diff --git a/.gitignore b/.gitignore
index a0fec98..2fc3fbc 100644
--- a/.gitignore
+++ b/.gitignore
@@ -18,6 +18,7 @@ data/*.sqlite
 data/*.sqlite3
 data/ai_settings.json
 data/box_types.json
+data/ai_chat_memory.json
 data/*.log
 *.db
 *.sqlite
diff --git a/README.md b/README.md
index 95a331c..85a6749 100644
--- a/README.md
+++ b/README.md
@@ -134,6 +134,34 @@ $env:SILICONFLOW_MODEL="Qwen/Qwen2.5-7B-Instruct"
 
 当 AI 补货建议出现“请求失败，请稍后重试”时，优先打开系统日志查看最近的 `ERROR` 或 `WARNING` 记录。
 
+### 2.7 AI 数据库聊天
+
+系统提供数据库聊天页，支持自然语言提问并自动生成只读 SQL 查询库存数据：
+
+- 页面：`/ai/chat`
+- 入口：`仓库概览` 右侧 AI 卡片中的 `聊天`
+
+安全边界：
+
+- 只允许 `SELECT/CTE` 查询
+- 禁止写操作（INSERT/UPDATE/DELETE/DDL 等）
+- 禁止访问 `users` 表
+- 默认自动补 `LIMIT`，避免一次性返回过多数据
+
+若提问失败，可在系统日志中查看 `ai_chat_*` 相关记录。
+
+联网补充（可选）：
+
+- 聊天页可勾选 `允许联网补充`
+- 开启后会在数据库查询结果基础上补充公开来源线索
+- 回答会区分“数据库结论”和“联网参考”，并展示来源可信度与链接
+
+本地记忆：
+
+- AI 聊天会为当前登录用户保存本地记忆（最近对话与“记住 ...”条目）
+- 记忆文件：`data/ai_chat_memory.json`
+- 聊天页可点击 `清空本地记忆` 按钮重置
+
 ## 3. 页面说明
 
 ### 3.1 首页 `/`
diff --git a/app.py b/app.py
index 740c62e..8d9f2c8 100644
--- a/app.py
+++ b/app.py
@@ -39,6 +39,7 @@ DB_DIR = os.path.join(BASE_DIR, "data")
 os.makedirs(DB_DIR, exist_ok=True)
 DB_PATH = os.path.join(DB_DIR, "inventory.db")
 APP_LOG_PATH = os.path.join(DB_DIR, "app.log")
+AI_CHAT_MEMORY_PATH = os.path.join(DB_DIR, "ai_chat_memory.json")
 
 # Flask 和 SQLAlchemy 基础初始化。
 app = Flask(__name__)
@@ -100,6 +101,105 @@ def _read_log_lines(limit: int = 200) -> list[str]:
         return [line.rstrip("\n") for line in file_obj.readlines()[-limit:]]
 
 
+def _load_ai_chat_memory_store() -> dict:
+    if not os.path.exists(AI_CHAT_MEMORY_PATH):
+        return {}
+    try:
+        with open(AI_CHAT_MEMORY_PATH, "r", encoding="utf-8") as file_obj:
+            payload = json.load(file_obj)
+        return payload if isinstance(payload, dict) else {}
+    except (OSError, json.JSONDecodeError):
+        return {}
+
+
+def _save_ai_chat_memory_store(payload: dict) -> None:
+    safe_payload = payload if isinstance(payload, dict) else {}
+    with open(AI_CHAT_MEMORY_PATH, "w", encoding="utf-8") as file_obj:
+        json.dump(safe_payload, file_obj, ensure_ascii=False, indent=2)
+
+
+def _normalize_memory_text(text: str, max_len: int = 220) -> str:
+    cleaned = re.sub(r"\s+", " ", str(text or "")).strip()
+    if len(cleaned) > max_len:
+        return cleaned[: max_len - 3] + "..."
+    return cleaned
+
+
+def _get_ai_chat_memory(username: str) -> dict:
+    user_key = (username or "guest").strip() or "guest"
+    store = _load_ai_chat_memory_store()
+    block = store.get(user_key, {}) if isinstance(store, dict) else {}
+    notes = block.get("notes", []) if isinstance(block, dict) else []
+    turns = block.get("turns", []) if isinstance(block, dict) else []
+    if not isinstance(notes, list):
+        notes = []
+    if not isinstance(turns, list):
+        turns = []
+    return {
+        "user": user_key,
+        "notes": [str(item) for item in notes[:30]],
+        "turns": [item for item in turns[-20:] if isinstance(item, dict)],
+    }
+
+
+def _memory_context_text(memory: dict) -> str:
+    notes = memory.get("notes", []) if isinstance(memory, dict) else []
+    turns = memory.get("turns", []) if isinstance(memory, dict) else []
+    lines = []
+    if notes:
+        lines.append("用户长期记忆:")
+        for note in notes[-12:]:
+            lines.append(f"- {_normalize_memory_text(note, max_len=120)}")
+    if turns:
+        lines.append("最近本地记忆对话:")
+        for turn in turns[-6:]:
+            q = _normalize_memory_text(turn.get("q", ""), max_len=90)
+            a = _normalize_memory_text(turn.get("a", ""), max_len=120)
+            lines.append(f"Q: {q}")
+            lines.append(f"A: {a}")
+    return "\n".join(lines)
+
+
+def _append_ai_chat_memory(username: str, question: str, answer: str) -> None:
+    user_key = (username or "guest").strip() or "guest"
+    store = _load_ai_chat_memory_store()
+    block = store.get(user_key)
+    if not isinstance(block, dict):
+        block = {"notes": [], "turns": []}
+
+    notes = block.get("notes")
+    turns = block.get("turns")
