diff --git a/.env.example b/.env.example
index faf7291..c89b184 100644
--- a/.env.example
+++ b/.env.example
@@ -5,6 +5,11 @@ APP_ENCRYPTION_KEY=
 # the system settings; all accounts use the same backend market snapshot.
 TUSHARE_TOKEN=your_tushare_token_here
 
+# Optional xiaobai-datahub client. All DATAHUB_READ_* / DATAHUB_SHADOW_* flags
+# default off in config/datahub.config.json, so the website keeps using Tushare.
+DATAHUB_BASE_URL=http://127.0.0.1:8766
+DATAHUB_TOKEN=
+
 # Optional iFinD HTTP credential. The backend exchanges it for a short-lived
 # access token and never exposes either token to browsers.
 IFIND_REFRESH_TOKEN=your_ifind_refresh_token_here
diff --git a/backend/bootstrap/settings.py b/backend/bootstrap/settings.py
index 37db457..80e952a 100644
--- a/backend/bootstrap/settings.py
+++ b/backend/bootstrap/settings.py
@@ -11,6 +11,8 @@ from backend.features.accounts.security import SecretVault
 def environment_credentials(environment: Mapping[str, str]) -> dict[str, str]:
     return {
         "tushare_token": str(environment.get("TUSHARE_TOKEN") or "").strip(),
+        "datahub_token": str(environment.get("DATAHUB_TOKEN") or "").strip(),
+        "datahub_base_url": str(environment.get("DATAHUB_BASE_URL") or "").strip(),
         "ifind_refresh_token": str(environment.get("IFIND_REFRESH_TOKEN") or "").strip(),
         "ifind_access_token": str(environment.get("IFIND_ACCESS_TOKEN") or "").strip(),
         "platform_llm_primary_api_key": str(
diff --git a/backend/data/datahub/__init__.py b/backend/data/datahub/__init__.py
new file mode 100644
index 0000000..71e0fbd
--- /dev/null
+++ b/backend/data/datahub/__init__.py
@@ -0,0 +1,15 @@
+from backend.data.datahub.bridge import DatahubAwareTushareClient, DatahubBridge
+from backend.data.datahub.client import DatahubClient, DatahubResponse
+from backend.data.datahub.errors import DatahubError
+from backend.data.datahub.settings import DATASETS, DatahubSettings, DatasetFlags
+
+__all__ = [
+    "DATASETS",
+    "DatahubAwareTushareClient",
+    "DatahubBridge",
+    "DatahubClient",
+    "DatahubError",
+    "DatahubResponse",
+    "DatahubSettings",
+    "DatasetFlags",
+]
diff --git a/backend/data/datahub/bridge.py b/backend/data/datahub/bridge.py
new file mode 100644
index 0000000..068ca30
--- /dev/null
+++ b/backend/data/datahub/bridge.py
@@ -0,0 +1,249 @@
+from __future__ import annotations
+
+import logging
+import sys
+from typing import Any, Callable
+
+from backend.data.datahub.client import DatahubClient, DatahubResponse
+from backend.data.datahub.compare import compare_rows
+from backend.data.datahub.errors import DatahubError
+from backend.data.datahub.native import (
+    API_TO_DATASET,
+    filter_calendar_rows,
+    filter_stock_rows,
+    project_fields,
+    to_native_rows,
+    yyyymmdd,
+)
+from backend.data.datahub.redact import redact_text, redact_value
+from backend.data.datahub.settings import DatahubSettings
+from backend.data.providers.tushare_client import TushareClient
+
+LOGGER = logging.getLogger("xiaobai.datahub")
+ShadowSink = Callable[[dict[str, Any]], None]
+EMPTY_FAIL_DATASETS = {"stocks", "daily", "index_daily", "valuation", "moneyflow", "auction"}
+
+
+def looks_like_heaven(module_name: str, filename: str = "") -> bool:
+    path = filename.replace("\\", "/")
+    return module_name.startswith("backend.features.heaven") or "/features/heaven/" in path
+
+
+def caller_is_heaven(depth: int = 24) -> bool:
+    frame = sys._getframe(1)
+    for _ in range(depth):
+        frame = frame.f_back if frame is not None else None
+        if frame is None:
+            return False
+        name = str(frame.f_globals.get("__name__") or "")
+        filename = str(frame.f_code.co_filename or "")
+        if looks_like_heaven(name, filename):
+            return True
+    return False
+
+
+class DatahubBridge:
+    def __init__(
+        self,
+        settings: DatahubSettings,
+        client: DatahubClient,
+        shadow_sink: ShadowSink | None = None,
+        heaven_guard: Callable[[], bool] | None = None,
+    ) -> None:
