fix: 连续上榜改为严格N日每天上榜，缺一日都不行

提示词：
- SYSTEM_PROMPT 重写为可执行原则（事实优先/时间锚定/判断挂数字/禁用套话）＋输出格式契约
- DAILY_ANALYSIS_PROMPT 重排注意力：取数指令置顶（要求同轮并行调用 4 个核心工具）、
  写作规约 8 条、环比快照居中、前日报告全文置底并改为「今日数据优先+核对昨日展望」
- 新增单位口径、符号规范（+/- 是前端红涨绿跌着色依据）、不荐股等硬规约
- 移除已废弃的「今日定调」摘要规则（与分段指令、前端渲染互相冲突）
- 八章重构：板块资金流独立为「四、资金与筹码」，重复的「关注方向+下个交易日建议」
  合并为「七、明日展望」

生成链路：
- 修复分段生成未回灌前文，后段看不到前段导致重复铺数据、数字口径打架
- 新增 REQUIRED_TOOLS 校验：核心工具不全时补一轮并在 user 消息点名缺失项
- 修复 _extract_summary 被 # 标题行占掉，改为剥离标题与 Markdown 标记后截断

AGENTS.md 同步八章结构与 prompt 工程约定

.gitignore

@@ -123,3 +123,6 @@ tmp/
 /.core.hmbtNy
 /.core.dump
 .v2-demo-backup/
+
+# 后端运行时数据（本地启动生成：SQLite 库、管理密码等）
+backend/data/

AGENTS.md

@@ -55,8 +55,9 @@
 - 生成分享链接时复用已有短链，避免每次生成新链接
 - 全页面适配移动端：响应式字体、间距、布局断点（sm/md/lg）
 - K线图历史数据获取失败时直接报错，禁止使用模拟数据降级（避免"刷新数据变化"问题）
-- K线天数逻辑：默认3个月（90天），自选日期到现在超过3个月则从自选日期开始
-- K线图使用 recharts Brush 组件实现移动端缩放和滑动查看
+- K线天数逻辑：日K一次拉取近365天（供拖动回看），默认视口只展示最近90个自然日（3个月），向左拖动查看更早数据；分钟K一次拉取320根，铺满展示
+- K线图基于 TradingView Lightweight Charts v5 实现（蜡烛/折线 + MA/MACD/RSI 副图）；拖动/缩放查看数据，切换指标或显示方式保留当前视口（仅在数据集变化时重置）
+- K线图时间格式自定义：时间轴刻度 年→`YYYY`、月→`YYYY-MM`、日→`MM-DD`，十字光标→`YYYY-MM-DD`（`tickMarkFormatter` + `localization.timeFormatter`）；格式化必须用 UTC 取值（`getUTCFullYear` 等），因为时间戳按"北京时间墙钟视作 UTC"存储，用本地时区方法会错位一天
 - 板块标记：688开头=科创(红)、300/301开头=创业(紫)、920/8/4开头=北交(橙)，主板不显示标签；标记位置：搜索候选、详情页标题、集合卡片股票列表、分享页股票卡片标题
 - 详情页右上角外部跳转按钮：①"东方财富" `https://wap.eastmoney.com/quote/stock/{market}.{code}.html?appfenxiang=1`，market映射 688→6/60→1/其他→0；②"金十数据" `https://search.jin10.com/?keyword={股票名称URL编码}`（按名称搜索金十资讯）
 - 详情页资金流向模块：展示近30日资金流向分析，包含：
@@ -70,3 +71,25 @@
 - 资金流向数据获取失败时，前端通过 `fundFlowError` 状态显示错误信息，便于排查问题
 - ❌ 东方财富 API（push2his.eastmoney.com）在 Edge Function 环境被拒绝访问（peer closed connection），主力净流入数据无法获取，显示为 "-"
 - ✅ 替代方案：使用腾讯实时行情API的外盘(索引7)和内盘(索引8)数据计算净主动买入额 = (外盘 - 内盘) × 当前价 × 100（单位：元）
+## 每日 AI 分析（backend FastAPI，15:10 自动触发）
+
+- 报告结构约定：正文以 `# {trade_date} A股收盘分析报告` 开头后直接进入第一章，无"今日定调"摘要（已移除：prompt 不生成、前端无高亮框）；`summary` 为去掉标题行与 Markdown 标记后的正文前200字，供管理列表展示
+- 八章固定顺序：一、市场总览｜二、题材热点分析｜三、核心股追踪｜四、资金与筹码｜五、重要消息面｜六、海外市场与国内期货｜七、明日展望｜八、风险提示；分段归属：part1=一+二、part2=三+四、part3=五~八（原"四、关注方向/五、下个交易日建议"内容重复，已合并为"七、明日展望"，板块资金流从第二章挪出独立成第四章）
+- prompt 工程约定（改模板必读）：①取数指令放 prompt 最前（要求同一轮并行调用 4 个核心工具，代码用 `REQUIRED_TOOLS` 校验，缺则补一轮并在 user 消息里点名缺哪些）；②环比快照紧随其后、前日报告全文放最后（避免长文本把取数指令挤出注意力区）；③凡写数字的规则必须同时写"找不到就写 —（今日无数据）"，否则模型会用常识补全；④涨跌幅/环比/净额一律带 + 或 -，这是前端 `colorizeText` 红涨绿跌的着色依据；⑤模板用 `.format()` 渲染，正文中出现裸 `{}` 会直接报错
+- 分段生成必须回灌前文（`messages.append({"role":"assistant","content":part_content})`）：否则后段看不到前段，会重复铺同一段数据、数字口径打架；生成失败的占位段不回灌
+- 环比数据：`collect_ai_analysis` 每次运行先采集当日盘面快照（指数+涨跌统计 JSON）写入 `daily_market_stats` 表（UNIQUE trade_date，upsert），再读上一交易日快照格式化为 prompt 中的"环比数据段"；首跑无昨日数据时 AI 须如实标注"暂无昨日基准"
+- 连板梯队：`get_market_dashboard` 返回 `limitLadder` 字段（完整版，2连板以上全量+首板前8），涨停原因/封单金额来自涨停池按 thscode 匹配；看板 events 里的天梯仍只取每层前2（保持 UI 精简），两处用途不同不要合并
+- tokens 口径：`ai_reports.tokens_used` 只记"当次生成"消耗（约7万/份）；同日重新生成走 UPSERT，`generation_count` 自增、`updated_at` 刷新、`created_at` 保留首次生成时间；表有 UNIQUE(trade_date, report_type)，禁止改回 INSERT OR REPLACE 之外还要注意别用 lastrowid（UPSERT 更新时不可靠，须回查 id）
+- 旧库补列用 init_db 里的 try/except ALTER TABLE 轻量迁移（CREATE TABLE IF NOT EXISTS 不会更新旧表结构）
+
+- 补充数据源（services/market_extra.py）：①财经快讯 get_news=新浪7x24 zhibo.sina.cn（feed.list.rich_text）；②板块主力资金流=东财 push2 的 clist 接口（f62 主力净额），**必须用 push2delay.eastmoney.com 镜像**——push2 对部分客户端 TLS 指纹拦截（peer closed），查询串保持字面量 `+` 号；③两融=datacenter-web 的 RPTA_RZRQ_LSHJ（T+1 披露），两融余额=RZYE+RQYE
+- AI 报告重要消息面规则：必须基于 get_news 快讯，5-8条，格式【宏观/政策/行业/公司/海外】新闻——影响解读，禁止编造；板块资金面规则：必须引用 sectorFundFlow 的行业净流入/流出TOP3+概念TOP3
+- 截断续写：报告长导致 finish_reason=length 时，拼接已有内容并向 messages 追加"继续"指令让模型续写（最多3次），truncated 仅在续写后仍截断时为 true；call_llm 读超时 300s（续写携带全部上下文）
+- 快照入库防护：_capture_market_snapshot 校验 indices 非空且涨跌统计不全 0，fuyao 失败时跳过入库，防止空快照污染环比
+- 调试注意：独立脚本跑 backend 代码必须显式 `load_dotenv("/path/to/repo/.env")`——fuyao_apikey 在仓库根目录 .env（uvicorn 靠 --env-file 参数加载），backend/.env 只有 MX keys；且 stdin 脚本里 load_dotenv() 无参调用会因 frame 断言报错，须显式传路径
+
+- 第三档呈现：报告页按 `## ` 二级标题拆分为多张卡片渲染（`splitReport`），首卡含 # 标题+定调引用块；表格单元格数字按 A股惯例红涨绿跌（`colorizeChildren` 只给带 +/- 号的数字着色）；标题栏显示"总第 N 期"（`issue_number`，latest 接口用 COUNT(trade_date<=) 子查询计算）
+- 海外指数/国内期货：`fetch_global_markets` 走 push2delay 的 ulist.np（海外 secid `100.NDX` 等）+ clist（期货 fs `m:8/113/142/114/115`，只取名称含"主连/主力合约"且排除"次主连"，按成交额降序取前12）；接入 dashboard `globalMarkets` 字段
+- ⚠️ openteam 网关对 LLM 单请求有约 120s 硬超时（超时返回 502 或空 SSE），GLM reasoning 长报告一次生成必死。解法：`call_llm` 全流式（SSE）+ `collect_ai_analysis` 分三段生成（REPORT_PARTS，每段约1200-1600字，各重试3次带退避，失败占位不阻塞）；工具轮拿到数据后立即 break 进分段（再问一轮只会空转120s）；`thinking:{type:disabled}` 参数网关返回400不可用
+- 工具结果必须瘦身：`get_active_core_stocks` 只保留出现次数前25只+题材前5（全量145KB会撑爆上下文）；get_market_dashboard 全量约13KB可接受
+- glm-5.3-flash 空返回特征：SSE 200 但只有 reasoning_content 无 content/tool_calls/finish_reason（可能流满120s被掐），按空轮次处理重试即可

backend/database.py

@@ -82,9 +82,20 @@ CREATE TABLE IF NOT EXISTS ai_reports (
   tools_used TEXT,
   model TEXT NOT NULL,
   tokens_used INTEGER,
+  generation_count INTEGER NOT NULL DEFAULT 1,
+  llm_calls INTEGER,
   created_at TEXT NOT NULL DEFAULT (datetime('now','localtime')),
+  updated_at TEXT,
   UNIQUE(trade_date, report_type)
 );
+
+CREATE TABLE IF NOT EXISTS daily_market_stats (
+  id INTEGER PRIMARY KEY AUTOINCREMENT,
+  trade_date TEXT NOT NULL,
+  payload TEXT NOT NULL,
+  created_at TEXT NOT NULL DEFAULT (datetime('now','localtime')),
+  UNIQUE(trade_date)
+);
 """
 
 
@@ -107,6 +118,17 @@ def init_db():
     conn = get_connection()
     conn.executescript(SCHEMA_SQL)
 