+    if not isinstance(notes, list):
+        notes = []
+    if not isinstance(turns, list):
+        turns = []
+
+    q = _normalize_memory_text(question)
+    a = _normalize_memory_text(answer)
+    turns.append({"q": q, "a": a, "ts": datetime.utcnow().isoformat(timespec="seconds")})
+    turns = turns[-30:]
+
+    match = re.match(r"^(记住|记一下|请记住)[:：\s]*(.+)$", q)
+    if match:
+        note = _normalize_memory_text(match.group(2), max_len=140)
+        if note and note not in notes:
+            notes.append(note)
+            notes = notes[-30:]
+
+    block["notes"] = notes
+    block["turns"] = turns
+    store[user_key] = block
+    _save_ai_chat_memory_store(store)
+
+
+def _clear_ai_chat_memory(username: str) -> None:
+    user_key = (username or "guest").strip() or "guest"
+    store = _load_ai_chat_memory_store()
+    if user_key in store:
+        store.pop(user_key, None)
+        _save_ai_chat_memory_store(store)
+
+
 _setup_app_logger()
 
 # 这里集中放全局常量，避免后面函数里散落硬编码。
@@ -2875,6 +2975,675 @@ def _call_siliconflow_chat(
         raise RuntimeError("AI 返回格式无法解析") from exc
 
 
+DB_CHAT_ALLOWED_TABLES = {"boxes", "components", "inventory_events"}
+DB_CHAT_FORBIDDEN_KEYWORDS = {
+    "insert",
+    "update",
+    "delete",
+    "drop",
+    "alter",
+    "create",
+    "replace",
+    "truncate",
+    "attach",
+    "detach",
+    "pragma",
+    "vacuum",
+    "reindex",
+    "begin",
+    "commit",
+    "rollback",
+}
+
+
+def _extract_json_object_text(raw_text: str) -> str:
+    text = (raw_text or "").strip()
+    if not text:
+        return ""
+    if text.startswith("```"):
+        text = re.sub(r"^```(?:json)?\s*", "", text)
+        text = re.sub(r"\s*```$", "", text)
+    first = text.find("{")
+    last = text.rfind("}")
+    if first >= 0 and last > first:
+        return text[first : last + 1]
+    return text
+
+
+def _build_db_chat_schema_hint() -> str:
+    return (
+        "可用表和字段:\n"
+        "1) boxes(id, name, description, box_type, slot_capacity, slot_prefix, start_number)\n"
+        "2) components(id, box_id, slot_index, part_no, name, specification, quantity, location, note, is_enabled)\n"
+        "3) inventory_events(id, box_id, box_type, component_id, part_no, event_type, delta, created_at)\n"
+        "禁止访问 users 表。"
+    )
+
+
+def _is_safe_readonly_sql(raw_sql: str) -> tuple[bool, str, str]:
+    sql = (raw_sql or "").strip()
+    if not sql:
+        return False, "SQL 为空", ""
+
+    if sql.endswith(";"):
+        sql = sql[:-1].strip()
+
+    lowered = re.sub(r"\s+", " ", sql.lower())
+    if ";" in sql:
+        return False, "仅允许单条 SQL 查询", ""
+
+    if not (lowered.startswith("select ") or lowered.startswith("with ")):
+        return False, "仅允许 SELECT 只读查询", ""
+
+    for keyword in DB_CHAT_FORBIDDEN_KEYWORDS:
+        if re.search(rf"\b{re.escape(keyword)}\b", lowered):
+            return False, f"SQL 包含危险关键字: {keyword}", ""
+
+    if re.search(r"\busers\b", lowered):
+        return False, "禁止访问 users 表", ""
+
+    referenced_tables = re.findall(r"\b(?:from|join)\s+([a-zA-Z_][\w]*)", lowered)
+    if not referenced_tables:
+        return False, "SQL 未包含可识别的数据表", ""
+
+    for table_name in referenced_tables:
+        if table_name not in DB_CHAT_ALLOWED_TABLES:
+            return False, f"不允许访问表: {table_name}", ""
+
+    return True, "", sql
+
+
+def _ensure_query_limit(sql: str, row_limit: int = 80) -> str:
+    if re.search(r"\blimit\s+\d+", sql, flags=re.IGNORECASE):
+        return sql
+    return f"{sql} LIMIT {row_limit}"
+
+
+def _serialize_sql_rows(rows, columns: list[str], *, max_cell_len: int = 120) -> list[dict]:
+    serialized = []
+    for row in rows:
+        item = {}
+        for idx, column in enumerate(columns):
+            value = row[idx] if idx < len(row) else None
+            if isinstance(value, datetime):
+                text = value.strftime("%Y-%m-%d %H:%M:%S")
+            else:
+                text = "" if value is None else str(value)
+            if len(text) > max_cell_len:
+                text = text[: max_cell_len - 3] + "..."
