feat: TODO DB迁移+no_agent执行器+阻塞升级机制

- 创建 mofin.db → todos 表（id/title/status/priority/fix_action/retry机制）
- 创建 self_todo_executor.py：no_agent脚本，纯代码逻辑
  - 有fix_action→执行→验证→标记completed
  - 无fix_action→标记blocked→输出通知
  - 失败重试3次→超限标blocked
  - 新blocked项首次输出后缓存不重复
- 修改 morning_health_check：
  - TODO写入DB取代JSON（sqlite3，无row_factory依赖）
  - 去重：含completed查重
  - 输出阻塞TODO列表
- 替换cron：LLM cron(2h间距) → no_agent(10min间距)
- 修复：deliver=origin两任务改为local
- 清理：废弃todo.json退役

scripts/morning_health_check.py

@@ -134,18 +134,18 @@ def write_todos_for_issues():
                 level = issue["level"]
                 pri = todo_priority.get(level, "medium")
                 
-                # 去重：检查是否已有相同 title
-                existing = conn.execute(
-                    "SELECT id, status FROM todos WHERE title=? AND status IN ('pending','in_progress','blocked')",
+                # 去重：检查是否已存在（含completed的也要查，避免重复加）
+                r_exist = conn.execute(
+                    "SELECT id, status FROM todos WHERE title=?",
                     (title,)
                 ).fetchone()
                 
-                if existing:
-                    if existing["status"] == "blocked":
+                if r_exist:
+                    if r_exist[1] == "blocked":
                         # 已阻塞的重新打开
                         conn.execute(
                             "UPDATE todos SET status='pending', priority=?, note='已重新打开', updated_at=CURRENT_TIMESTAMP WHERE id=?",
-                            (pri, existing["id"])
+                            (pri, r_exist[0])
                         )
                 else:
                     # 生成fix_action（如果可推导）
@@ -160,18 +160,15 @@ def write_todos_for_issues():
                     new_count += 1
             
             conn.commit()
-            conn.close()
             
             if new_count > 0:
                 print()
                 print(f"📋 已加入TODO({new_count}条):")
-                # 重新查刚插入的pending记录
-                cur2 = conn.cursor()
-                for r2 in cur2.execute(
+                for r2 in conn.execute(
                     "SELECT title, priority FROM todos WHERE status='pending' AND source='health_check' "
                     "ORDER BY created_at DESC LIMIT ?", (new_count,)
                 ).fetchall():
-                    print(f"  [{r2['priority']}] {r2['title'][:70]}")
+                    print(f"  [{r2[1]}] {r2[0][:70]}")
             conn.close()
         except Exception as e:
             print(f"  TODO写入异常: {e}")
@@ -660,6 +657,22 @@ def main():
             if entry["level"] in ("critical", "error"):
                 print(f"  [{entry['level'].upper()}] {entry['category']}: {entry['msg']}")
     
+    # 检查是否有阻塞的 TODO 需要知微处理
+    try:
+        conn = sqlite3.connect(str(DB_PATH))
+        blocked = conn.execute(
+            "SELECT id, title, priority, created_at FROM todos WHERE status='blocked' "
+            "ORDER BY CASE priority WHEN 'high' THEN 0 WHEN 'medium' THEN 1 ELSE 2 END, created_at ASC"
+        ).fetchall()
+        if blocked:
+            print()
+            print("⛔ 阻塞TODO（需知微人工处理）:")
+            for b in blocked:
+                print(f"  [{b[2]}] #{b[0]} {b[1][:70]} ({b[3][:10]})")
+        conn.close()
+    except:
+        pass
+    
     # 将异常写入 TODO 系统
     write_todos_for_issues()