+        self.settings = settings
+        self.client = client
+        self.shadow_sink = shadow_sink
+        self.heaven_guard = heaven_guard or caller_is_heaven
+
+    def dataset_status(self, trade_date: str) -> list[dict[str, Any]] | None:
+        flags = self.settings.flags("status")
+        if not flags.read and not flags.shadow:
+            return None
+        try:
+            response = self._require_fresh(self.client.dataset_status(yyyymmdd(trade_date)), "status")
+            rows = list(response.data or [])
+            if flags.shadow:
+                self._emit_shadow(compare_rows("status", [], rows, response.meta))
+            if flags.read:
+                return rows
+            return None
+        except Exception as exc:
+            self._log_failure("status", exc)
+            if flags.shadow:
+                self._emit_shadow(compare_rows("status", [], [], {}, self._error_text(exc)))
+            return None
+
+    def batches(self, trade_date: str, dataset: str = "") -> list[dict[str, Any]] | None:
+        flags = self.settings.flags("status")
+        if not flags.read:
+            return None
+        try:
+            response = self._require_fresh(
+                self.client.batches(yyyymmdd(trade_date), dataset),
+                "status",
+            )
+            return list(response.data or [])
+        except Exception as exc:
+            self._log_failure("status", exc)
+            return None
+
+    def query(
+        self,
+        api_name: str,
+        params: dict[str, Any] | None,
+        fields: str,
+        legacy_query: Callable[..., list[dict[str, Any]]],
+    ) -> list[dict[str, Any]]:
+        dataset = API_TO_DATASET.get(api_name)
+        if not dataset or self.heaven_guard():
+            return legacy_query(api_name, params, fields)
+        flags = self.settings.flags(dataset)
+        if not flags.read and not flags.shadow:
+            return legacy_query(api_name, params, fields)
+
+        hub_rows: list[dict[str, Any]] | None = None
+        hub_meta: dict[str, Any] = {}
+        hub_error: str | None = None
+        hub_canonical: list[dict[str, Any]] = []
+        try:
+            response = self._fetch_dataset(dataset, params or {})
+            hub_canonical = self._extract_rows(dataset, response, params or {})
+            hub_rows = to_native_rows(dataset, hub_canonical)
+            hub_meta = dict(response.meta)
+            self._validate_usable(dataset, hub_rows, response)
+        except Exception as exc:
+            hub_error = self._error_text(exc)
+            self._log_failure(dataset, exc)
+
+        if flags.shadow:
+            try:
+                legacy_rows = legacy_query(api_name, params, fields)
+            except Exception as exc:
+                if flags.read and hub_rows is not None and hub_error is None:
+                    self._emit_shadow(compare_rows(dataset, [], hub_canonical, hub_meta, self._error_text(exc)))
+                    return project_fields(hub_rows, fields)
+                raise
+            self._emit_shadow(compare_rows(dataset, legacy_rows, hub_canonical, hub_meta, hub_error))
+            if flags.read and hub_rows is not None and hub_error is None:
+                return project_fields(hub_rows, fields)
+            return legacy_rows
+
+        if flags.read and hub_rows is not None and hub_error is None:
+            return project_fields(hub_rows, fields)
+        return legacy_query(api_name, params, fields)
+
+    def _fetch_dataset(self, dataset: str, params: dict[str, Any]) -> DatahubResponse:
+        date = yyyymmdd(params.get("trade_date") or params.get("date"))
+        start = yyyymmdd(params.get("start_date") or params.get("from") or date)
+        end = yyyymmdd(params.get("end_date") or params.get("to") or date)
+        code = str(params.get("ts_code") or params.get("code") or "").strip()
+        if dataset == "calendar":
+            if not start or not end:
+                raise DatahubError("INVALID_ARGUMENT", "calendar requires start_date and end_date")