+            item[column] = text
+        serialized.append(item)
+    return serialized
+
+
+def _build_db_chat_sql_plan(
+    question: str,
+    history: list[dict],
+    memory_text: str,
+    settings: dict,
+) -> tuple[str, str, str]:
+    history_lines = []
+    for row in history[-6:]:
+        role = str(row.get("role", "user") or "user")
+        content = str(row.get("content", "") or "").strip()
+        if not content:
+            continue
+        history_lines.append(f"[{role}] {content[:200]}")
+
+    system_prompt = (
+        "你是库存数据库查询规划助手。"
+        "请根据用户问题生成一条 SQLite 只读 SQL。"
+        "只允许 SELECT/CTE，不要写入操作，不要访问 users 表。"
+        "输出必须是 JSON: {\"sql\":string,\"reason\":string}，不要输出其他文字。"
+    )
+    user_prompt = (
+        _build_db_chat_schema_hint()
+        + "\n\n对话历史:\n"
+        + "\n".join(history_lines or ["(无)"])
+        + "\n\n当前问题:\n"
+        + question
+    )
+    if memory_text:
+        user_prompt += "\n\n本地记忆摘要:\n" + memory_text
+
+    raw_text = _call_siliconflow_chat(
+        system_prompt,
+        user_prompt,
+        api_url=settings["api_url"],
+        model=settings["model"],
+        api_key=settings["api_key"],
+        timeout=settings["timeout"],
+    )
+    parsed = json.loads(_extract_json_object_text(raw_text))
+    sql = str(parsed.get("sql", "") or "").strip()
+    reason = str(parsed.get("reason", "") or "").strip() or "AI 未提供 SQL 说明"
+    if not sql:
+        raise RuntimeError("AI 未生成有效 SQL")
+    return sql, reason, raw_text
+
+
+def _build_db_chat_web_context(question: str, rows: list[dict], timeout: int, max_queries: int = 2) -> list[dict]:
+    """为数据库聊天构建可选联网参考上下文。"""
+    query_candidates = []
+    base_query = (question or "").strip()
+    if base_query:
+        query_candidates.append(base_query)
+
+    for row in rows[:3]:
+        part_no = str(row.get("part_no", "") or "").strip()
+        name = str(row.get("name", "") or "").strip()
+        if part_no and name:
+            query_candidates.append(f"{part_no} {name}")
+        elif part_no:
+            query_candidates.append(part_no)
+        elif name:
+            query_candidates.append(name)
+
+    contexts = []
+    used = set()
+    for query in query_candidates:
+        q = query.strip()
+        if len(q) < 2 or q in used:
+            continue
+        used.add(q)
+        result = _fetch_open_search_context(q, timeout=timeout)
+        sources = result.get("sources") or []
+        if not sources:
+            continue
+        contexts.append(
+            {
+                "query": result.get("query", q),
+                "sources": sources[:4],
+            }
+        )
+        if len(contexts) >= max_queries:
+            break
+
+    return contexts
+
+
+def _extract_weather_location(question: str) -> str:
+    text = (question or "").strip()
+    if not text:
+        return ""
+
+    patterns = [
+        r"查询(.{1,20}?)的?天气",
+        r"(.{1,20}?)的?天气",
+    ]
+    for pattern in patterns:
+        match = re.search(pattern, text)
+        if not match:
+            continue
+        candidate = re.sub(r"[\s，。！？?、]+", "", match.group(1) or "").strip()
+        if candidate:
+            return candidate
+    return ""
+
+
+def _weather_code_to_text(code: int) -> str:
+    mapping = {
+        0: "晴",
+        1: "大部晴",
+        2: "多云",
+        3: "阴",
+        45: "雾",
+        48: "雾凇",
+        51: "小毛毛雨",
+        53: "毛毛雨",
+        55: "浓毛毛雨",
+        61: "小雨",
+        63: "中雨",
+        65: "大雨",
+        71: "小雪",
+        73: "中雪",
+        75: "大雪",
+        80: "阵雨",
+        81: "较强阵雨",
+        82: "强阵雨",
+        95: "雷阵雨",
+    }
+    return mapping.get(int(code), f"天气代码 {code}")
+
+
+def _fetch_weather_context(question: str, timeout: int) -> dict | None:
+    """天气问题联网兜底：使用 Open-Meteo 地理编码与实时天气接口。"""
+    location = _extract_weather_location(question)
+    if not location:
+        location = "广州"
+
+    try:
+        geo_params = urllib.parse.urlencode(
+            {
+                "name": location,
+                "count": "1",
+                "language": "zh",
+                "format": "json",
+            }
+        )
+        geo_url = f"https://geocoding-api.open-meteo.com/v1/search?{geo_params}"
+        geo_req = urllib.request.Request(
+            geo_url,
+            method="GET",
+            headers={"User-Agent": "inventory-ai-chat/1.0"},
+        )
+        with urllib.request.urlopen(geo_req, timeout=timeout) as resp:
+            geo_raw = resp.read().decode("utf-8", errors="ignore")
+        geo_data = json.loads(geo_raw)
+        results = geo_data.get("results") or []
+        if not results:
+            return None
+
+        first = results[0]
+        latitude = first.get("latitude")
+        longitude = first.get("longitude")
+        if latitude is None or longitude is None:
+            return None
+
+        display_name = str(first.get("name") or location).strip() or location
+        country = str(first.get("country") or "").strip()
+        admin1 = str(first.get("admin1") or "").strip()
+        place_name = " / ".join([v for v in [display_name, admin1, country] if v])
+
+        weather_params = urllib.parse.urlencode(
+            {
+                "latitude": str(latitude),
+                "longitude": str(longitude),
+                "current": "temperature_2m,weather_code,wind_speed_10m",
+                "timezone": "Asia/Shanghai",
+            }
+        )
+        weather_url = f"https://api.open-meteo.com/v1/forecast?{weather_params}"
+        weather_req = urllib.request.Request(
+            weather_url,
+            method="GET",
+            headers={"User-Agent": "inventory-ai-chat/1.0"},
+        )
+        with urllib.request.urlopen(weather_req, timeout=timeout) as resp:
+            weather_raw = resp.read().decode("utf-8", errors="ignore")
+        weather_data = json.loads(weather_raw)
+        current = weather_data.get("current") or {}
+        if not current:
+            return None
+
+        temp = current.get("temperature_2m")