+    # 轻量迁移：为已存在的旧表补列（CREATE TABLE IF NOT EXISTS 不会更新旧表结构）
+    for alter_sql in (
+        "ALTER TABLE ai_reports ADD COLUMN generation_count INTEGER NOT NULL DEFAULT 1",
+        "ALTER TABLE ai_reports ADD COLUMN updated_at TEXT",
+        "ALTER TABLE ai_reports ADD COLUMN llm_calls INTEGER",
+    ):
+        try:
+            conn.execute(alter_sql)
+        except sqlite3.OperationalError:
+            pass  # 列已存在
+
     # 清理过期缓存
     from services.cache import clean_expired
     clean_expired()

backend/dist/admin.html

@@ -334,7 +334,7 @@ function renderReports(list) {
   document.getElementById('recentBody').innerHTML = list.map(r =>
     `<tr>
       <td>${r.trade_date}</td>
-      <td><a href="/ai-analysis" target="_blank" style="color:var(--primary);text-decoration:none">${r.title || '-'}</a></td>
+      <td><a href="/ai-analysis?date=${r.trade_date}" target="_blank" style="color:var(--primary);text-decoration:none">${r.title || '-'}</a></td>
       <td>${r.tokens_used || 0}</td>
     </tr>`
   ).join('') || '<tr><td colspan="3" style="color:var(--muted-foreground);text-align:center">暂无数据</td></tr>';
@@ -349,7 +349,7 @@ function renderAllReports(list) {
       <td>${r.tokens_used || 0}</td>
       <td>${r.created_at || '-'}</td>
       <td style="white-space:nowrap">
-        <a href="/ai-analysis" target="_blank" class="btn btn-ghost btn-sm" style="padding:3px 10px;font-size:11px;text-decoration:none">查看</a>
+        <a href="/ai-analysis?date=${r.trade_date}" target="_blank" class="btn btn-ghost btn-sm" style="padding:3px 10px;font-size:11px;text-decoration:none">查看</a>
         <button class="btn btn-destructive btn-sm" style="padding:3px 10px;font-size:11px" onclick="deleteReport(${r.id})">删除</button>
       </td>
     </tr>`

backend/routes/admin.py

@@ -143,7 +143,8 @@ async def admin_list_reports(request: Request):
     conn = get_connection()
     try:
         rows = conn.execute(
-            "SELECT id, trade_date, report_type, title, summary, tools_used, model, tokens_used, created_at "
+            "SELECT id, trade_date, report_type, title, summary, tools_used, model, tokens_used, "
+            "generation_count, llm_calls, created_at, updated_at "
             "FROM ai_reports ORDER BY trade_date DESC, id DESC LIMIT 50"
         ).fetchall()
         items = []

backend/routes/ai_analysis.py

@@ -43,8 +43,11 @@ def _can_trigger(trade_date: str) -> tuple[bool, str]:
 async def get_latest_report():
     conn = get_connection()
     try:
+        # issue_number：按报告日期序数作为总期号
         row = conn.execute(
-            "SELECT * FROM ai_reports ORDER BY trade_date DESC, id DESC LIMIT 1"
+            """SELECT r.*,
+                      (SELECT COUNT(*) FROM ai_reports WHERE trade_date <= r.trade_date) AS issue_number
+               FROM ai_reports r ORDER BY r.trade_date DESC, r.id DESC LIMIT 1"""
         ).fetchone()
         if not row:
             raise HTTPException(status_code=404, detail="暂无分析报告")
@@ -60,7 +63,8 @@ async def list_reports():
     conn = get_connection()
     try:
         rows = conn.execute(
-            "SELECT id, trade_date, report_type, title, summary, tools_used, model, tokens_used, created_at "
+            "SELECT id, trade_date, report_type, title, summary, tools_used, model, tokens_used, "
+            "generation_count, llm_calls, created_at, updated_at "
             "FROM ai_reports ORDER BY trade_date DESC, id DESC LIMIT 50"
         ).fetchall()
         items = []
@@ -73,6 +77,29 @@ async def list_reports():
         conn.close()
 
 
+@router.get("/ai-analysis/by-date/{trade_date}", summary="按交易日获取 AI 分析报告")
+async def get_report_by_date(trade_date: str):
+    try:
+        datetime.strptime(trade_date, "%Y-%m-%d")
+    except ValueError:
+        raise HTTPException(status_code=400, detail="日期格式错误，应为 YYYY-MM-DD")
+    conn = get_connection()
+    try:
+        row = conn.execute(
+            """SELECT r.*,
+                      (SELECT COUNT(*) FROM ai_reports WHERE trade_date <= r.trade_date) AS issue_number
+               FROM ai_reports r WHERE r.trade_date = ? AND r.report_type = 'daily' LIMIT 1""",
+            (trade_date,)
+        ).fetchone()
+        if not row:
+            raise HTTPException(status_code=404, detail=f"{trade_date} 暂无分析报告")
+        item = dict_from_row(row)
+        item["toolsUsed"] = json.loads(item.pop("tools_used") or "[]")
+        return JSONResponse({"data": item}, headers=_NO_CACHE_HEADERS)
+    finally:
+        conn.close()
+
+
 @router.get("/ai-analysis/{report_id}", summary="AI 分析报告详情")
 async def get_report(report_id: int):
     conn = get_connection()

backend/routes/core_stocks.py

@@ -20,11 +20,11 @@ def _recent_trade_dates(conn, n: int = 10) -> list[str]:
     return [r["trade_date"] for r in reversed(rows)]
 
 
-@router.get("/active", summary="活跃核心股 + 最近10日涨幅矩阵")
-async def active_core_stocks():
+@router.get("/active", summary="活跃核心股 + 最近N日涨幅矩阵")
+async def active_core_stocks(days: int = Query(10, ge=1, le=30, description="交易日天数")):
     conn = get_connection()
     try:
-        dates = _recent_trade_dates(conn, 10)
+        dates = _recent_trade_dates(conn, days)
         if not dates:
             return JSONResponse({"dates": [], "stocks": []}, headers=_NO_CACHE_HEADERS)
 
@@ -90,6 +90,56 @@ async def active_core_stocks():
         conn.close()
 
 
+@router.get("/consecutive", summary="最近N日每天上榜的核心股")
+async def consecutive_core_stocks(days: int = Query(3, ge=2, le=10, description="交易日天数")):
+    conn = get_connection()
+    try:
+        rows = conn.execute(
+            "SELECT DISTINCT trade_date FROM daily_core_stocks ORDER BY trade_date DESC LIMIT ?",
+            (days,)
+        ).fetchall()
+        recent_dates = [r["trade_date"] for r in reversed(rows)]
+
+        if len(recent_dates) < days:
+            return JSONResponse({"days": days, "stocks": []}, headers=_NO_CACHE_HEADERS)
+
+        placeholders = ",".join("?" * len(recent_dates))
+        rows = conn.execute(
+            f"""SELECT trade_date, stock_code, stock_name, f3, cover_count
+                FROM daily_core_stocks
+                WHERE trade_date IN ({placeholders})""",
+            recent_dates,
+        ).fetchall()
+
+        stock_dates: dict[str, dict] = {}
+        for r in rows:
+            code = r["stock_code"]
+            s = stock_dates.setdefault(code, {
+                "stockCode": code,
+                "stockName": r["stock_name"],
+                "dates": set(),
+                "dailyGains": {},
+            })
+            s["dates"].add(r["trade_date"])
+            s["dailyGains"][r["trade_date"]] = r["f3"]
+
+        results = []
+        for code, s in stock_dates.items():
+            if len(s["dates"]) >= days:
+                results.append({
+                    "stockCode": code,
+                    "stockName": s["stockName"],
+                    "consecutiveDays": len(s["dates"]),
+                    "recentDates": sorted(s["dates"]),
+                    "dailyGains": s["dailyGains"],
+                })
+
+        results.sort(key=lambda x: -x["consecutiveDays"])
+        return JSONResponse({"days": days, "recentDates": recent_dates, "stocks": results}, headers=_NO_CACHE_HEADERS)
+    finally:
+        conn.close()
+
+
 @router.get("/history", summary="指定交易日核心股(含所属题材)")
 async def core_stock_history(date: str = Query(..., description="交易日 YYYY-MM-DD")):
     conn = get_connection()

backend/routes/market_dashboard.py

@@ -17,7 +17,7 @@ from datetime import datetime, timezone, timedelta
 from fastapi import APIRouter, HTTPException
 from fastapi.responses import JSONResponse
 
-from services import fuyao_client
+from services import fuyao_client, market_extra
 
 router = APIRouter()
 
@@ -163,12 +163,18 @@ async def _build_dashboard() -> dict:
             anomaly_data,
             limit_ladder_data,
             skyrocket_data,
+            sector_flow_data,
+            margin_data,
+            global_markets_data,
         ) = await asyncio.gather(
             fuyao_client.hot_stock_list("day"),
             fuyao_client.dragon_tiger_list("all"),
             fuyao_client.anomaly_analysis_list(["SHARP_RISE", "RAPID_RALLY", "LIMIT_UP"]),
             fuyao_client.limit_up_ladder(),
             fuyao_client.skyrocket_list("day"),
+            market_extra.fetch_sector_fund_flow(),
+            market_extra.fetch_margin_summary(),
+            market_extra.fetch_global_markets(),
             return_exceptions=True,
         )
     except Exception:
@@ -177,6 +183,9 @@ async def _build_dashboard() -> dict:
         anomaly_data = {}
         limit_ladder_data = {}
         skyrocket_data = {}
+        sector_flow_data = None
+        margin_data = None
+        global_markets_data = None
 
     # ── 解析指数 ──
     indices = []
@@ -291,6 +300,12 @@ async def _build_dashboard() -> dict:
         "auctionSignal": auction_signal,
     }
 
+    # 两融余额（交易所 T+1 披露），供看板展示、AI 分析与次日环比快照使用
+    if isinstance(margin_data, dict):
+        market_stats["marginBalanceYi"] = margin_data.get("balanceYi")
+        market_stats["marginChangeYi"] = margin_data.get("changeYi")
+        market_stats["marginDate"] = margin_data.get("date")
+
     # ── 行业强度榜（使用行业指数真实涨幅） ──
     sector_strength = []
     industries = industry_catalog if isinstance(industry_catalog, list) else []
@@ -485,13 +500,77 @@ async def _build_dashboard() -> dict:
         "detail": f"涨停 {limit_up_count} 跌停 {limit_down_count} 炸板 {break_count}",
     })
 
+    # ── 连板梯队（完整版，供 AI 分析用；上方 events 天梯只取前2保持看板精简） ──
+    # 涨停池里带 limit_up_reason / seal_money，按代码匹配给梯队个股
+    reason_by_code = {}
+    if isinstance(limit_up_data, dict):
+        for lu in limit_up_data.get("item", []) or []:
+            code = lu.get("thscode", "")
+            if code:
+                reason_by_code[code] = {
+                    "reason": (lu.get("limit_up_reason") or "").strip(),
+                    "sealWan": round(_safe_float(lu.get("seal_money")) / 10000),
+                }
+
+    _LADDER_LEVELS = [
+        ("seven_over", 7, "7连板+"),
+        ("six_board", 6, "6连板"),
+        ("five_board", 5, "5连板"),
+        ("four_board", 4, "4连板"),
+        ("three_board", 3, "3连板"),
+        ("two_board", 2, "2连板"),
+        ("first_board", 1, "首板"),
+    ]
+    limit_ladder = []
+    if isinstance(limit_ladder_data, dict):
+        ladder_items = limit_ladder_data.get("item", [])
+        if ladder_items:
+            today_boards = ladder_items[0].get("boards", {}) or {}
+            seen_keys = set()
+            for key, board_num, label in _LADDER_LEVELS:
+                seen_keys.add(key)
+                # 首板数量多（几十家）只取前8家；2连板以上全量保留
+                cap = 8 if board_num == 1 else None
+                entries = today_boards.get(key, []) or []
+                if cap is not None:
+                    entries = entries[:cap]
+                for item in entries:
+                    code = item.get("thscode", "")
+                    extra = reason_by_code.get(code, {})
+                    limit_ladder.append({
+                        "board": board_num,
+                        "label": label,
+                        "name": item.get("name", ""),
+                        "code": code,
+                        "reason": extra.get("reason", ""),
+                        "sealWan": extra.get("sealWan"),
+                    })
+            # 兜底：天梯返回了未知层级 key 时也带上（跳过已处理的已知 key）
+            for key, entries in today_boards.items():
+                if key in seen_keys:
+                    continue
+                for item in entries or []:
+                    code = item.get("thscode", "")
+                    extra = reason_by_code.get(code, {})
+                    limit_ladder.append({
+                        "board": _safe_float(item.get("board_num")) or 1,
+                        "label": f"{int(_safe_float(item.get('board_num')) or 1)}连板",
+                        "name": item.get("name", ""),
+                        "code": code,
+                        "reason": extra.get("reason", ""),
+                        "sealWan": extra.get("sealWan"),
+                    })
+
     # ── 组装结果 ──
     result = {
         "indices": indices,
         "marketStats": market_stats,
         "sectorStrength": sector_strength[:31],
         "conceptStrength": concept_strength[:10],
+        "sectorFundFlow": sector_flow_data if isinstance(sector_flow_data, dict) else None,
+        "globalMarkets": global_markets_data if isinstance(global_markets_data, dict) else None,
         "events": events,
+        "limitLadder": limit_ladder,
         "updateTime": datetime.now(BJT).strftime("%Y-%m-%d %H:%M:%S"),
     }
 

backend/services/ai_service.py

@@ -3,10 +3,15 @@
 负责：
 1. 调用 OpenAI 兼容 API 进行分析
 2. Function Calling 循环（AI 可主动获取数据）
-3. 保存报告到数据库
+3. 采集当日盘面快照（供次日环比）
+4. 保存报告到数据库
 """
 