+            return self.client.calendar(start, end)
+        if dataset == "stocks":
+            return self._paginate(self.client.stocks, {})
+        fetchers = {
+            "daily": self.client.daily_bars,
+            "index_daily": self.client.index_bars,
+            "valuation": self.client.valuation,
+            "moneyflow": self.client.moneyflow,
+            "auction": self.client.auction,
+        }
+        fetcher = fetchers[dataset]
+        query: dict[str, Any] = {}
+        if code:
+            query["code"] = code
+        if date and not (params.get("start_date") or params.get("end_date")):
+            query["date"] = date
+        else:
+            if start:
+                query["from"] = start
+            if end:
+                query["to"] = end
+        if dataset == "daily":
+            query["adjust"] = "none"
+        return self._paginate(fetcher, query)
+
+    def _paginate(self, fetcher: Callable[..., DatahubResponse], params: dict[str, Any]) -> DatahubResponse:
+        limit = self.settings.page_limit
+        offset = 0
+        rows: list[Any] = []
+        meta: dict[str, Any] = {}
+        schema_version = 1
+        while True:
+            page = fetcher(**{**params, "limit": limit, "offset": offset})
+            meta = dict(page.meta)
+            schema_version = page.schema_version
+            data = page.data or []
+            if not isinstance(data, list):
+                raise DatahubError("INTERNAL", "datahub returned a non-list payload")
+            rows.extend(data)
+            if len(data) < limit:
+                break
+            offset += limit
+            if offset > 200_000:
+                break
+        return DatahubResponse(data=rows, meta=meta, schema_version=schema_version)
+
+    def _extract_rows(
+        self,
+        dataset: str,
+        response: DatahubResponse,
+        params: dict[str, Any],
+    ) -> list[dict[str, Any]]:
+        rows = [dict(item) for item in (response.data or [])]
+        if dataset == "calendar":
+            return filter_calendar_rows(rows, params)
+        if dataset == "stocks":
+            return filter_stock_rows(rows, params)
+        return rows
+
+    def _validate_usable(self, dataset: str, rows: list[dict[str, Any]], response: DatahubResponse) -> None:
+        meta = response.meta or {}
+        stale_seconds = int(meta.get("staleness_seconds") or 0)
+        if meta.get("stale") or stale_seconds > self.settings.stale_seconds_max:
+            raise DatahubError("STALE", f"{dataset} data is stale")
+        if dataset in EMPTY_FAIL_DATASETS and not rows:
+            raise DatahubError("EMPTY", f"{dataset} returned no rows")
+
+    def _require_fresh(self, response: DatahubResponse, dataset: str) -> DatahubResponse:
+        self._validate_usable(dataset, list(response.data or []) if isinstance(response.data, list) else [], response)
+        return response
+
+    def _emit_shadow(self, report: dict[str, Any]) -> None:
+        safe = redact_value(report, secrets=self.settings.secrets())
+        LOGGER.info("datahub shadow %s", safe)
+        if self.shadow_sink is not None:
+            self.shadow_sink(report)
+
+    def _log_failure(self, dataset: str, exc: Exception) -> None:
+        LOGGER.warning(
+            "datahub fallback dataset=%s error=%s",
+            dataset,
+            redact_text(self._error_text(exc), self.settings.secrets()),
+        )
+
+    def _error_text(self, exc: Exception) -> str:
+        if isinstance(exc, DatahubError):
+            text = f"{exc.code}: {exc.message}"
+        else:
+            text = str(exc)
+        return redact_text(text, self.settings.secrets())
+
+
+class DatahubAwareTushareClient:
+    def __init__(self, legacy: TushareClient, bridge: DatahubBridge) -> None:
+        self._legacy = legacy
+        self._bridge = bridge
+
+    def query(
+        self,
+        api_name: str,
+        params: dict[str, Any] | None = None,
+        fields: str = "",
+    ) -> list[dict[str, Any]]:
+        return self._bridge.query(api_name, params, fields, self._legacy.query)