+        code = int(current.get("weather_code", 0) or 0)
+        wind = current.get("wind_speed_10m")
+        time_text = str(current.get("time") or "").strip()
+        weather_text = _weather_code_to_text(code)
+
+        snippet = f"{place_name} 当前{weather_text}，气温约 {temp}°C，风速约 {wind} km/h，时间 {time_text}。"
+        source = {
+            "title": f"Open-Meteo 实时天气（{place_name}）",
+            "snippet": snippet,
+            "url": "https://open-meteo.com/",
+            "reliability_level": "high",
+            "reliability_label": "高可信",
+            "reliability_reason": "公开气象 API 实时数据",
+            "domain": "open-meteo.com",
+        }
+        return {
+            "query": f"{location} 天气",
+            "sources": [source],
+        }
+    except Exception:
+        return None
+
+
+def _build_db_chat_answer(
+    question: str,
+    sql: str,
+    sql_reason: str,
+    rows: list[dict],
+    web_context: list[dict],
+    settings: dict,
+) -> str:
+    if not rows:
+        return "查询已执行，但没有匹配数据。你可以换个条件，例如指定料号、时间范围或盒型。"
+
+    system_prompt = (
+        "你是库存分析助手。"
+        "请仅根据提供的 SQL 结果回答，禁止虚构不存在的数据。"
+        "回答用简明中文，优先给结论，再给关键证据。"
+        "若提供了联网参考，请明确标注为参考信息，不得当作数据库事实。"
+    )
+    user_prompt = (
+        "用户问题:\n"
+        + question
+        + "\n\n执行SQL:\n"
+        + sql
+        + "\n\nSQL说明:\n"
+        + sql_reason
+        + "\n\n查询结果(JSON):\n"
+        + json.dumps(rows, ensure_ascii=False)
+    )
+    if web_context:
+        user_prompt += "\n\n联网参考(JSON):\n" + json.dumps(web_context, ensure_ascii=False)
+
+    return _call_siliconflow_chat(
+        system_prompt,
+        user_prompt,
+        api_url=settings["api_url"],
+        model=settings["model"],
+        api_key=settings["api_key"],
+        timeout=settings["timeout"],
+    )
+
+
+def _is_general_chat_question(question: str) -> bool:
+    text = (question or "").strip().lower()
+    if not text:
+        return False
+
+    general_hints = [
+        "你是什么",
+        "你是谁",
+        "你能做什么",
+        "聊天",
+        "模型",
+        "介绍",
+        "hello",
+        "hi",
+        "你好",
+        "谢谢",
+    ]
+    data_hints = [
+        "库存",
+        "器件",
+        "料号",
+        "盒",
+        "袋",
+        "出库",
+        "入库",
+        "数量",
+        "part_no",
+        "component",
+        "box",
+    ]
+
+    has_general_hint = any(hint in text for hint in general_hints)
+    has_data_hint = any(hint in text for hint in data_hints)
+    return has_general_hint and not has_data_hint
+
+
+def _build_general_chat_answer(
+    question: str,
+    history: list[dict],
+    web_context: list[dict],
+    memory_text: str,
+    settings: dict,
+) -> str:
+    q = (question or "").strip()
+    q_lower = q.lower()
+    if "模型" in q or "model" in q_lower:
+        model_name = str(settings.get("model", "") or "未配置")
+        return f"当前系统配置的 AI 模型是：{model_name}。"
+
+    history_lines = []
+    for row in history[-8:]:
+        role = str(row.get("role", "user") or "user")
+        content = str(row.get("content", "") or "").strip()
+        if not content:
+            continue
+        history_lines.append(f"[{role}] {content[:220]}")
+
+    system_prompt = (
+        "你是系统内置通用助手。"
+        "请直接回答用户问题，不要把自己限制为仅库存问题。"
+        "对于时效性问题（如天气、新闻、实时价格），若提供了联网参考就优先依据参考；"
+        "若没有联网参考，需明确说明可能不够实时。"
+        "回答保持简洁中文。"
+    )
+    user_prompt = (
+        "对话历史:\n"
+        + "\n".join(history_lines or ["(无)"])
+        + "\n\n用户当前问题:\n"
+        + question
+    )
+    if memory_text:
+        user_prompt += "\n\n本地记忆摘要:\n" + memory_text
+    if web_context:
+        user_prompt += "\n\n联网参考(JSON):\n" + json.dumps(web_context, ensure_ascii=False)
+
+    return _call_siliconflow_chat(
+        system_prompt,
+        user_prompt,
+        api_url=settings["api_url"],
+        model=settings["model"],
+        api_key=settings["api_key"],
+        timeout=settings["timeout"],
+    )
+
+
+@app.route("/ai/chat")
+def ai_chat_page():
+    return render_template("ai_chat.html")
+
+
+@app.route("/ai/chat/memory/clear", methods=["POST"])
+def ai_chat_memory_clear():
+    username = (session.get("username") or "guest").strip() or "guest"
+    _clear_ai_chat_memory(username)
+    _log_event(logging.INFO, "ai_chat_memory_cleared", user=username)
+    return {"ok": True, "message": "本地记忆已清空"}
+
+
+@app.route("/ai/chat/query", methods=["POST"])
+def ai_chat_query():
+    payload = request.get_json(silent=True) or {}
+    question = str(payload.get("question", "") or "").strip()
+    history = payload.get("history", [])
+    allow_web_search = _is_truthy_form_value(str(payload.get("allow_web_search", "")))
+    allow_db_query = _is_truthy_form_value(str(payload.get("allow_db_query", "")))
+    if not isinstance(history, list):
+        history = []
+
+    if not question:
+        return {"ok": False, "message": "请输入问题"}, 400
+
+    username = (session.get("username") or "guest").strip() or "guest"
+    memory_payload = _get_ai_chat_memory(username)
+    memory_text = _memory_context_text(memory_payload)
+
+    settings = _get_ai_settings()
+    if not settings.get("api_key") or not settings.get("api_url") or not settings.get("model"):
+        return {"ok": False, "message": "AI 参数不完整，请先到参数页配置"}, 400
+
+    if not allow_db_query:
+        web_context = []
+        if allow_web_search:
+            web_context = _build_db_chat_web_context(question, [], timeout=settings.get("timeout", 30), max_queries=1)
+            if not web_context and ("天气" in question or "weather" in question.lower()):
+                weather_ctx = _fetch_weather_context(question, timeout=settings.get("timeout", 30))
+                if weather_ctx:
+                    web_context = [weather_ctx]
+        answer = _build_general_chat_answer(question, history, web_context, memory_text, settings)
+        _append_ai_chat_memory(username, question, answer)
+        _log_event(
+            logging.INFO,
+            "ai_chat_general_success",
+            question=question,