+import asyncio
 import json
+import re
+import traceback
+
 import httpx
 from datetime import datetime, timezone, timedelta
 
@@ -16,7 +21,7 @@ from database import get_connection
 
 _CST = timezone(timedelta(hours=8))
 
-SYSTEM_PROMPT = """你是一位专业的A股市场分析师，擅长从数据中发现投资机会。
+SYSTEM_PROMPT = """你是一位专业的A股市场分析师，擅长从数据中发现投资机会。为收盘后的每日复盘报告供稿。读者是有一定经验、但没时间盯盘的短线与波段投资者。
 
 你的分析风格：
 - 数据驱动，基于真实数据而非主观臆断
@@ -25,68 +30,163 @@ SYSTEM_PROMPT = """你是一位专业的A股市场分析师，擅长从数据中
 - 风险提示，每次推荐都需说明风险点
 
 可用工具：
-- get_market_dashboard: 获取市场整体数据（指数/涨跌统计/行业强度/事件情报/市场温度）
+- get_market_dashboard: 获取市场整体数据（指数/涨跌统计/市场温度/连板梯队/行业强度/板块资金流/两融/海外指数与国内期货/事件情报）
 - get_theme_history: 获取指定日期的题材涨幅排行
 - get_active_core_stocks: 获取核心股追踪数据（10日涨幅矩阵+所属题材）
+- get_consecutive_core_stocks: 获取最近N个交易日每天都上榜的核心股（缺一日都不行），用于识别持续活跃的热点股
 - get_stock_quote: 获取个股实时行情
 - get_fund_flow: 获取个股资金流向
+- get_news: 获取财经快讯（新浪7x24，用于重要消息面）
 
 重要规则：
 1. 你必须先调用工具获取数据，然后基于数据进行分析
 2. 不要凭空编造数据，所有数据必须来自工具返回
 3. 如果工具返回空数据，如实说明数据不可用
-4. 分析完成后给出明确的结论和建议"""
+4. 分析完成后给出明确的结论和建议
 
-DAILY_ANALYSIS_PROMPT = """请对 {trade_date} 的A股市场进行收盘分析，生成一份完整的分析报告。
+工作原则：
+1. 事实优先：所有数字、股票名称与代码、涨停原因、封单金额必须能在工具返回中找到出处；工具没有的写「—（今日无数据）」，严禁凭记忆、常识或推理补全。
+2. 时间锚定：只分析给定交易日当天及之前的真实数据，禁止引用训练记忆中的行情、政策或题材。
+3. 判断必须挂数字：每个结论后面要有价格/涨跌幅/金额/家数支撑，不做无数据的定性判断。
+4. 说人话：结论先行、短句、少形容词；禁用「整体来看」「值得注意的是」「情绪有所回暖」这类无信息量的套话。
+
+输出格式（前端按 ## 二级标题切卡片渲染，红涨绿跌依赖数字符号）：
+- 章节标题严格用 `## 一、xxx` 形式，标题不超过 12 字
+- 涨跌幅、环比变化、资金净额等有方向的数字必须带 + 或 - 号（如 +2.31%、-15.6亿）；不带符号的数字不会被着色
+- 多维度对比一律用 Markdown 表格，表头标注单位"""
+
+# 一次工具轮就必须拿全的核心工具：模型常只调 1-2 个就开写，缺失会让对应章节数据空洞
+REQUIRED_TOOLS = ("get_market_dashboard", "get_theme_history",
+                  "get_active_core_stocks", "get_consecutive_core_stocks", "get_news")
+
+DAILY_ANALYSIS_PROMPT = """请对 {trade_date}（A股交易日） 的A股市场进行收盘分析，生成一份完整的分析报告。
 
 {prev_report_section}
+{prev_snapshot_section}
 
-请先调用以下工具获取数据：
-1. get_market_dashboard - 获取市场整体数据
-2. get_theme_history(date="{trade_date}") - 获取今日题材涨幅
-3. get_active_core_stocks - 获取核心股数据
+你的分析风格：
+- 数据驱动，基于真实数据而非主观臆断
+- 逻辑清晰，先总后分，层层递进
+- 观点明确，给出具体的操作建议
+- 风险提示，每次推荐都需说明风险点
 
-然后基于数据生成报告，结构如下：
+重要规则：
+1. 你必须先调用工具获取数据，然后基于数据进行分析
+2. 不要凭空编造数据，所有数据必须来自工具返回
+3. 如果工具返回空数据，如实说明数据不可用
+4. 分析完成后给出明确的结论和建议
 
-## 一、市场总览
-- 主要指数表现（上证、深证、创业板）
-- 涨跌家数统计
-- 市场温度评估
+■ 第一步 · 取数（必须一次并行完成）
+在同一次回复中并行发起以下 5 个工具调用，不要拆成多轮、不要只调其中一部分：
+1. get_market_dashboard() —— 指数、涨跌统计、市场温度、连板梯队 limitLadder、板块资金流 sectorFundFlow、两融、海外指数与国内期货 globalMarkets
+2. get_theme_history(date="{trade_date}") —— 今日题材涨幅榜
+3. get_active_core_stocks() —— 核心股 10 日涨幅矩阵与所属题材
+4. get_consecutive_core_stocks(days=3) —— 最近3个交易日每天都上榜的核心股（缺一日都不行）
+5. get_news(limit=30) —— 今日财经快讯
+只有在需要核实某只具体个股时，才额外调用 get_stock_quote / get_fund_flow，合计不超过 2 次。
 
-## 二、题材热点分析
-- 今日涨幅前5题材
-- 持续活跃的题材
-- 新兴热点题材
+■ 第二步 · 写作（以下规约对每一部分都生效）
+0. 报告标题（# 一级标题）之后的第一行，必须是一个引用块"定调摘要"，格式严格为：
+> 今日定调：<一句话核心结论，不超过80字，必须包含1-2个关键数字（如成交额、涨停家数、市场温度）>
+1. 事实约束：所有数字、个股代码与名称、涨停原因、封单金额必须出自工具返回，找不到就写「—（今日无数据）」，禁止推测或用常识填补；禁止用「预计/有望/大概率」代替数据。
+2. 环比约束：量能与情绪类数字必须给环比,若上方提供了"前一交易日盘面数据快照"（成交额、涨跌家数、涨停/跌停/炸板、市场温度、两融余额、主要指数），首次提到时必须写成「今日值（较昨日 ±变化）」；快照缺失时写「暂无昨日基准」，不得编造环比。
+3. 单位口径（全文统一，表头标注单位）：成交额→亿元（≥1万亿时写 x.xx 万亿）；资金净额→亿元；封单金额→万元（limitLadder 的 sealWan 字段）；指数→点。
+4. 符号规范：涨跌幅、环比变化、净流入/流出一律带 + 或 -，这是前端红涨绿跌的着色依据。
+5. 连板梯队：连板梯队必须完整呈现 get_market_dashboard 返回的 limitLadder：从最高连板到2连板逐级列表格，每只标注涨停原因（reason字段）与封单金额（sealWan，单位万，为空则不写）；首板只挑3-5只人气最高的点评。
+6. 板块资金：板块资金面必须引用 get_market_dashboard 返回的 sectorFundFlow：行业主力净流入TOP5、净流出TOP5、概念净流入TOP5（亿元，带符号），结合题材分析说明资金动向。
+7. 海外市场与国内期货：必须引用 get_market_dashboard 返回的 globalMarkets：overseas 为海外主要指数（纳斯达克/道琼斯/标普500/恒生/日经/富时），futures 为国内期货主力合约（按成交额降序，已含价格与涨跌幅）；点评与A股关联度高的品种（股指期货、原油、贵金属、黑色系），数据缺失则如实说明。
+8. 消息面：重要消息面必须基于 get_news 返回的快讯整理：挑5-8条对次日盘面影响最大的消息，每条格式为"【分类】一句话新闻 —— 一句影响解读"（分类用：宏观/政策/行业/公司/海外）；快讯中若没有某方面的重要消息，如实说明「今日无重要消息」，严禁编造工具中不存在的新闻。
+9. 适当使用表格展示数据对比。
+基于数据生成报告。报告共八章，固定顺序：一、市场总览｜二、题材热点分析｜三、核心股追踪｜四、资金与筹码｜五、重要消息面｜六、海外市场与国内期货｜七、下个交易日建议｜八、风险提示。
+同一段数据只允许在它归属的那一章出现：指数与量能在第一章、题材涨幅在第二章、连板与核心股在第三章、资金流与两融在第四章，其他章节只做结论引用，不重复铺数据。
+系统会把八章拆成三次生成，每次只写被指定的部分。
+"""
 
-## 三、核心股追踪
-- 连板股分析
-- 核心股表现
-- 龙头股辨识
+# 分段生成指令：网关对单次 LLM 请求有约120s硬超时，整篇报告一次生成必被掐断，
+# 故拆为三段（每段约1200-1600字），各自独立调用后拼接
+REPORT_PARTS = [
+    """现在写报告的【第1部分】，只输出这一部分，直接输出 Markdown，不要任何开场白或说明：
+1. 以 `# {title} A股收盘分析报告` 一级标题开头
+2. `## 一、市场总览`（指数表格、涨跌统计须环比、市场温度）
+3. `## 二、题材热点分析`（涨幅前5、持续活跃、新兴热点、退潮警示、板块主力资金流TOP3）
+全文控制在1600字以内。""",
+    """现在写报告的【第2部分】，只输出这一部分，直接输出 Markdown，不要重复之前内容：
+- `## 三、核心股追踪`（连板梯队完整表格：层级/股票/涨停原因/封单，首板挑3-5只人气股点评；连续上榜股票表格：股票/连续天数/近3日涨幅，点评持续活跃的核心标的；核心股表现；龙头辨识）
+- `## 四、资金与筹码`：sectorFundFlow 的行业净流入 TOP3 / 净流出 TOP3 / 概念净流入 TOP3（亿元，带符号）、两融余额与变化、事件情报中值得注意的筹码信号
+全文控制在 1500 字以内。""",
+    """前两部分已在上下文中，现在续写【第3部分】，直接输出 Markdown，不要复述前文、不要写过渡句：
+- `## 五、重要消息面`：5-8条，格式【分类】新闻——影响解读
+- `## 六、海外市场与国内期货`（点评对次日A股的影响）
+全文控制在1300字以内。""",
+    """现在写报告的【第3部分】，只输出这一部分，直接输出 Markdown，不要重复之前内容：
+- `## 七、下个交易日建议`（大盘预判、题材方向、核心股、仓位策略、潜在交易机会、规避方向，结合市场情绪、资金流向、热点题材、核心股表现、消息面、海外市场与国内期货、风险提示，给出明确的结论和建议。)
+- `## 八、风险提示`
+全文控制在1900字以内。""",
+]
+## 五、下个交易日建议`（大盘预判、题材方向、核心股、仓位策略、规避方向）
 