+
+    def __getattr__(self, name: str) -> Any:
+        return getattr(self._legacy, name)
diff --git a/backend/data/datahub/client.py b/backend/data/datahub/client.py
new file mode 100644
index 0000000..7f247f2
--- /dev/null
+++ b/backend/data/datahub/client.py
@@ -0,0 +1,184 @@
+from __future__ import annotations
+
+import json
+import logging
+import urllib.error
+import urllib.parse
+import urllib.request
+from dataclasses import dataclass, field
+from typing import Any, Callable
+
+from backend.data.datahub.errors import DatahubError
+from backend.data.datahub.redact import redact_text
+from backend.data.datahub.settings import DatahubSettings
+
+LOGGER = logging.getLogger("xiaobai.datahub")
+UrlOpen = Callable[..., Any]
+
+
+@dataclass(frozen=True)
+class DatahubResponse:
+    data: Any
+    meta: dict[str, Any] = field(default_factory=dict)
+    schema_version: int = 1
+    status: int = 200
+
+
+class DatahubClient:
+    def __init__(
+        self,
+        settings: DatahubSettings,
+        urlopen: UrlOpen = urllib.request.urlopen,
+    ) -> None:
+        self.settings = settings
+        self._urlopen = urlopen
+
+    def health(self) -> DatahubResponse:
+        return self.get("/v1/health")
+
+    def calendar(self, start: str, end: str) -> DatahubResponse:
+        return self.get("/v1/calendar", {"from": start, "to": end})
+
+    def stocks(self, updated_since: str = "", limit: int | None = None, offset: int = 0) -> DatahubResponse:
+        params: dict[str, Any] = {"offset": offset, "limit": limit or self.settings.page_limit}
+        if updated_since:
+            params["updated_since"] = updated_since
+        return self.get("/v1/stocks", params)
+
+    def daily_bars(self, **params: Any) -> DatahubResponse:
+        return self.get("/v1/bars/daily", params)
+
+    def index_bars(self, **params: Any) -> DatahubResponse:
+        return self.get("/v1/indexes/bars", params)
+
+    def valuation(self, **params: Any) -> DatahubResponse:
+        return self.get("/v1/valuation", params)
+
+    def moneyflow(self, **params: Any) -> DatahubResponse:
+        return self.get("/v1/moneyflow", params)
+
+    def auction(self, **params: Any) -> DatahubResponse:
+        return self.get("/v1/auction", params)
+
+    def dataset_status(self, date: str) -> DatahubResponse:
+        return self.get("/v1/datasets/status", {"date": date})
+
+    def batches(self, date: str, dataset: str = "") -> DatahubResponse:
+        params: dict[str, Any] = {"date": date}
+        if dataset:
+            params["dataset"] = dataset
+        return self.get("/v1/batches", params)
+
+    def get(self, path: str, params: dict[str, Any] | None = None) -> DatahubResponse:
+        if not self.settings.token:
+            raise DatahubError("NOT_CONFIGURED", "DATAHUB_TOKEN is not configured")
+        query = {
+            key: value
+            for key, value in (params or {}).items()
+            if value is not None and value != ""
+        }
+        url = self.settings.base_url + path
+        if query:
+            url = f"{url}?{urllib.parse.urlencode(query)}"
+        attempts = 1 + max(0, self.settings.retries)
+        last_error: DatahubError | None = None
+        for attempt in range(attempts):
+            try:
+                return self._request(url)
+            except DatahubError as exc:
+                last_error = exc
+                if exc.code not in {"TIMEOUT", "UNAVAILABLE"} or attempt + 1 >= attempts:
+                    raise
+                LOGGER.warning(
+                    "datahub retry %s/%s %s",
+                    attempt + 1,
+                    attempts,
+                    redact_text(str(exc), self.settings.secrets()),
+                )
+        raise last_error or DatahubError("INTERNAL", "datahub request failed")
+
+    def _request(self, url: str) -> DatahubResponse:
+        request = urllib.request.Request(
+            url,
+            headers={
+                "Accept": "application/json",
+                "X-Datahub-Token": self.settings.token,