+            allow_web_search=allow_web_search,
+            allow_db_query=allow_db_query,
+            web_sources=sum(len(item.get("sources", [])) for item in web_context),
+            memory_notes=len(memory_payload.get("notes", [])),
+        )
+        return {
+            "ok": True,
+            "answer": answer,
+            "sql": "",
+            "planner_reason": "未启用数据库查询，按通用问答模式处理",
+            "planner_raw": "",
+            "row_count": 0,
+            "rows_preview": [],
+            "web_context": web_context,
+            "allow_web_search": allow_web_search,
+            "allow_db_query": allow_db_query,
+            "chat_mode": "general",
+        }
+
+    try:
+        sql_raw, planner_reason, planner_text = _build_db_chat_sql_plan(
+            question,
+            history,
+            memory_text,
+            settings,
+        )
+    except Exception as exc:
+        _log_event(
+            logging.WARNING,
+            "ai_chat_plan_error",
+            error=str(exc),
+            question=question,
+            model=settings.get("model", ""),
+        )
+        if not _is_general_chat_question(question):
+            return {"ok": False, "message": f"SQL 规划失败: {exc}"}, 400
+
+        web_context = []
+        if allow_web_search:
+            web_context = _build_db_chat_web_context(question, [], timeout=settings.get("timeout", 30), max_queries=1)
+            if not web_context and ("天气" in question or "weather" in question.lower()):
+                weather_ctx = _fetch_weather_context(question, timeout=settings.get("timeout", 30))
+                if weather_ctx:
+                    web_context = [weather_ctx]
+        answer = _build_general_chat_answer(question, history, web_context, memory_text, settings)
+        _append_ai_chat_memory(username, question, answer)
+        return {
+            "ok": True,
+            "answer": answer,
+            "sql": "",
+            "planner_reason": "已切换到通用对话模式",
+            "planner_raw": "",
+            "row_count": 0,
+            "rows_preview": [],
+            "web_context": web_context,
+            "allow_web_search": allow_web_search,
+            "allow_db_query": allow_db_query,
+            "chat_mode": "general",
+        }
+
+    safe, reject_reason, safe_sql = _is_safe_readonly_sql(sql_raw)
+    if not safe:
+        _log_event(
+            logging.WARNING,
+            "ai_chat_sql_rejected",
+            reason=reject_reason,
+            sql=sql_raw,
+            question=question,
+        )
+        can_fallback_general = (
+            "未包含可识别的数据表" in reject_reason
+            or _is_general_chat_question(question)
+        )
+        if not can_fallback_general:
+            return {"ok": False, "message": f"SQL 被安全策略拒绝: {reject_reason}"}, 400
+
+        web_context = []
+        if allow_web_search:
+            web_context = _build_db_chat_web_context(question, [], timeout=settings.get("timeout", 30), max_queries=1)
+            if not web_context and ("天气" in question or "weather" in question.lower()):
+                weather_ctx = _fetch_weather_context(question, timeout=settings.get("timeout", 30))
+                if weather_ctx:
+                    web_context = [weather_ctx]
+        answer = _build_general_chat_answer(question, history, web_context, memory_text, settings)
+        _append_ai_chat_memory(username, question, answer)
+        return {
+            "ok": True,
+            "answer": answer,
+            "sql": "",
+            "planner_reason": "已切换到通用对话模式",
+            "planner_raw": planner_text,
+            "row_count": 0,
+            "rows_preview": [],
+            "web_context": web_context,
+            "allow_web_search": allow_web_search,
+            "allow_db_query": allow_db_query,
+            "chat_mode": "general",
+        }
+
+    final_sql = _ensure_query_limit(safe_sql, row_limit=80)
+    try:
+        query_result = db.session.execute(db.text(final_sql))
+        columns = list(query_result.keys())
+        rows = query_result.fetchall()
+    except Exception as exc:
+        _log_event(
+            logging.ERROR,
+            "ai_chat_query_execute_error",
+            error=str(exc),
+            sql=final_sql,
+            question=question,
+            traceback=traceback.format_exc(),
+        )
+        return {"ok": False, "message": "SQL 执行失败，请调整问题后重试"}, 400
+
+    serialized_rows = _serialize_sql_rows(rows, columns)
+    web_context = []
+    if allow_web_search:
+        try:
+            web_context = _build_db_chat_web_context(
+                question,
+                serialized_rows,
+                timeout=settings.get("timeout", 30),
+            )
+        except Exception as exc:
+            _log_event(
+                logging.WARNING,
+                "ai_chat_web_context_error",
+                error=str(exc),
+                question=question,
+            )
+            web_context = []
+
+    try:
+        answer = _build_db_chat_answer(
+            question,
+            final_sql,
+            planner_reason,
+            serialized_rows,
+            web_context,
+            settings,
+        )
+    except Exception as exc:
+        _log_event(
+            logging.WARNING,
+            "ai_chat_answer_error",
+            error=str(exc),
+            sql=final_sql,
+            row_count=len(serialized_rows),
+        )
+        answer = (
+            "查询已完成，但 AI 总结超时/失败。"
+            f"共返回 {len(serialized_rows)} 行数据，你可以查看本次 SQL 并缩小问题范围后重试。"
+        )
+
+    _append_ai_chat_memory(username, question, answer)
+
+    _log_event(
+        logging.INFO,
+        "ai_chat_query_success",
+        question=question,
+        sql=final_sql,
+        row_count=len(serialized_rows),
+        web_sources=sum(len(item.get("sources", [])) for item in web_context),
+        memory_notes=len(memory_payload.get("notes", [])),
+    )
+    return {
+        "ok": True,
+        "answer": answer,
+        "sql": final_sql,
+        "planner_reason": planner_reason,