-## 四、关注方向
-- 明日值得关注的题材方向
-- 潜在的交易机会
 
-## 五、下个交易日建议
-- 明日大盘预判（支撑/压力位）
-- 建议关注的题材方向（2-3个）
-- 建议关注的核心股（附理由）
-- 操作策略（仓位建议、买卖时机）
-- 需要规避的方向
+async def _consume_sse(resp: httpx.Response) -> dict:
+    """消费 OpenAI 兼容 SSE 流，拼装为与非流式响应相同的结构"""
+    content_parts: list[str] = []
+    finish_reason = ""
+    usage: dict = {}
+    # tool_calls 按 index 拼装（流式下 arguments 分片到达）
+    tool_acc: dict[int, dict] = {}
 
-## 六、风险提示
-- 需要警惕的风险因素
-- 操作建议
+    async for line in resp.aiter_lines():
+        if not line.startswith("data:"):
+            continue
+        data = line[5:].strip()
+        if not data or data == "[DONE]":
+            continue
+        try:
+            chunk = json.loads(data)
+        except json.JSONDecodeError:
+            continue
+        if chunk.get("usage"):
+            usage = chunk["usage"]
+        choices = chunk.get("choices") or []
+        if not choices:
+            continue
+        delta = choices[0].get("delta") or {}
+        if delta.get("content"):
+            content_parts.append(delta["content"])
+        for tc in delta.get("tool_calls") or []:
+            idx = tc.get("index", 0)
+            slot = tool_acc.setdefault(idx, {"id": "", "type": "function",
+                                             "function": {"name": "", "arguments": ""}})
+            if tc.get("id"):
+                slot["id"] = tc["id"]
+            fn = tc.get("function") or {}
+            if fn.get("name"):
+                slot["function"]["name"] += fn["name"]
+            if fn.get("arguments"):
+                slot["function"]["arguments"] += fn["arguments"]
+        if choices[0].get("finish_reason"):
+            finish_reason = choices[0]["finish_reason"]
 
-请用 Markdown 格式输出，适当使用表格展示数据对比。"""
+    message: dict = {"role": "assistant", "content": "".join(content_parts) or None}
+    if tool_acc:
+        message["tool_calls"] = [
+            {"id": tool_acc[i]["id"], "type": "function",
+             "function": tool_acc[i]["function"]}
+            for i in sorted(tool_acc)
+        ]
+    return {"choices": [{"message": message, "finish_reason": finish_reason}], "usage": usage}
 
 
 async def call_llm(messages: list, tools: list = None) -> dict:
-    """调用 OpenAI 兼容 API"""
+    """调用 OpenAI 兼容 API（流式）。
+
+    必须用 stream：网关对非流式请求有约120s的代理超时，长生成会被 502 掐断；
+    流式下字节持续到达不会被判定超时。返回结构与非流式一致。
+    """
     async with httpx.AsyncClient() as client:
         payload = {
             "model": AI_MODEL,
             "messages": messages,
+            "stream": True,
         }
         if AI_MAX_TOKENS is not None:
             payload["max_tokens"] = AI_MAX_TOKENS
@@ -96,17 +196,40 @@ async def call_llm(messages: list, tools: list = None) -> dict:
             payload["tools"] = tools
             payload["tool_choice"] = "auto"
 
-        resp = await client.post(
-            f"{AI_API_BASE}/chat/completions",
-            headers={
-                "Authorization": f"Bearer {AI_API_KEY}",
-                "Content-Type": "application/json",
-            },
-            json=payload,
-            timeout=120,
-        )
-        resp.raise_for_status()
-        return resp.json()
+        max_attempts = 4
+        for attempt in range(1, max_attempts + 1):
+            try:
+                async with client.stream(
+                    "POST",
+                    f"{AI_API_BASE}/chat/completions",
+                    headers={
+                        "Authorization": f"Bearer {AI_API_KEY}",
+                        "Content-Type": "application/json",
+                        "Accept": "text/event-stream",
+                    },
+                    json=payload,
+                    timeout=600,
+                ) as resp:
+                    if resp.status_code == 429 and attempt < max_attempts:
+                        wait = min(30 * attempt, 90)
+                        print(f"[ai-service] LLM 429 限流，{wait}s 后重试（第 {attempt}/{max_attempts - 1} 次）")
+                        await asyncio.sleep(wait)
+                        continue
+                    if resp.status_code >= 500 and attempt < max_attempts:
+                        print(f"[ai-service] LLM {resp.status_code}，{min(15 * attempt, 60)}s 后重试")
+                        await asyncio.sleep(min(15 * attempt, 60))
+                        continue
+                    resp.raise_for_status()
+                    return await _consume_sse(resp)
+            except httpx.TransportError as e:
+                # 网络层错误（超时/断连）也值得重试
+                if attempt < max_attempts:
+                    wait = min(15 * attempt, 60)
+                    print(f"[ai-service] LLM 网络错误（{type(e).__name__}），{wait}s 后重试")
+                    await asyncio.sleep(wait)
+                    continue
+                raise
+        raise RuntimeError("LLM 调用重试次数耗尽")
 
 
 async def collect_ai_analysis(trade_date: str) -> dict:
@@ -119,72 +242,257 @@ async def collect_ai_analysis(trade_date: str) -> dict:
         {"id": int, "tokens_used": int, "tools_used": list}
     """
     prev_report_section = _get_prev_report_section(trade_date)
+    # 采集当日盘面快照（供次日环比），并读取上一交易日快照注入 prompt
+    await _capture_market_snapshot(trade_date)
+    prev_snapshot_section = _get_prev_snapshot_section(trade_date)
+
     messages = [
         {"role": "system", "content": SYSTEM_PROMPT},
         {"role": "user", "content": DAILY_ANALYSIS_PROMPT.format(
             trade_date=trade_date,
-            prev_report_section=prev_report_section
+            prev_report_section=prev_report_section,
+            prev_snapshot_section=prev_snapshot_section,
         )}
     ]
 
+    # ── 阶段一：工具轮（只获取数据，模型若直接开写报告则丢弃，由阶段二重写） ──
     tools_used = []
     total_tokens = 0
-    final_content = ""
+    llm_calls = 0
     was_truncated = False
-    max_rounds = 30  # 安全上限，正常分析约 3-8 轮
+    max_rounds = 8
+    tool_rounds = 0
 
     for i in range(max_rounds):
         response = await call_llm(messages, tools=TOOLS)
+        llm_calls += 1
         total_tokens += response.get("usage", {}).get("total_tokens", 0)
 
         choice = response["choices"][0]
         message = choice["message"]
-        messages.append(message)
-
         finish_reason = choice.get("finish_reason", "")
+        print(f"[ai-service] tool round {i}: finish={finish_reason}, "
+              f"content_len={len(message.get('content') or '')}, "
+              f"tool_calls={len(message.get('tool_calls') or [])}")
 
-        if finish_reason == "stop":
-            final_content = message.get("content") or ""
-            break
-
-        if finish_reason == "length":
-            was_truncated = True
-            final_content = message.get("content") or ""
-            if not final_content:
-                for msg in reversed(messages):
-                    if msg.get("role") == "assistant" and msg.get("content"):
-                        final_content = msg["content"]
-                        break
-            print(f"[ai-service] 警告：响应被截断 (finish_reason=length)")
-            break
-
-        if finish_reason == "tool_calls":
-            for tool_call in message.get("tool_calls", []):
+        if finish_reason == "tool_calls" and message.get("tool_calls"):
+            tool_rounds += 1
+            messages.append(message)
+            for tool_call in message["tool_calls"]:
                 func_name = tool_call["function"]["name"]
                 func_args = json.loads(tool_call["function"]["arguments"])
                 tools_used.append(func_name)
-
                 result = await execute_tool(func_name, func_args)
                 messages.append({
                     "role": "tool",
                     "tool_call_id": tool_call["id"],
                     "content": result
                 })
+            # 核心工具齐了就立即进分段写作（再问一轮只会空转120s）；
+            # 不齐则最多再补一轮，否则对应章节会无数据可写
+            if all(t in tools_used for t in REQUIRED_TOOLS):
+                break
+            if tool_rounds >= 2:
+                print(f"[ai-service] 核心工具仍不齐（已取：{sorted(set(tools_used))}），直接进入写作")
+                break
+            print(f"[ai-service] 核心工具不全，补一轮取数（{tool_rounds}/2）")
+            missing = [t for t in REQUIRED_TOOLS if t not in tools_used]
+            messages.append({
+                "role": "user",
+                "content": f"还缺以下数据，请立即调用后停止取数：{', '.join(missing)}"
+                           f"（get_theme_history 的 date 参数为 {trade_date}）"
+            })
+            continue
 
-    summary = final_content[:200].replace("\n", " ") if final_content else ""
-    report_id = _save_report(trade_date, final_content, summary, tools_used, total_tokens)
+        # 非工具轮（模型直接开写/空返回）：只要有工具结果就直接进入分段写作；
+        # 一轮工具都没拿到则重试
+        if tools_used:
+            break
+        print(f"[ai-service] 未获取到工具数据，重试（{i + 1}/{max_rounds}）")
+        await asyncio.sleep(5)
 
-    return {"id": report_id, "tokens_used": total_tokens, "tools_used": tools_used, "truncated": was_truncated}
+    if not tools_used:
+        raise RuntimeError("工具数据获取失败，无法生成报告")
+
+    # ── 阶段二：分段生成报告（绕开网关单请求约120s硬超时） ──
+    final_content = ""
+    for part_idx, part_prompt in enumerate(REPORT_PARTS):
+        part_content = ""
+        for attempt, backoff in ((1, 0), (2, 10), (3, 30)):
+            if backoff:
+                await asyncio.sleep(backoff)
+            part_messages = messages + [{
+                "role": "user",
+                "content": part_prompt.format(title=trade_date) if part_idx == 0 else part_prompt,
+            }]
+            response = await call_llm(part_messages)
+            llm_calls += 1
+            total_tokens += response.get("usage", {}).get("total_tokens", 0)
+            choice = response["choices"][0]
+            part_content = choice["message"].get("content") or ""
+            finish = choice.get("finish_reason", "")
+            print(f"[ai-service] part {part_idx + 1} attempt {attempt}: finish={finish}, len={len(part_content)}")
+            if part_content and finish in ("stop", "length"):
+                break
+            print(f"[ai-service] part {part_idx + 1} 生成异常，重试")
+        if not part_content:
+            was_truncated = True
+            print(f"[ai-service] 警告：part {part_idx + 1} 三次尝试均失败")
+            part_content = f"\n\n> （第{part_idx + 1}部分生成失败，请稍后重新生成）\n"
+        else:
+            # 回灌前文：后段看不到前段内容会导致重复铺数据、数字口径打架
+            messages.append({"role": "assistant", "content": part_content})
+        final_content += (final_content and "\n\n" or "") + part_content
+
+    summary = _extract_summary(final_content)
+    report_id = _save_report(trade_date, final_content, summary, tools_used, total_tokens, llm_calls)
+    print(f"[ai-service] 生成完成：LLM调用 {llm_calls} 次，tokens {total_tokens}")
+
+    return {"id": report_id, "tokens_used": total_tokens, "llm_calls": llm_calls, "tools_used": tools_used, "truncated": was_truncated}
 