+                "User-Agent": "XiaobaiReviewDatahub/1.0",
+            },
+            method="GET",
+        )
+        try:
+            with self._urlopen(request, timeout=self.settings.timeout_seconds) as response:
+                status = int(getattr(response, "status", 200) or 200)
+                raw = response.read().decode("utf-8")
+        except TimeoutError as exc:
+            raise DatahubError("TIMEOUT", "datahub request timed out") from exc
+        except urllib.error.HTTPError as exc:
+            body = _read_error_body(exc)
+            raise _http_error(exc.code, body, self.settings.secrets()) from exc
+        except urllib.error.URLError as exc:
+            reason = redact_text(str(getattr(exc, "reason", exc)), self.settings.secrets())
+            if "timed out" in reason.lower():
+                raise DatahubError("TIMEOUT", "datahub request timed out") from exc
+            raise DatahubError("UNAVAILABLE", f"datahub unavailable: {reason}") from exc
+        payload = _parse_json(raw, self.settings.secrets())
+        return _as_response(payload, status, self.settings.secrets())
+
+
+def _parse_json(raw: str, secrets: tuple[str, ...]) -> dict[str, Any]:
+    try:
+        payload = json.loads(raw)
+    except json.JSONDecodeError as exc:
+        raise DatahubError("INTERNAL", "datahub returned invalid json") from exc
+    if not isinstance(payload, dict):
+        raise DatahubError("INTERNAL", "datahub returned a non-object payload")
+    return payload
+
+
+def _as_response(payload: dict[str, Any], status: int, secrets: tuple[str, ...]) -> DatahubResponse:
+    error = payload.get("error")
+    if isinstance(error, dict):
+        raise _mapped_error(str(error.get("code") or "INTERNAL"), str(error.get("message") or "datahub error"), status)
+    if status >= 400:
+        raise DatahubError("UNAVAILABLE", f"datahub http {status}", status)
+    return DatahubResponse(
+        data=payload.get("data"),
+        meta=dict(payload.get("meta") or {}),
+        schema_version=int(payload.get("schema_version") or 1),
+        status=status,
+    )
+
+
+def _http_error(status: int, payload: dict[str, Any], secrets: tuple[str, ...]) -> DatahubError:
+    error = payload.get("error") if isinstance(payload.get("error"), dict) else {}
+    code = str((error or {}).get("code") or "")
+    message = str((error or {}).get("message") or payload.get("message") or f"datahub http {status}")
+    message = redact_text(message, secrets)
+    if status == 401 or code == "UNAUTHORIZED":
+        return DatahubError("UNAUTHORIZED", message, status)
+    if status == 404 or code == "DATASET_NOT_PUBLISHED":
+        return DatahubError("DATASET_NOT_PUBLISHED", message, status)
+    if status == 400 or code == "INVALID_ARGUMENT":
+        return DatahubError("INVALID_ARGUMENT", message, status)
+    if status in {429, 503} or code in {"RATE_LIMITED", "SOURCE_UNAVAILABLE"}:
+        return DatahubError("UNAVAILABLE", message, status)
+    return DatahubError(code or "INTERNAL", message, status)
+
+
+def _mapped_error(code: str, message: str, status: int) -> DatahubError:
+    if code == "STALE_DATA":
+        return DatahubError("STALE", message, status)
+    if code in {"UNAUTHORIZED", "DATASET_NOT_PUBLISHED", "INVALID_ARGUMENT"}:
+        return DatahubError(code, message, status)
+    if code in {"RATE_LIMITED", "SOURCE_UNAVAILABLE"}:
+        return DatahubError("UNAVAILABLE", message, status)
+    return DatahubError(code or "INTERNAL", message, status)
+
+
+def _read_error_body(exc: urllib.error.HTTPError) -> dict[str, Any]:
+    try:
+        raw = exc.read().decode("utf-8")
+        payload = json.loads(raw)
+        return payload if isinstance(payload, dict) else {"message": raw}
+    except Exception:
+        return {"message": str(exc)}
diff --git a/backend/data/datahub/compare.py b/backend/data/datahub/compare.py
new file mode 100644
index 0000000..08b45de
--- /dev/null
+++ b/backend/data/datahub/compare.py
@@ -0,0 +1,134 @@
+from __future__ import annotations
+
+from typing import Any
+
