")