 
-def _save_report(trade_date: str, content: str, summary: str, tools_used: list, tokens_used: int) -> int:
-    """保存报告到数据库"""
+def _extract_summary(content: str) -> str:
+    """摘要：去掉标题行与 Markdown 标记后截断 200 字（供管理列表展示）
+
+    正文首行是 `# XXXX A股收盘分析报告`，直接截断会让它占掉摘要大半。
+    """
+    text = re.sub(r"^#.*$", "", content or "", flags=re.MULTILINE)
+    text = re.sub(r"[>#*`|]", "", text)
+    return re.sub(r"\s+", " ", text).strip()[:200]
+
+
+def _num(v) -> str:
+    """快照数值安全转字符串"""
+    if isinstance(v, float):
+        return f"{v:g}"
+    return str(v if v is not None else "-")
+
+
+def _fmt_amount(v) -> str:
+    """成交额（元）→ 万亿/亿 可读格式"""
+    try:
+        v = float(v)
+    except (TypeError, ValueError):
+        return "-"
+    if v >= 1e12:
+        return f"{v / 1e12:.2f}万亿"
+    if v >= 1e8:
+        return f"{v / 1e8:.0f}亿"
+    return f"{v:.0f}元"
+
+
+def _fmt_index(idx: dict) -> str:
+    if not idx:
+        return "-"
+    try:
+        pct = round(float(idx.get("changePct", 0)), 2)
+    except (TypeError, ValueError):
+        pct = 0
+    sign = "+" if pct >= 0 else ""
+    return f"{idx.get('price', '-')}（{sign}{pct}%）"
+
+
+def _fmt_temperature(v) -> str:
+    """温度可能是 dict（score/label/factors），取分数与标签"""
+    if isinstance(v, dict):
+        score = v.get("score")
+        label = v.get("label") or ""
+        return f"{score}分{('（' + label + '）') if label else ''}"
+    return _num(v)
+
+
+async def _capture_market_snapshot(trade_date: str) -> bool:
+    """采集当日盘面快照入库（指数+涨跌统计），供次日分析做环比"""
+    try:
+        from routes.market_dashboard import _build_dashboard
+        data = await _build_dashboard()
+        stats = data.get("marketStats") or {}
+        # 数据有效性校验：fuyao 拉取失败时涨跌统计全 0，空快照会污染次日环比
+        if not data.get("indices") or (stats.get("upCount", 0) + stats.get("downCount", 0) == 0):
+            print(f"[ai-service] 盘面数据无效，跳过快照入库 {trade_date}")
+            return False
+        payload = json.dumps(
+            {"indices": data.get("indices", []), "marketStats": stats},
+            ensure_ascii=False,
+            default=str,
+        )
+        conn = get_connection()
+        try:
+            conn.execute(
+                """INSERT INTO daily_market_stats (trade_date, payload) VALUES (?, ?)
+                   ON CONFLICT(trade_date) DO UPDATE SET payload = excluded.payload""",
+                (trade_date, payload),
+            )
+            conn.commit()
+        finally:
+            conn.close()
+        return True
+    except Exception:
+        print(f"[ai-service] 市场快照采集失败 {trade_date}：")
+        traceback.print_exc()
+        return False
+
+
+def _get_prev_snapshot_section(trade_date: str) -> str:
+    """读取上一交易日盘面快照，格式化为 prompt 中的环比数据段"""
     conn = get_connection()
     try:
-        cursor = conn.execute(
-            """INSERT OR REPLACE INTO ai_reports
-               (trade_date, report_type, title, content, summary, tools_used, model, tokens_used)
-               VALUES (?, 'daily', ?, ?, ?, ?, ?, ?)""",
+        row = conn.execute(
+            "SELECT trade_date, payload FROM daily_market_stats WHERE trade_date < ? ORDER BY trade_date DESC LIMIT 1",
+            (trade_date,),
+        ).fetchone()
+        if not row:
+            return ""
+        try:
+            snap = json.loads(row["payload"])
+        except (TypeError, ValueError):
+            return ""
+        stats = snap.get("marketStats") or {}
+        indices = {i.get("name"): i for i in (snap.get("indices") or []) if isinstance(i, dict)}
+        idx_line = "、".join(
+            f"{name} {_fmt_index(indices.get(name))}"
+            for name in ("上证指数", "深证成指", "创业板指", "科创50")
+        )
+        margin_line = ""
+        if stats.get("marginBalanceYi"):
+            change = stats.get("marginChangeYi")
+            change_txt = ""
+            if change is not None:
+                sign = "+" if float(change) >= 0 else ""
+                change_txt = f"（较前一日 {sign}{_num(change)}亿）"
+            margin_line = f"\n- 两融余额：{_num(stats.get('marginBalanceYi'))}亿{change_txt}，数据日期 {stats.get('marginDate') or '-'}（T+1）"
+        return f"""以下是前一交易日（{row["trade_date"]}）的盘面数据快照，报告中的量能与情绪数字必须给出与它的环比对比：
+
+- 两市成交额：{_fmt_amount(stats.get("totalTurnover"))}
+- 上涨/下跌/平盘：{_num(stats.get("upCount"))}/{_num(stats.get("downCount"))}/{_num(stats.get("flatCount"))}，涨停 {_num(stats.get("limitUp"))} 家、跌停 {_num(stats.get("limitDown"))} 家、炸板 {_num(stats.get("limitBreak"))} 家（炸板率 {_num(stats.get("breakRate"))}%）
+- 强势/弱势股：{_num(stats.get("strongCount"))}/{_num(stats.get("weakCount"))}，市场宽度 {_num(stats.get("marketBreadth"))}%
+- 市场温度：{_fmt_temperature(stats.get("temperature"))}，竞价信号：{stats.get("auctionSignal") or "-"}{margin_line}
+- 指数收盘：{idx_line}
+
+---
+"""
+    finally:
+        conn.close()
+
+
+def _save_report(trade_date: str, content: str, summary: str, tools_used: list, tokens_used: int, llm_calls: int = 0) -> int:
+    """保存报告到数据库（同日重生成：覆盖内容、generation_count+1、tokens 记当次消耗）"""
+    conn = get_connection()
+    try:
+        conn.execute(
+            """INSERT INTO ai_reports
+               (trade_date, report_type, title, content, summary, tools_used, model, tokens_used, llm_calls, updated_at)
+               VALUES (?, 'daily', ?, ?, ?, ?, ?, ?, ?, datetime('now','localtime'))
+               ON CONFLICT(trade_date, report_type) DO UPDATE SET
+                 title = excluded.title,
+                 content = excluded.content,
+                 summary = excluded.summary,
+                 tools_used = excluded.tools_used,
+                 model = excluded.model,
+                 tokens_used = excluded.tokens_used,
+                 llm_calls = excluded.llm_calls,
+                 updated_at = excluded.updated_at,
+                 generation_count = ai_reports.generation_count + 1""",
             (
                 trade_date,
                 f"{trade_date} A股收盘分析",
@@ -193,10 +501,15 @@ def _save_report(trade_date: str, content: str, summary: str, tools_used: list,
                 json.dumps(tools_used),
                 AI_MODEL,
                 tokens_used,
-            )
+                llm_calls,
+            ),
         )
         conn.commit()
-        return cursor.lastrowid
+        row = conn.execute(
+            "SELECT id FROM ai_reports WHERE trade_date = ? AND report_type = 'daily'",
+            (trade_date,),
+        ).fetchone()
+        return row["id"]
     finally:
         conn.close()
 
@@ -213,7 +526,11 @@ def _get_prev_report_section(trade_date: str) -> str:
             return ""
         prev_date = row["trade_date"]
         prev_content = row["content"] or ""
-        return f"""以下是前一个交易日（{prev_date}）的分析报告，请参考其中的分析逻辑和关注方向，结合今日数据进行对比分析：
+        return f"""以下是前一个交易日（{prev_date}）的分析报告，仅用于保持分析连续性：
+
+- 今日数据优先：两者冲突时以今日工具数据为准，并在正文点出变化（如"昨日提示的××今日××"）
+- 禁止照搬昨日结论、禁止复制昨日段落
+- 重点核对昨日"明日展望"与"风险提示"里提到的方向是否兑现；未兑现的要在正文中说明
 
 {prev_content}
 

backend/services/ai_tools.py

@@ -16,7 +16,7 @@ TOOLS = [
         "type": "function",
         "function": {
             "name": "get_market_dashboard",
-            "description": "获取A股市场看板数据，包含主要指数行情、涨跌统计、行业强度、概念热度、事件情报、市场温度评分",
+            "description": "获取A股市场看板数据，包含主要指数行情、全市场涨跌统计（涨跌家数/涨停/跌停/炸板率/成交额）、市场温度评分与竞价信号、行业强度榜、概念热度、板块主力资金流、两融余额、海外主要指数与国内期货主力合约（globalMarkets字段）、完整连板梯队（limitLadder字段，含涨停原因与封单金额）、事件情报（热门股/龙虎榜/飙升/异动）",
             "parameters": {"type": "object", "properties": {}, "required": []}
         }
     },
@@ -42,6 +42,20 @@ TOOLS = [
             "parameters": {"type": "object", "properties": {}, "required": []}
         }
     },
+    {
+        "type": "function",
+        "function": {
+            "name": "get_consecutive_core_stocks",
+            "description": "获取最近N个交易日每天都上榜的核心股（缺一日都不行），用于识别持续活跃的热点股",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "days": {"type": "integer", "description": "交易日天数，默认3"}
+                },
+                "required": []
+            }
+        }
+    },
     {
         "type": "function",
         "function": {
@@ -71,6 +85,20 @@ TOOLS = [
                 "required": ["code", "name"]
             }
         }
+    },
+    {
+        "type": "function",
+        "function": {
+            "name": "get_news",
+            "description": "获取最近的财经快讯（新浪7x24，含宏观/行业/公司/海外动态），用于重要消息面梳理",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "limit": {"type": "integer", "description": "获取条数，默认30，最大50"}
+                },
+                "required": []
+            }
+        }
     }
 ]
 
@@ -84,6 +112,8 @@ async def execute_tool(tool_name: str, arguments: dict) -> str:
             return await _get_theme_history(arguments.get("date", ""))
         elif tool_name == "get_active_core_stocks":
             return await _get_active_core_stocks()
+        elif tool_name == "get_consecutive_core_stocks":
+            return await _get_consecutive_core_stocks(arguments.get("days", 3))
         elif tool_name == "get_stock_quote":
             return await _get_stock_quote(arguments.get("code", ""))
         elif tool_name == "get_fund_flow":
@@ -92,12 +122,23 @@ async def execute_tool(tool_name: str, arguments: dict) -> str:
                 arguments.get("name", ""),
                 arguments.get("days", 30)
             )
+        elif tool_name == "get_news":
+            return await _get_news(arguments.get("limit", 30))
         else:
             return json.dumps({"error": f"未知工具: {tool_name}"})
     except Exception as e:
         return json.dumps({"error": str(e)})
 
 
+async def _get_news(limit: int = 30) -> str:
+    """获取财经快讯"""
+    from services.market_extra import fetch_news
+    news = await fetch_news(limit)
+    if not news:
+        return json.dumps({"error": "快讯获取失败或暂无数据"}, ensure_ascii=False)
+    return json.dumps({"count": len(news), "items": news}, ensure_ascii=False)
+
+
 async def _get_market_dashboard() -> str:
     """获取市场看板数据"""
     from routes.market_dashboard import _build_dashboard
@@ -159,9 +200,9 @@ async def _get_active_core_stocks() -> str:
             if s["lastAppear"] is None or r["trade_date"] > s["lastAppear"]:
                 s["lastAppear"] = r["trade_date"]
 
-        stocks = list(stock_days.values())
-        stocks.sort(key=lambda x: x.get("lastAppear") or "", reverse=True)
-        stocks.sort(key=lambda x: -x["appearCount"])
+        # 只保留最近10日中出现次数最多的前25只（全量可达145KB，会把上下文撑爆）
+        stocks = sorted(stock_days.values(), key=lambda x: -x["appearCount"])[:25]
+        keep_codes = {s["stockCode"] for s in stocks}
 
         themes_rows = conn.execute(
             f"""SELECT stock_code, theme_code, theme_name FROM daily_core_stock_themes
@@ -170,10 +211,12 @@ async def _get_active_core_stocks() -> str:
         ).fetchall()
         themes_by_stock = {}
         for t in themes_rows:
+            if t["stock_code"] not in keep_codes:
+                continue
             per = themes_by_stock.setdefault(t["stock_code"], {})
             per.setdefault(t["theme_code"], {"theme_code": t["theme_code"], "theme_name": t["theme_name"]})
         for s in stocks:
-            s["themes"] = list(themes_by_stock.get(s["stockCode"], {}).values())
+            s["themes"] = list(themes_by_stock.get(s["stockCode"], {}).values())[:5]
 
         latest = dates[-1] if dates else None
         for s in stocks:
@@ -192,6 +235,66 @@ async def _get_active_core_stocks() -> str:
         conn.close()
 