+from backend.data.datahub.native import SCALE_FIELDS, row_key, to_canonical_row, yyyymmdd
+
+NUMERIC_TOLERANCE = 1e-4
+
+
+def compare_rows(
+    dataset: str,
+    legacy_rows: list[dict[str, Any]],
+    hub_rows: list[dict[str, Any]] | None,
+    hub_meta: dict[str, Any] | None = None,
+    hub_error: str | None = None,
+) -> dict[str, Any]:
+    hub = hub_rows or []
+    legacy_map = {row_key(dataset, row): row for row in legacy_rows}
+    hub_map = {row_key(dataset, _align_hub_row(row)): row for row in hub}
+    missing_hub = sorted(key for key in legacy_map if key not in hub_map)
+    missing_legacy = sorted(key for key in hub_map if key not in legacy_map)
+    value_diffs: list[dict[str, Any]] = []
+    unit_conversion: list[dict[str, Any]] = []
+    matched = 0
+    for key, legacy in legacy_map.items():
+        hub_row = hub_map.get(key)
+        if hub_row is None:
+            continue
+        field_report = _compare_fields(dataset, legacy, hub_row)
+        if field_report["unit_conversion"]:
+            unit_conversion.append({"key": list(key), "fields": field_report["unit_conversion"]})
+        if field_report["value_diff"]:
+            value_diffs.append({"key": list(key), "fields": field_report["value_diff"]})
+        if not field_report["unit_conversion"] and not field_report["value_diff"]:
+            matched += 1
+    stale_seconds = int((hub_meta or {}).get("staleness_seconds") or 0)
+    time_skew = bool((hub_meta or {}).get("stale")) or stale_seconds > 0
+    return {
+        "dataset": dataset,
+        "legacy_rows": len(legacy_rows),
+        "hub_rows": len(hub),
+        "matched": matched,
+        "missing_hub": [list(item) for item in missing_hub[:20]],
+        "missing_legacy": [list(item) for item in missing_legacy[:20]],
+        "missing_hub_count": len(missing_hub),
+        "missing_legacy_count": len(missing_legacy),
+        "value_diff_count": len(value_diffs),
+        "unit_conversion_count": len(unit_conversion),
+        "value_diffs": value_diffs[:20],
+        "unit_conversion": unit_conversion[:20],
+        "time_skew": time_skew,
+        "staleness_seconds": stale_seconds,
+        "published_at": (hub_meta or {}).get("published_at"),
+        "trade_date": yyyymmdd((hub_meta or {}).get("trade_date")),
+        "hub_error": hub_error,
+        "equal": (
+            not hub_error
+            and not missing_hub
+            and not missing_legacy
+            and not value_diffs
+            and not unit_conversion
+            and not time_skew
+        ),
+    }
+
+
+def _align_hub_row(row: dict[str, Any]) -> dict[str, Any]:
+    aligned = dict(row)
+    if "volume" in aligned and "vol" not in aligned:
+        aligned["vol"] = aligned.get("volume")
+    return aligned
+
+
+def _compare_fields(dataset: str, legacy: dict[str, Any], hub: dict[str, Any]) -> dict[str, list[dict[str, Any]]]:
+    canonical_legacy = to_canonical_row(dataset, legacy)
+    hub_canonical = _hub_canonical(dataset, hub)
+    native_hub = _align_hub_row(hub)
+    value_diff: list[dict[str, Any]] = []
+    unit_conversion: list[dict[str, Any]] = []
+    keys = (set(canonical_legacy) | set(hub_canonical)) - {"batch_id", "updated_at", "volume"}
+    scales = SCALE_FIELDS.get(dataset) or {}
+    for field in sorted(keys):
+        left = canonical_legacy.get(field)
+        right = hub_canonical.get(field)
+        if _same(left, right):
+            continue
+        native_left = legacy.get(field)
+        hub_raw = native_hub.get(field)
+        if field in scales and _near(_optional(native_left), _optional(hub_raw)):
+            unit_conversion.append(
+                {"field": field, "legacy": native_left, "hub": hub_raw, "reason": "unit_conversion"}
+            )
+            continue
+        value_diff.append({"field": field, "legacy": left, "hub": right, "reason": "value_diff"})
+    return {"value_diff": value_diff, "unit_conversion": unit_conversion}
+
+
+def _hub_canonical(dataset: str, row: dict[str, Any]) -> dict[str, Any]:
+    """Hub API rows are already canonical; only align field names."""