+        "planner_raw": planner_text,
+        "row_count": len(serialized_rows),
+        "rows_preview": serialized_rows[:8],
+        "web_context": web_context,
+        "allow_web_search": allow_web_search,
+        "allow_db_query": allow_db_query,
+        "chat_mode": "db",
+    }
+
+
 def _is_safe_next_path(path: str) -> bool:
     candidate = (path or "").strip()
     if not candidate:
diff --git a/data/ai_settings.json b/data/ai_settings.json
index 995b9a3..2c33b7f 100644
--- a/data/ai_settings.json
+++ b/data/ai_settings.json
@@ -1,6 +1,6 @@
 {
   "api_url": "https://api.siliconflow.cn/v1/chat/completions",
-  "model": "Pro/zai-org/GLM-5",
+  "model": "Pro/moonshotai/Kimi-K2.5",
   "api_key": "sk-pekgnbdvwgydxzteabnykswjadkitoopwcekmksydfoslmlo",
   "timeout": 120,
   "restock_threshold": 2,
diff --git a/static/css/style.css b/static/css/style.css
index b147816..1ce36ea 100644
--- a/static/css/style.css
+++ b/static/css/style.css
@@ -618,6 +618,86 @@ body {
     overflow: auto;
 }
 
+.ai-chat-shell {
+    display: grid;
+    gap: var(--space-2);
+}
+
+.ai-chat-messages {
+    display: grid;
+    gap: var(--space-1);
+    max-height: 58vh;
+    overflow: auto;
+    padding-right: 4px;
+}
+
+.ai-chat-item {
+    border: 1px solid var(--line);
+    border-radius: var(--radius);
+    background: color-mix(in srgb, var(--card) 88%, var(--card-alt));
+    padding: var(--space-1) var(--space-2);
+}
+
+.ai-chat-item.user {
+    border-color: color-mix(in srgb, var(--accent) 55%, var(--line));
+}
+
+.ai-chat-item h3 {
+    margin: 0 0 6px;
+    font-size: 14px;
+}
+
+.ai-chat-item p {
+    margin: 0;
+}
+
+.md-content h1,
+.md-content h2,
+.md-content h3 {
+    margin: 0 0 8px;
+    line-height: 1.35;
+}
+
+.md-content p {
+    margin: 0 0 8px;
+}
+
+.md-content ul {
+    margin: 0 0 8px;
+    padding-left: 20px;
+}
+
+.md-content li {
+    margin: 4px 0;
+}
+
+.md-content .md-code {
+    margin: 8px 0;
+    border: 1px solid var(--line);
+    border-radius: var(--radius);
+    background: color-mix(in srgb, var(--card) 90%, var(--card-alt));
+    padding: 10px;
+    overflow: auto;
+}
+
+.md-content code {
+    font: 12px/1.45 Consolas, "Cascadia Mono", monospace;
+}
+
+.md-content a {
+    color: var(--accent-press);
+    text-decoration: none;
+}
+
+.md-content a:hover {
+    text-decoration: underline;
+}
+
+.ai-chat-form-wrap {
+    border-top: 1px dashed var(--line);
+    padding-top: var(--space-2);
+}
+
 .box-list {
     display: grid;
     grid-template-columns: repeat(auto-fill, minmax(300px, 1fr));
diff --git a/templates/ai_chat.html b/templates/ai_chat.html
new file mode 100644
index 0000000..4275421
--- /dev/null
+++ b/templates/ai_chat.html
@@ -0,0 +1,326 @@
+<!DOCTYPE html>
+<html lang="zh-CN">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>AI数据库聊天</title>
+    <link rel="stylesheet" href="{{ url_for('static', filename='css/style.css') }}">
+</head>
+<body>
+    <header class="hero slim">
+        <div>
+            <h1>AI数据库聊天</h1>
+            <p>用自然语言提问，系统会生成只读 SQL 查询库存数据并给出结论</p>
+        </div>
+        <div class="hero-actions">
+            <a class="btn btn-light" href="{{ url_for('types_page') }}">返回仓库概览</a>
+            <a class="btn btn-light" href="{{ url_for('ai_settings_page') }}">AI参数</a>
+            <a class="btn btn-light" href="{{ url_for('system_logs_page') }}">系统日志</a>
+            {% include '_account_menu.html' %}
+        </div>
+    </header>
+
+    <main class="container">
+        <section class="panel ai-chat-shell">
+            <div class="ai-chat-messages" id="ai-chat-messages"></div>
+            <div class="ai-chat-form-wrap">
+                <p class="hint" id="ai-chat-status">提示：默认是通用聊天 + 联网补充；勾选“启用数据库查询”后再做库存数据分析。</p>
+                <textarea id="ai-chat-input" class="batch-input" rows="4" placeholder="请输入问题，例如：低库存里哪些器件近30天出库最快？"></textarea>
+                <label>
+                    <input type="checkbox" id="ai-chat-web-search" checked>
+                    允许联网补充（回答中会区分“数据库结论”和“联网参考”）
+                </label>
+                <label>
+                    <input type="checkbox" id="ai-chat-db-query">
+                    启用数据库查询（不勾选则不执行 SQL，仅通用问答）
+                </label>
+                <div class="actions">
+                    <button class="btn" id="ai-chat-send" type="button">发送</button>
+                    <button class="btn btn-light" id="ai-chat-clear" type="button">清空会话</button>
+                    <button class="btn btn-light" id="ai-chat-clear-memory" type="button">清空本地记忆</button>
+                </div>
+            </div>
+        </section>
+    </main>
+
+    <script>
+        (function () {
+            var sendBtn = document.getElementById('ai-chat-send');
+            var clearBtn = document.getElementById('ai-chat-clear');
+            var clearMemoryBtn = document.getElementById('ai-chat-clear-memory');
+            var input = document.getElementById('ai-chat-input');
+            var webSearchToggle = document.getElementById('ai-chat-web-search');
+            var dbQueryToggle = document.getElementById('ai-chat-db-query');
+            var statusNode = document.getElementById('ai-chat-status');
+            var messagesNode = document.getElementById('ai-chat-messages');
+            var history = [];
+
+            function escapeHtml(text) {
+                return String(text || '').replace(/[&<>"']/g, function (ch) {
+                    return ({'&': '&amp;', '<': '&lt;', '>': '&gt;', '"': '&quot;', "'": '&#39;'})[ch];
+                });
+            }
+
+            function renderInlineMarkdown(escapedText) {
+                var html = String(escapedText || '');
+                html = html.replace(/\[([^\]]+)\]\((https?:\/\/[^\s)]+)\)/g, '<a href="$2" target="_blank" rel="noopener noreferrer">$1</a>');
+                html = html.replace(/\*\*([^*]+)\*\*/g, '<strong>$1</strong>');
+                html = html.replace(/`([^`]+)`/g, '<code>$1</code>');
+                return html;
+            }
+
+            function renderMarkdown(text) {