 
+async def _get_consecutive_core_stocks(days: int = 3) -> str:
+    """获取最近N个交易日**每个交易日都上榜**的核心股，缺一日都不行"""
+    from database import get_connection
+    conn = get_connection()
+    try:
+        # 最近N个有数据的交易日
+        rows = conn.execute(
+            "SELECT DISTINCT trade_date FROM daily_core_stocks ORDER BY trade_date DESC LIMIT ?",
+            (days,)
+        ).fetchall()
+        recent_dates = [r["trade_date"] for r in reversed(rows)]
+
+        if len(recent_dates) < days:
+            return json.dumps({"days": days, "stocks": []}, ensure_ascii=False)
+
+        # 查询这些日期的上榜记录
+        placeholders = ",".join("?" * len(recent_dates))
+        rows = conn.execute(
+            f"""SELECT trade_date, stock_code, stock_name, f3, cover_count
+                FROM daily_core_stocks
+                WHERE trade_date IN ({placeholders})""",
+            recent_dates,
+        ).fetchall()
+
+        # 按股票分组，记录上榜日期
+        stock_dates: dict[str, dict] = {}
+        for r in rows:
+            code = r["stock_code"]
+            s = stock_dates.setdefault(code, {
+                "stockCode": code,
+                "stockName": r["stock_name"],
+                "dates": set(),
+                "dailyGains": {},
+            })
+            s["dates"].add(r["trade_date"])
+            s["dailyGains"][r["trade_date"]] = r["f3"]
+
+        # 只保留N个交易日全部上榜的股票
+        results = []
+        for code, s in stock_dates.items():
+            if len(s["dates"]) >= days:
+                results.append({
+                    "stockCode": code,
+                    "stockName": s["stockName"],
+                    "consecutiveDays": len(s["dates"]),
+                    "recentDates": sorted(s["dates"]),
+                    "dailyGains": s["dailyGains"],
+                })
+
+        results.sort(key=lambda x: -x["consecutiveDays"])
+
+        return json.dumps({
+            "days": days,
+            "recentDates": recent_dates,
+            "stocks": results,
+        }, ensure_ascii=False)
+    finally:
+        conn.close()
+
+
 async def _get_stock_quote(code: str) -> str:
     """获取个股实时行情"""
     from services.tencent import fetch_quote

backend/services/market_extra.py

@@ -0,0 +1,210 @@
+"""市场补充数据源（供 AI 分析与看板扩展）
+
+- fetch_news: 新浪财经 7x24 快讯
+- fetch_sector_fund_flow: 东方财富板块主力资金流排行（行业/概念）
+- fetch_margin_summary: 东方财富两融余额汇总（T+1 数据）
+
+均为公开接口，失败时返回 []/None，不阻塞主流程。
+"""
+
+import asyncio
+
+import httpx
+
+_TIMEOUT = 10.0
+_UA = ("Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 "
+       "(KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36")
+
+
+async def fetch_news(limit: int = 30) -> list[dict]:
+    """新浪财经 7x24 快讯，返回 [{time: 'MM-DD HH:MM', content}]；失败返回 []"""
+    try:
+        limit = max(1, min(int(limit or 30), 50))
+    except (TypeError, ValueError):
+        limit = 30
+    try:
+        async with httpx.AsyncClient(timeout=_TIMEOUT, headers={"User-Agent": _UA}) as client:
+            resp = await client.get(
+                "https://zhibo.sina.com.cn/api/zhibo/feed",
+                params={"page": 1, "page_size": limit, "zhibo_id": 152, "tag_id": 0},
+            )
+            resp.raise_for_status()
+            items = resp.json()["result"]["data"]["feed"]["list"]
+        news = []
+        for it in items or []:
+            text = (it.get("rich_text") or "").strip()
+            if not text:
+                continue
+            news.append({
+                "time": (it.get("create_time") or "")[5:16],  # 'MM-DD HH:MM'
+                "content": text[:300],
+            })
+        return news
+    except Exception:
+        return []
+
+
+_FLOW_HOSTS = (
+    # push2 对部分客户端有 TLS 指纹拦截（peer closed），delay 镜像同接口且稳定
+    "https://push2delay.eastmoney.com",
+    "https://push2.eastmoney.com",
+)
+
+
+async def _fetch_flow_boards(client: httpx.AsyncClient, fs: str, po: int, pz: int) -> list[dict]:
+    """拉取一类板块的主力净流入排行。po=1 降序（净流入最多），po=0 升序（净流出最多）
+
+    查询串保持字面量 + 号（与东财网页请求一致），逐 host 尝试。
+    """
+    qs = (f"/api/qt/clist/get?fid=f62&po={po}&pz={pz}&pn=1&np=1"
+          f"&fltt=2&invt=2&fs={fs}&fields=f12,f14,f62,f184")
+    last_err: Exception | None = None
+    for host in _FLOW_HOSTS:
+        try:
+            resp = await client.get(host + qs)
+            resp.raise_for_status()
+            diff = (resp.json().get("data") or {}).get("diff") or []
+            break
+        except Exception as e:
+            last_err = e
+            diff = []
+    else:
+        raise ConnectionError(f"板块资金流全部数据源失败: {last_err}")
+    if isinstance(diff, dict):  # 兼容旧版 {index: item} 结构
+        diff = list(diff.values())
+    boards = []
+    for d in diff:
+        amt = d.get("f62")
+        if not isinstance(amt, (int, float)):
+            continue
+        boards.append({
+            "code": d.get("f12", ""),
+            "name": d.get("f14", ""),
+            "mainNet": round(amt / 1e8, 1),  # 亿元
+            "mainNetPct": d.get("f184"),     # 主力净占比 %
+        })
+    return boards
+
+
+_GLOBAL_INDICES = "100.NDX,100.DJIA,100.SPX,100.HSI,100.N225,100.FTSE"
+_FUT_MARKETS = ("m:8", "m:113", "m:142", "m:114", "m:115")  # 中金所/上期所/上期能源/大商所/郑商所
+
+
+async def fetch_global_markets() -> dict | None:
+    """海外主要指数 + 国内期货主力合约；失败返回 None"""
+    try:
+        async with httpx.AsyncClient(timeout=_TIMEOUT, headers={"User-Agent": _UA}) as client:
+            overseas, futures = await asyncio.gather(
+                _fetch_overseas_indices(client),
+                _fetch_futures_main(client),
+            )
+        if not overseas and not futures:
+            return None
+        return {"overseas": overseas, "futures": futures}
+    except Exception:
+        return None
+
+
+async def _fetch_overseas_indices(client: httpx.AsyncClient) -> list[dict]:
+    resp = await client.get(
+        "https://push2delay.eastmoney.com/api/qt/ulist.np/get",
+        params={"secids": _GLOBAL_INDICES, "fields": "f12,f14,f2,f3", "fltt": 2, "invt": 2},
+    )
+    resp.raise_for_status()
+    diff = (resp.json().get("data") or {}).get("diff") or []
+    return [
+        {"name": d.get("f14", ""), "price": d.get("f2"), "changePct": d.get("f3")}
+        for d in diff
+    ]
+
+
+async def _fetch_futures_main(client: httpx.AsyncClient) -> list[dict]:
+    """国内期货主力合约（名称含"主连/主力合约"），按成交额降序取前12"""
+    import re
+
+    results: list[dict] = []
+    for fs in _FUT_MARKETS:
+        try:
+            resp = await client.get(
+                "https://push2delay.eastmoney.com/api/qt/clist/get",
+                params={"fid": "f6", "po": 1, "pz": 200, "pn": 1, "np": 1,
+                        "fltt": 2, "invt": 2, "fs": fs, "fields": "f12,f14,f2,f3,f6"},
+            )
+            resp.raise_for_status()
+            diff = (resp.json().get("data") or {}).get("diff") or []
+        except Exception:
+            continue
+        for d in diff:
+            name = d.get("f14") or ""
+            if "主连" not in name and "主力合约" not in name:
+                continue
+            if "次主连" in name:  # 次主力合约，排除
+                continue
+            amount = d.get("f6")
+            if not isinstance(amount, (int, float)):
+                continue
+            results.append({
+                "name": re.sub(r"(主连|主力合约)$", "", name),
+                "price": d.get("f2"),
+                "changePct": d.get("f3"),
+                "amountYi": round(amount / 1e8, 1),
+            })
+    results.sort(key=lambda x: -x["amountYi"])
+    return results[:12]
+
+
+async def fetch_sector_fund_flow() -> dict | None:
+    """板块主力资金流排行：行业净流入/净流出 TOP6 + 概念净流入 TOP6；失败返回 None"""
+    try:
+        async with httpx.AsyncClient(timeout=_TIMEOUT, headers={"User-Agent": _UA}) as client:
+            industry_in, industry_out, concept_in = await asyncio.gather(
+                _fetch_flow_boards(client, "m:90+t:2", 1, 6),
+                _fetch_flow_boards(client, "m:90+t:2", 0, 6),
+                _fetch_flow_boards(client, "m:90+t:3", 1, 6),
+            )
+        return {
+            "industryInflow": industry_in,    # 主力净流入降序
+            "industryOutflow": industry_out,  # 升序（净流出最多在前）
+            "conceptInflow": concept_in,
+            "unit": "亿元",
+        }
+    except Exception:
+        return None
+
+
+async def fetch_margin_summary() -> dict | None:
+    """沪深北两融余额汇总（交易所 T+1 披露）；失败返回 None"""
+    try:
+        async with httpx.AsyncClient(timeout=_TIMEOUT, headers={"User-Agent": _UA}) as client:
+            resp = await client.get(
+                "https://datacenter-web.eastmoney.com/api/data/v1/get",
+                params={
+                    "reportName": "RPTA_RZRQ_LSHJ",
+                    "columns": "ALL",
+                    "source": "WEB",
+                    "sortColumns": "dim_date",
+                    "sortTypes": "-1",
+                    "pageSize": 2,
+                    "pageNumber": 1,
+                },
+            )
+            resp.raise_for_status()
+            rows = ((resp.json().get("result") or {}).get("data")) or []
+        if not rows:
+            return None
+
+        def _balance(row: dict) -> float:
+            return float(row.get("RZYE") or 0) + float(row.get("RQYE") or 0)
+
+        latest = rows[0]
+        prev = rows[1] if len(rows) > 1 else None
+        balance = _balance(latest)
+        change = (balance - _balance(prev)) if prev else None
+        return {
+            "date": (latest.get("DIM_DATE") or "")[:10],
+            "balanceYi": round(balance / 1e8),  # 亿元
+            "changeYi": round(change / 1e8) if change is not None else None,
+            "rzjmeYi": round(float(latest.get("RZJME") or 0) / 1e8, 1),  # 融资净买入
+        }
+    except Exception:
+        return None

src/components/kline-card.tsx

@@ -137,6 +137,22 @@ export function KLineCard({ code, addedAt, ready }: Props) {
   const displayData = chartPeriod === "d" ? dailyData : minuteData;
   const currentLoading = chartPeriod === "d" ? chartLoading : minuteLoading;
 