+                var source = String(text || '').replace(/\r\n?/g, '\n');
+                var codeBlocks = [];
+
+                source = source.replace(/```([a-zA-Z0-9_-]+)?\n([\s\S]*?)```/g, function (_, lang, code) {
+                    var language = escapeHtml(lang || 'text');
+                    var body = escapeHtml(code || '');
+                    var token = '@@CODEBLOCK_' + codeBlocks.length + '@@';
+                    codeBlocks.push('<pre class="md-code"><code class="lang-' + language + '">' + body + '</code></pre>');
+                    return token;
+                });
+
+                var escaped = escapeHtml(source);
+                var lines = escaped.split('\n');
+                var htmlParts = [];
+                var paragraph = [];
+                var inList = false;
+
+                function flushParagraph() {
+                    if (!paragraph.length) {
+                        return;
+                    }
+                    htmlParts.push('<p>' + renderInlineMarkdown(paragraph.join('<br>')) + '</p>');
+                    paragraph = [];
+                }
+
+                function closeList() {
+                    if (inList) {
+                        htmlParts.push('</ul>');
+                        inList = false;
+                    }
+                }
+
+                lines.forEach(function (line) {
+                    var trimmed = line.trim();
+                    if (!trimmed) {
+                        flushParagraph();
+                        closeList();
+                        return;
+                    }
+
+                    var heading = trimmed.match(/^(#{1,3})\s+(.+)$/);
+                    if (heading) {
+                        flushParagraph();
+                        closeList();
+                        var level = heading[1].length;
+                        htmlParts.push('<h' + level + '>' + renderInlineMarkdown(heading[2]) + '</h' + level + '>');
+                        return;
+                    }
+
+                    var listItem = trimmed.match(/^[-*]\s+(.+)$/);
+                    if (listItem) {
+                        flushParagraph();
+                        if (!inList) {
+                            htmlParts.push('<ul>');
+                            inList = true;
+                        }
+                        htmlParts.push('<li>' + renderInlineMarkdown(listItem[1]) + '</li>');
+                        return;
+                    }
+
+                    closeList();
+                    paragraph.push(trimmed);
+                });
+
+                flushParagraph();
+                closeList();
+
+                var html = htmlParts.join('');
+                codeBlocks.forEach(function (block, index) {
+                    html = html.replace('@@CODEBLOCK_' + index + '@@', block);
+                });
+                return html || '<p></p>';
+            }
+
+            function appendMessage(role, title, content, extraHtml) {
+                var card = document.createElement('article');
+                card.className = 'ai-chat-item ' + role;
+                var bodyHtml = role === 'assistant'
+                    ? '<div class="md-content">' + renderMarkdown(content) + '</div>'
+                    : '<p>' + escapeHtml(content) + '</p>';
+                card.innerHTML =
+                    '<h3>' + escapeHtml(title) + '</h3>' +
+                    bodyHtml +
+                    (extraHtml || '');
+                messagesNode.appendChild(card);
+                messagesNode.scrollTop = messagesNode.scrollHeight;
+            }
+
+            function toHistoryPayload() {
+                return history.slice(-8).map(function (item) {
+                    return { role: item.role, content: item.content };
+                });
+            }
+
+            function renderWebContextHtml(contexts) {
+                if (!Array.isArray(contexts) || !contexts.length) {
+                    return '';
+                }
+
+                var blocks = contexts.map(function (ctx) {
+                    var query = escapeHtml(ctx.query || '');
+                    var sources = Array.isArray(ctx.sources) ? ctx.sources : [];
+                    var itemsHtml = sources.map(function (src) {
+                        var title = escapeHtml(src.title || '未命名来源');
+                        var snippet = escapeHtml(src.snippet || '');
+                        var url = escapeHtml(src.url || '');
+                        var badgeLabel = escapeHtml(src.reliability_label || '中可信');
+                        var badgeLevel = escapeHtml(src.reliability_level || 'medium');
+                        var badgeClass = 'source-' + badgeLevel;
+                        var reason = escapeHtml(src.reliability_reason || '来源类型未知，建议人工确认');
+                        var domain = escapeHtml(src.domain || '-');
+                        var linkHtml = url ? '<a href="' + url + '" target="_blank" rel="noopener noreferrer">查看来源</a>' : '';
+
+                        return (
+                            '<li>' +
+                                '<div class="source-head">' +
+                                    '<strong>' + title + '</strong>' +
+                                    '<span class="source-badge ' + badgeClass + '">' + badgeLabel + '</span>' +
+                                '</div>' +
+                                '<p>' + snippet + '</p>' +
+                                '<p>依据: ' + reason + ' | 域名: ' + domain + '</p>' +
+                                linkHtml +
+                            '</li>'
+                        );
+                    }).join('');
+
+                    return (
+                        '<section class="panel">' +
+                            '<h4>检索词: ' + query + '</h4>' +
+                            '<ul class="ai-source-items">' + itemsHtml + '</ul>' +
+                        '</section>'
+                    );
+                }).join('');
+
+                return '<details class="box-overview"><summary>联网来源明细</summary><div class="ai-source-list">' + blocks + '</div></details>';
+            }
+
+            sendBtn.addEventListener('click', function () {
+                var question = (input.value || '').trim();
+                if (!question) {
+                    statusNode.textContent = '请输入问题后再发送。';
+                    return;
+                }
+
+                sendBtn.disabled = true;
+                input.value = '';
+                var useWeb = !!(webSearchToggle && webSearchToggle.checked);
+                var useDb = !!(dbQueryToggle && dbQueryToggle.checked);
+                if (useDb && useWeb) {
+                    statusNode.textContent = 'AI 正在查询数据库并联网补充，请稍候...';
+                } else if (useDb) {
+                    statusNode.textContent = 'AI 正在查询数据库，请稍候...';
+                } else if (useWeb) {
+                    statusNode.textContent = 'AI 正在通用问答并联网补充，请稍候...';
+                } else {
+                    statusNode.textContent = 'AI 正在通用问答，请稍候...';
+                }
+                appendMessage('user', '你', question, '');
+
+                fetch('{{ url_for("ai_chat_query") }}', {
+                    method: 'POST',
+                    headers: { 'Content-Type': 'application/json' },
+                    body: JSON.stringify({
+                        question: question,
+                        history: toHistoryPayload(),
+                        allow_web_search: useWeb,
+                        allow_db_query: useDb
+                    })
+                })
+                .then(function (resp) {
+                    return resp.json().then(function (data) {
+                        return { ok: resp.ok, data: data || {} };
+                    });
+                })
+                .then(function (result) {
+                    var data = result.data || {};
+                    if (!result.ok || !data.ok) {
+                        statusNode.textContent = '提问失败';
+                        appendMessage('assistant', 'AI助手', data.message || '请求失败，请稍后重试。', '');
+                        return;
+                    }
+
+                    var extra = '';
+                    if (data.sql) {
+                        extra += '<details class="box-overview"><summary>本次SQL</summary><pre class="ai-panel-content">' + escapeHtml(data.sql) + '</pre></details>';
+                    }
+                    if (data.planner_reason && data.chat_mode !== 'general') {
+                        extra += '<p class="hint">SQL思路: ' + escapeHtml(data.planner_reason) + '</p>';
+                    }
+                    if (typeof data.row_count === 'number' && data.chat_mode !== 'general') {
+                        extra += '<p class="hint">返回行数: ' + data.row_count + '</p>';
+                    }
+                    if (data.chat_mode === 'general') {
+                        extra += '<p class="hint">模式: 通用对话' + (data.allow_web_search ? ' + 联网补充' : '') + '</p>';
+                    } else if (data.allow_web_search) {
+                        extra += '<p class="hint">模式: 数据库 + 联网补充</p>';
+                    } else {
+                        extra += '<p class="hint">模式: 仅数据库</p>';
+                    }
+                    extra += renderWebContextHtml(data.web_context || []);
+
+                    appendMessage('assistant', 'AI助手', data.answer || '已完成查询。', extra);
+                    history.push({ role: 'user', content: question });
+                    history.push({ role: 'assistant', content: data.answer || '' });
+                    statusNode.textContent = '已完成本轮查询。';
+                })
+                .catch(function () {
+                    statusNode.textContent = '提问失败';
+                    appendMessage('assistant', 'AI助手', '网络请求失败，请稍后重试。', '');
+                })
+                .finally(function () {
+                    sendBtn.disabled = false;
+                });
+            });
+
+            clearBtn.addEventListener('click', function () {
+                history = [];
+                messagesNode.innerHTML = '';
+                input.value = '';
+                statusNode.textContent = '会话已清空。';
+            });
+
+            if (clearMemoryBtn) {
+                clearMemoryBtn.addEventListener('click', function () {
+                    clearMemoryBtn.disabled = true;
+                    fetch('{{ url_for("ai_chat_memory_clear") }}', {
+                        method: 'POST',
+                        headers: { 'Content-Type': 'application/json' },
+                        body: '{}'
+                    })
+                    .then(function (resp) {
+                        return resp.json().then(function (data) {
+                            return { ok: resp.ok, data: data || {} };
+                        });
+                    })
+                    .then(function (result) {
+                        if (!result.ok || !result.data.ok) {
+                            statusNode.textContent = '清空本地记忆失败';
+                            return;
+                        }
+                        statusNode.textContent = '本地记忆已清空。';
+                    })
+                    .catch(function () {
+                        statusNode.textContent = '清空本地记忆失败';
+                    })
+                    .finally(function () {
+                        clearMemoryBtn.disabled = false;
+                    });
+                });
+            }
+        })();
+    </script>
+</body>
+</html>
diff --git a/templates/types.html b/templates/types.html
index dbefb78..1cfd020 100644
--- a/templates/types.html
+++ b/templates/types.html
@@ -92,6 +92,7 @@
                 <div class="ai-panel-head">
                     <h2>AI补货建议</h2>
                     <div class="hero-actions">
+                        <a class="btn btn-light" href="{{ url_for('ai_chat_page') }}">聊天</a>
                         <a class="btn btn-light" href="{{ url_for('system_logs_page') }}">日志</a>
                         <a class="btn btn-light" href="{{ url_for('ai_settings_page') }}">参数</a>
                     </div>