+  // 传给图表的数据引用需稳定（仅随 displayData 变化），
+  // 否则切换指标/显示方式时 KLineChart 会把新数组当成新数据集重置视口
+  const chartData = useMemo(
+    () =>
+      displayData.map((d) => ({
+        time: d.dateMs,
+        open: d.open,
+        close: d.close,
+        high: d.high,
+        low: d.low,
+        volume: d.volume,
+        isAddedDate: d.isAddedDate,
+      })),
+    [displayData],
+  );
+
   return (
     <Card className="shadow-lg">
       <CardHeader className="pb-2 md:pb-4 px-3 md:px-6 pt-4 md:pt-6">
@@ -213,7 +229,7 @@ export function KLineCard({ code, addedAt, ready }: Props) {
       </CardHeader>
       <CardContent className="px-2 md:px-6 pb-2 md:pb-6">
         {currentLoading ? (
-          <div className="h-[320px] sm:h-[380px] md:h-[440px] w-full flex items-center justify-center">
+          <div className="h-[390px] w-full flex items-center justify-center">
             <div className="flex items-center text-muted-foreground text-sm">
               <div className="animate-pulse mr-2 h-2 w-2 rounded-full bg-primary"></div>
               K线数据加载中...
@@ -221,18 +237,11 @@ export function KLineCard({ code, addedAt, ready }: Props) {
           </div>
         ) : (
           <KLineChart
-            data={displayData.map((d) => ({
-              time: d.dateMs,
-              open: d.open,
-              close: d.close,
-              high: d.high,
-              low: d.low,
-              volume: d.volume,
-              isAddedDate: d.isAddedDate,
-            }))}
+            data={chartData}
             mode={chartMode}
             hasAddedDate={ready}
             indicators={indicators}
+            defaultVisibleDays={chartPeriod === "d" ? 90 : undefined}
           />
         )}
       </CardContent>

src/components/kline-chart.tsx

@@ -12,6 +12,7 @@ import {
   ColorType,
   CrosshairMode,
   LineStyle,
+  TickMarkType,
   createSeriesMarkers,
   type IChartApi,
   type ISeriesApi,
@@ -41,6 +42,8 @@ interface Props {
   mode: "line" | "candle";
   hasAddedDate?: boolean;
   indicators: IndicatorToggles;
+  /** 默认只展示最近 N 个自然日（日线用），其余数据靠拖动查看；不传则铺满全部数据 */
+  defaultVisibleDays?: number;
 }
 
 // lightweight-charts 无法解析 CSS 变量或 oklch() 颜色，直接用具体 hex 色。
@@ -61,9 +64,41 @@ const MACD_DIF = "#3b82f6";
 const MACD_DEA = "#f59e0b";
 const RSI_COLOR = "#a855f7";
 
-export function KLineChart({ data, mode, hasAddedDate, indicators }: Props) {
+// 时间统一按 UTC 取值格式化（kline-card 把北京时间墙钟视作 UTC 存储），
+// 用本地时区方法会导致日期错位一天
+function timeToDate(t: Time): Date {
+  if (typeof t === "number") return new Date(t * 1000);
+  if (typeof t === "string") return new Date(t.length === 10 ? `${t}T00:00:00Z` : t);
+  return new Date(Date.UTC(t.year, t.month - 1, t.day));
+}
+
+const pad2 = (n: number) => String(n).padStart(2, "0");
+
+/** 十字光标处的完整日期：2026-09-01 */
+function formatCrosshairTime(t: Time): string {
+  const d = timeToDate(t);
+  return `${d.getUTCFullYear()}-${pad2(d.getUTCMonth() + 1)}-${pad2(d.getUTCDate())}`;
+}
+
+/** 底部时间轴刻度：年→2026、月→2026-09、日→09-01（替代默认的 "01 9月 '26"） */
+function formatTickMark(t: Time, tickMarkType: TickMarkType): string {
+  const d = timeToDate(t);
+  if (tickMarkType === TickMarkType.Year) return `${d.getUTCFullYear()}`;
+  if (tickMarkType === TickMarkType.Month) return `${d.getUTCFullYear()}-${pad2(d.getUTCMonth() + 1)}`;
+  return `${pad2(d.getUTCMonth() + 1)}-${pad2(d.getUTCDate())}`;
+}
+
+// 各 pane 固定高度：主图加高、成交量压低，指标副图居中
+const MAIN_PANE_H = 320;
+const VOL_PANE_H = 70;
+const IND_PANE_H = 95;
+
+export function KLineChart({ data, mode, hasAddedDate, indicators, defaultVisibleDays }: Props) {
   const containerRef = useRef<HTMLDivElement>(null);
   const chartRef = useRef<IChartApi | null>(null);
+  // 记录上次填充的数据集：仅数据集变化（首次加载/切换周期）时调整视口，
+  // 指标开关/显示方式切换时保留用户拖动缩放后的位置
+  const lastDataRef = useRef<KLineItem[] | null>(null);
   const priceSeriesRef = useRef<ISeriesApi<"Candlestick" | "Line"> | null>(null);
   const volSeriesRef = useRef<ISeriesApi<"Histogram"> | null>(null);
   // 指标 series（重建用）
@@ -73,8 +108,15 @@ export function KLineChart({ data, mode, hasAddedDate, indicators }: Props) {
 
   // 容器高度随指标 pane 数量增长，避免主图被压缩
   const extraPanes = [indicators.macd, indicators.rsi].filter(Boolean).length;
-  // 固定图表高度：主图+成交量 300，每个指标副图 +95
-  const chartHeight = 300 + extraPanes * 95;
+  // 图表总高 = 各 pane 高度之和（主图 + 成交量 + 指标副图×N）
+  const chartHeight = MAIN_PANE_H + VOL_PANE_H + extraPanes * IND_PANE_H;
+
+  // pane 顺序固定：0=主图 1=成交量 2+=指标副图；指标开关增删 pane 后重新应用高度
+  function applyPaneHeights(chart: IChartApi) {
+    chart.panes().forEach((pane, i) => {
+      pane.setHeight(i === 0 ? MAIN_PANE_H : i === 1 ? VOL_PANE_H : IND_PANE_H);
+    });
+  }
 
   // 创建图表（仅一次）：pane0 主图 + pane1 成交量
   useEffect(() => {
@@ -94,7 +136,12 @@ export function KLineChart({ data, mode, hasAddedDate, indicators }: Props) {
         horzLines: { color: GRID, style: LineStyle.Dashed, visible: true },
       },
       rightPriceScale: { borderColor: GRID },
-      timeScale: { borderColor: GRID, timeVisible: false },
+      localization: { timeFormatter: formatCrosshairTime },
+      timeScale: {
+        borderColor: GRID,
+        timeVisible: false,
+        tickMarkFormatter: formatTickMark,
+      },
       crosshair: { mode: CrosshairMode.Normal },
     });
     chartRef.current = chart;
@@ -109,6 +156,7 @@ export function KLineChart({ data, mode, hasAddedDate, indicators }: Props) {
       1,
     );
     volSeriesRef.current = vol;
+    applyPaneHeights(chart);
 
     return () => {
       chart.remove();
@@ -118,6 +166,8 @@ export function KLineChart({ data, mode, hasAddedDate, indicators }: Props) {
       maSeriesRef.current = [];
       macdSeriesRef.current = [];
       rsiSeriesRef.current = [];
+      // 图表实例销毁后重置，重建时重新应用默认视口（StrictMode 重挂载同样生效）
+      lastDataRef.current = null;
     };
   }, []);
 
@@ -223,6 +273,8 @@ export function KLineChart({ data, mode, hasAddedDate, indicators }: Props) {
       );
       rsiSeriesRef.current.push(s);
     }
+
+    applyPaneHeights(chart);
   }
 
   // 模式切换：重建价格 series + 指标
@@ -310,7 +362,23 @@ export function KLineChart({ data, mode, hasAddedDate, indicators }: Props) {
       createSeriesMarkers(ps, []);
     }
 
-    chart.timeScale().fitContent();
+    if (lastDataRef.current === data) return;
+    lastDataRef.current = data;
+    if (defaultVisibleDays) {
+      // 默认视口只展示最近 N 个自然日，更早的数据靠向左拖动查看
+      const toMs = (t: string | number) =>
+        typeof t === "number" ? t * 1000 : new Date(t).getTime();
+      const lastMs = toMs(data[data.length - 1].time);
+      let fromIdx = data.findIndex(
+        (d) => toMs(d.time) >= lastMs - defaultVisibleDays * 86400_000,
+      );
+      if (fromIdx < 0) fromIdx = 0;
+      chart
+        .timeScale()
+        .setVisibleLogicalRange({ from: fromIdx, to: data.length - 1 + 2 });
+    } else {
+      chart.timeScale().fitContent();
+    }
   }
 
   // 颜色图例：根据启用的指标生成

src/lib/ai-analysis-api.ts

@@ -14,7 +14,11 @@ export interface AiReport {
   toolsUsed: string[];
   model: string;
   tokens_used: number;
+  generation_count?: number;
+  llm_calls?: number | null;
+  issue_number?: number;
   created_at: string;
+  updated_at?: string | null;
 }
 
 export async function fetchAiLatestReport(): Promise<AiReport> {
@@ -31,6 +35,13 @@ export async function fetchAiReports(): Promise<AiReport[]> {
   return result.data || [];
 }
 
+export async function fetchAiReportByDate(tradeDate: string): Promise<AiReport> {
+  const resp = await fetch(`${API_BASE}/api/ai-analysis/by-date/${tradeDate}`);
+  if (!resp.ok) throw new Error(`请求失败 (${resp.status})`);
+  const result = await resp.json();
+  return result.data;
+}
+
 export async function fetchAiReport(id: number): Promise<AiReport> {
   const resp = await fetch(`${API_BASE}/api/ai-analysis/${id}`);
   if (!resp.ok) throw new Error(`请求失败 (${resp.status})`);

src/lib/core-stock-api.ts

@@ -33,11 +33,11 @@ export interface CoreStockHistoryResponse {
   items: CoreStockHistoryItem[];
 }
 
-/** 获取活跃核心股 + 最近10日涨幅矩阵 */
-export async function fetchActiveCoreStocks(): Promise<ActiveCoreStocksResponse> {
+/** 获取活跃核心股 + 最近N日涨幅矩阵 */
+export async function fetchActiveCoreStocks(days: number = 10): Promise<ActiveCoreStocksResponse> {
   const baseUrl = getApiBaseUrl();
   try {
-    const resp = await fetch(`${baseUrl}/api/core-stocks/active`, { method: "GET", cache: "no-store" });
+    const resp = await fetch(`${baseUrl}/api/core-stocks/active?days=${days}`, { method: "GET", cache: "no-store" });
     if (!resp.ok) return { dates: [], stocks: [] };
     return resp.json();
   } catch (err) {

src/routes/ai-analysis.tsx

@@ -1,36 +1,153 @@
 import * as React from "react";
-import { Link, createFileRoute } from "@tanstack/react-router";
+import { Link, createFileRoute, useNavigate } from "@tanstack/react-router";
 import { ArrowLeft, Clock, Loader2, AlertCircle, Wrench, Calendar } from "lucide-react";
 import { useQuery } from "@tanstack/react-query";
 import Markdown from "react-markdown";
 import remarkGfm from "remark-gfm";
+import type { Components } from "react-markdown";
 import { Mermaid } from "../components/Mermaid";
-import { fetchAiLatestReport } from "../lib/ai-analysis-api";
+import { fetchAiLatestReport, fetchAiReportByDate, fetchAiReports } from "../lib/ai-analysis-api";
 import { Card, CardContent } from "../components/ui/card";
 
 export const Route = createFileRoute("/ai-analysis")({
+  validateSearch: (search: Record<string, unknown>) => {
+    return {
+      date: typeof search.date === "string" ? search.date : undefined,
+      // 兼容 ?data=YYYY-MM-DD 写法
+      data: typeof search.data === "string" ? search.data : undefined,
+    };
+  },
   component: AiAnalysisPage,
 });
 
+/** A股惯例红涨绿跌：只给带显式 +/- 号的数字着色（+2.3% / -1.2亿 / +56万...），无符号数字语义不明保持默认 */
+const SIGNED_NUM_RE = /([+-]\d+(?:\.\d+)?(?:%|亿|万亿|万)?)/g;
+
+function colorizeText(text: string): React.ReactNode[] {
+  return text.split(SIGNED_NUM_RE).map((part, i) => {
+    if (part.startsWith("+") || part.startsWith("-")) {
+      const val = parseFloat(part);
+      if (!Number.isNaN(val) && val !== 0) {
+        if (val > 0) return <span key={i} className="text-red-500">{part}</span>;
+        return <span key={i} className="text-green-600">{part}</span>;
+      }
+    }
+    return <React.Fragment key={i}>{part}</React.Fragment>;
+  });
+}
+
+function colorizeChildren(children: React.ReactNode): React.ReactNode {
+  return React.Children.map(children, (child) => {
+    if (typeof child === "string") return colorizeText(child);
+    if (React.isValidElement(child)) {
+      const kids = (child.props as { children?: React.ReactNode }).children;
+      if (kids != null) {
+        return React.cloneElement(
+          child as React.ReactElement<{ children?: React.ReactNode }>,
+          {},
+          colorizeChildren(kids),
+        );
+      }
+    }
+    return child;
+  });
+}
+
+/** 按 "## " 二级标题拆分报告为多张卡片；首段（# 标题等）若无正文则不单独成卡 */
+function splitReport(content: string): { intro: string; sections: { title: string; body: string }[] } {
+  const parts = content.split(/\n(?=## )/);
+  const sections = parts.slice(1).map((p) => {
+    const nl = p.indexOf("\n");
+    const title = (nl === -1 ? p : p.slice(0, nl)).replace(/^##\s*/, "").trim();
+    const body = nl === -1 ? "" : p.slice(nl + 1).trim();
+    return { title, body };
+  });
+  const intro = parts[0].trim();
+  // intro 只剩标题（无其他内容）时置空，避免渲染一张只有大标题的空卡片
+  const introBody = intro.replace(/^#[^\n]*$/m, "").trim();
+  return { intro: introBody ? intro : "", sections };
+}
+
 function AiAnalysisPage() {
+  const navigate = useNavigate();
+  const { date, data } = Route.useSearch();
+  const tradeDate = date ?? data ?? undefined;
+  const isHistorical = !!tradeDate;
+
   const { data: report, isLoading, isError } = useQuery({
-    queryKey: ["ai-report-latest"],
-    queryFn: fetchAiLatestReport,
+    queryKey: isHistorical ? ["ai-report", tradeDate] : ["ai-report-latest"],
+    queryFn: () => (isHistorical ? fetchAiReportByDate(tradeDate!) : fetchAiLatestReport()),
     retry: false,
   });
 
+  const { data: history = [] } = useQuery({
+    queryKey: ["ai-reports"],
+    queryFn: fetchAiReports,
+    retry: false,
+  });
+
+  const goDate = (d?: string) => {
+    navigate({ to: "/ai-analysis", search: d ? { date: d } : {} });
+  };
+
+  const selectedDate = tradeDate ?? report?.trade_date ?? "";
+
+  const mdComponents: Components = {
+    table({ children, ...props }) {
+      return (
+        <div className="overflow-x-auto -mx-3 sm:mx-0 px-3 sm:px-0">
+          <table {...props}>{children}</table>
+        </div>
+      );
+    },
+    td({ children, ...props }) {
+      return <td {...props}>{colorizeChildren(children)}</td>;
+    },
+    code({ className, children, ...props }) {
+      const match = /language-(\w+)/.exec(className || "");
+      if (match && match[1] === "mermaid") {
+        return <Mermaid chart={String(children).replace(/\n$/, "")} />;
+      }
+      return (
+        <code className={className} {...props}>
+          {children}
+        </code>
+      );
+    },
+  };
+
   return (
     <div className="min-h-screen bg-background text-foreground">
       <div className="max-w-4xl mx-auto px-3 sm:px-4 py-4 sm:py-6">
         {/* Header */}
-        <div className="flex items-center gap-2 sm:gap-3 mb-4 sm:mb-6">
-          <Link to="/" className="text-muted-foreground hover:text-foreground transition-colors">
-            <ArrowLeft className="h-5 w-5" />
-          </Link>
-          <h1 className="text-lg sm:text-xl md:text-2xl font-bold flex items-center gap-2">
-            <Calendar className="h-5 w-5 text-primary" />
-            AI 收盘分析
-          </h1>
+        <div className="flex items-center justify-between gap-2 sm:gap-3 mb-4 sm:mb-6">
+          <div className="flex items-center gap-2 sm:gap-3 min-w-0">
+            <Link to="/" className="text-muted-foreground hover:text-foreground transition-colors shrink-0">
+              <ArrowLeft className="h-5 w-5" />
+            </Link>
+            <h1 className="text-lg sm:text-xl md:text-2xl font-bold flex items-center gap-2 truncate">
+              <Calendar className="h-5 w-5 text-primary shrink-0" />
+              AI 收盘分析
+              {report?.issue_number ? (
+                <span className="text-xs font-normal text-muted-foreground whitespace-nowrap">总第 {report.issue_number} 期</span>
+              ) : null}
+            </h1>
+          </div>
+          {history.length > 0 && (
+            <select
+              value={history.some((h) => h.trade_date === selectedDate) ? selectedDate : ""}
+              onChange={(e) => goDate(e.target.value || undefined)}
+              className="text-xs sm:text-sm border border-border rounded-md bg-background px-2 py-1.5 text-foreground shrink-0"
+              aria-label="选择历史报告日期"
+            >
+              <option value="">最新</option>
+              {history.map((h) => (
+                <option key={h.id} value={h.trade_date}>
+                  {h.trade_date}
+                </option>
+              ))}
+            </select>
+          )}
         </div>
 
         {/* Content */}
@@ -42,8 +159,20 @@ function AiAnalysisPage() {
         ) : isError || !report ? (
           <div className="flex flex-col items-center gap-3 py-20">
             <AlertCircle className="h-6 w-6 text-destructive" />
-            <p className="text-sm text-muted-foreground">暂无分析报告</p>
-            <p className="text-xs text-muted-foreground/60">请通过管理面板触发 AI 分析</p>
+            <p className="text-sm text-muted-foreground">
+              {isHistorical ? `${tradeDate} 暂无分析报告` : "暂无分析报告"}
+            </p>
+            {isHistorical ? (
+              <Link
+                to="/ai-analysis"
+                search={{}}
+                className="text-xs text-primary hover:underline"
+              >
+                查看最新报告
+              </Link>
+            ) : (
+              <p className="text-xs text-muted-foreground/60">请通过管理面板触发 AI 分析</p>
+            )}
           </div>
         ) : (
           <>
@@ -53,9 +182,13 @@ function AiAnalysisPage() {
                 <div className="flex flex-wrap items-center gap-x-3 gap-y-1 sm:gap-4 text-xs sm:text-sm text-muted-foreground">
                   <span className="flex items-center gap-1.5">
                     <Clock className="h-3.5 w-3.5 shrink-0" />
-                    <span className="truncate">{report.created_at}</span>
+                    <span className="truncate">{report.updated_at || report.created_at}</span>
                   </span>
                   <span>{report.tokens_used?.toLocaleString()} tokens</span>
+                  {report.llm_calls ? <span>{report.llm_calls} 次 LLM 调用</span> : null}
+                  {(report.generation_count ?? 1) > 1 && (
+                    <span>第 {report.generation_count} 次生成</span>
+                  )}
                   {report.toolsUsed && report.toolsUsed.length > 0 && (
                     <span className="flex items-center gap-1.5">
                       <Wrench className="h-3.5 w-3.5 shrink-0" />
@@ -66,44 +199,45 @@ function AiAnalysisPage() {
               </CardContent>
             </Card>
 
-            {/* Report Content */}
-            <Card className="bg-card border-border">
-              <CardContent className="p-3 sm:p-4 md:p-6 ai-report-content">
-                <Markdown
-                  remarkPlugins={[remarkGfm]}
-                  components={{
-                    table({ children, ...props }) {
-                      return (
-                        <div className="overflow-x-auto -mx-3 sm:mx-0 px-3 sm:px-0">
-                          <table {...props}>{children}</table>
-                        </div>
-                      );
-                    },
-                    code({ className, children, ...props }) {
-                      const match = /language-(\w+)/.exec(className || "");
-                      if (match && match[1] === "mermaid") {
-                        return <Mermaid chart={String(children).replace(/\n$/, "")} />;
-                      }
-                      return (
-                        <code className={className} {...props}>
-                          {children}
-                        </code>
-                      );
-                    },
-                  }}
-                >
-                  {report.content}
-                </Markdown>
-              </CardContent>
-            </Card>
-
-            {/* Disclaimer */}
-            <p className="text-xs text-muted-foreground/60 text-center mt-4 sm:mt-6">
-              本报告由 AI 生成，仅供参考，不构成投资建议
-            </p>
+            {/* 报告正文：按 ## 章节分卡片渲染 */}
+            {(() => {
+              const { intro, sections } = splitReport(report.content);
+              return (
+                <div className="space-y-3 sm:space-y-4">
+                  {intro && (
+                    <Card className="bg-card border-border">
+                      <CardContent className="p-3 sm:p-4 md:p-6 ai-report-content">
+                        <Markdown remarkPlugins={[remarkGfm]} components={mdComponents}>
+                          {intro}
+                        </Markdown>
+                      </CardContent>
+                    </Card>
+                  )}
+                  {sections.map((sec) => (
+                    <Card key={sec.title} className="bg-card border-border overflow-hidden">
+                      <div className="px-3 sm:px-4 md:px-6 pt-3 sm:pt-4 pb-2 sm:pb-3 border-b border-border/60">
+                        <h2 className="text-base sm:text-lg font-semibold">{sec.title}</h2>
+                      </div>
+                      <CardContent className="p-3 sm:p-4 md:p-6 ai-report-content">
+                        <Markdown remarkPlugins={[remarkGfm]} components={mdComponents}>
+                          {sec.body}
+                        </Markdown>
+                      </CardContent>
+                    </Card>
+                  ))}
+                </div>
+              );
+            })()}
           </>
         )}
+
+        {/* Disclaimer */}
+        <p className="text-xs text-muted-foreground/60 text-center mt-4 sm:mt-6">
+          本报告由 AI 生成，仅供参考，不构成投资建议
+        </p>
       </div>
     </div>
   );
 }
+
+export default AiAnalysisPage;

src/routes/core-stocks.tsx

@@ -1,8 +1,9 @@
 import { createFileRoute, Link } from "@tanstack/react-router";
 import { useQuery } from "@tanstack/react-query";
-import { Fragment } from "react";
+import { Fragment, useState } from "react";
 import { fetchActiveCoreStocks } from "@/lib/core-stock-api";
 import { ArrowLeft, RefreshCw, Flame } from "lucide-react";
+import { Tabs, TabsList, TabsTrigger } from "@/components/ui/tabs";
 
 export const Route = createFileRoute("/core-stocks")({
   component: CoreStocksPage,
@@ -16,9 +17,10 @@ function formatGain(v: number | null | undefined): string {
 }
 
 function CoreStocksPage() {
+  const [days, setDays] = useState(10);
   const { data, isLoading, isFetching, refetch } = useQuery({
-    queryKey: ["core-stocks", "active"],
-    queryFn: fetchActiveCoreStocks,
+    queryKey: ["core-stocks", "active", days],
+    queryFn: () => fetchActiveCoreStocks(days),
     staleTime: 60_000,
     retry: false,
   });
@@ -48,9 +50,18 @@ function CoreStocksPage() {
       </header>
 
       <div className="max-w-5xl mx-auto px-4 mt-3 pb-8">
-        <p className="text-[10px] text-muted-foreground mb-2">
-          活跃热点股（最近 10 个交易日内上榜）· 按上榜次数排序 · 共 {stocks.length} 只
-        </p>
+        <div className="flex items-center justify-between mb-3">
+          <p className="text-[10px] text-muted-foreground">
+            活跃热点股（最近 {days} 个交易日内上榜）· 按上榜次数排序 · 共 {stocks.length} 只
+          </p>
+          <Tabs value={String(days)} onValueChange={(v) => setDays(Number(v))}>
+            <TabsList>
+              <TabsTrigger value="3">近3日</TabsTrigger>
+              <TabsTrigger value="5">近5日</TabsTrigger>
+              <TabsTrigger value="10">近10日</TabsTrigger>
+            </TabsList>
+          </Tabs>
+        </div>
 
         {isLoading ? (
           <div className="animate-pulse rounded-xl bg-muted h-32" />