+    aligned = dict(row)
+    if "volume" in aligned and "vol" not in aligned:
+        aligned["vol"] = aligned.get("volume")
+    if dataset == "calendar":
+        is_open = aligned.get("is_open")
+        aligned["is_open"] = 1 if is_open in (True, 1, "1", "Y", "y") else 0
+        aligned["cal_date"] = yyyymmdd(aligned.get("cal_date"))
+        aligned["pretrade_date"] = yyyymmdd(aligned.get("pretrade_date")) or None
+        aligned["exchange"] = str(aligned.get("exchange") or "SSE")
+    return aligned
+
+
+def _same(left: Any, right: Any) -> bool:
+    if left in (None, "") and right in (None, ""):
+        return True
+    if isinstance(left, (int, float)) or isinstance(right, (int, float)):
+        return _near(_optional(left), _optional(right))
+    return str(left or "") == str(right or "")
+
+
+def _near(left: float | None, right: float | None) -> bool:
+    if left is None and right is None:
+        return True
+    if left is None or right is None:
+        return False
+    return abs(left - right) <= max(NUMERIC_TOLERANCE, abs(left) * 1e-9, abs(right) * 1e-9)
+
+
+def _optional(value: Any) -> float | None:
+    if value in (None, ""):
+        return None
+    try:
+        return float(value)
+    except (TypeError, ValueError):
+        return None
diff --git a/backend/data/datahub/errors.py b/backend/data/datahub/errors.py
new file mode 100644
index 0000000..ce51473
--- /dev/null
+++ b/backend/data/datahub/errors.py
@@ -0,0 +1,12 @@
+from __future__ import annotations
+
+
+class DatahubError(RuntimeError):
+    def __init__(self, code: str, message: str, status: int | None = None) -> None:
+        super().__init__(message)
+        self.code = code
+        self.message = message
+        self.status = status
+
+    def __str__(self) -> str:
+        return f"{self.code}: {self.message}"
diff --git a/backend/data/datahub/native.py b/backend/data/datahub/native.py
new file mode 100644
index 0000000..e68b7ef
--- /dev/null
+++ b/backend/data/datahub/native.py
@@ -0,0 +1,153 @@
+from __future__ import annotations
+
+from typing import Any
+
+from backend.data.numbers import finite_number
+
+
+AMOUNT_THOUSAND_YUAN = 1000.0
+AMOUNT_WAN_YUAN = 10000.0
+VOLUME_LOT = 100.0
+
+API_TO_DATASET = {
+    "trade_cal": "calendar",
+    "stock_basic": "stocks",
+    "daily": "daily",
+    "daily_basic": "valuation",
+    "index_daily": "index_daily",
+    "moneyflow": "moneyflow",
+    "stk_auction": "auction",
+}
+
+SCALE_FIELDS = {
+    "daily": {"vol": VOLUME_LOT, "amount": AMOUNT_THOUSAND_YUAN},
+    "index_daily": {"vol": VOLUME_LOT, "amount": AMOUNT_THOUSAND_YUAN},
+    "valuation": {"total_mv": AMOUNT_WAN_YUAN, "circ_mv": AMOUNT_WAN_YUAN},
+    "moneyflow": {
+        "buy_sm_amount": AMOUNT_WAN_YUAN,
+        "sell_sm_amount": AMOUNT_WAN_YUAN,
+        "buy_md_amount": AMOUNT_WAN_YUAN,
+        "sell_md_amount": AMOUNT_WAN_YUAN,
+        "buy_lg_amount": AMOUNT_WAN_YUAN,
+        "sell_lg_amount": AMOUNT_WAN_YUAN,
+        "buy_elg_amount": AMOUNT_WAN_YUAN,
+        "sell_elg_amount": AMOUNT_WAN_YUAN,
+        "net_mf_amount": AMOUNT_WAN_YUAN,
+    },
+    "auction": {"vol": VOLUME_LOT, "float_share": AMOUNT_WAN_YUAN},
+}
+
+
+def yyyymmdd(value: Any) -> str:
+    return str(value or "").replace("-", "")[:8]
+
+
+def to_native_rows(dataset: str, rows: list[dict[str, Any]]) -> list[dict[str, Any]]:
+    return [to_native_row(dataset, row) for row in rows]
+
+
+def to_native_row(dataset: str, row: dict[str, Any]) -> dict[str, Any]:
+    if dataset == "calendar":
+        is_open = row.get("is_open")
+        return {
+            "exchange": str(row.get("exchange") or "SSE"),
+            "cal_date": yyyymmdd(row.get("cal_date")),
+            "is_open": 1 if is_open in (True, 1, "1", "Y", "y") else 0,
+            "pretrade_date": yyyymmdd(row.get("pretrade_date")) or None,
+        }
+    converted = dict(row)
+    converted.pop("batch_id", None)
+    if "volume" in converted and "vol" not in converted:
+        converted["vol"] = converted.pop("volume")
+    elif "volume" in converted:
+        converted.pop("volume", None)
+    scales = SCALE_FIELDS.get(dataset) or {}
+    for field, factor in scales.items():
+        if field in converted:
+            converted[field] = _unscale(converted.get(field), factor)
+    if dataset == "stocks":
+        converted.pop("updated_at", None)
+    return converted
+
+
+def to_canonical_row(dataset: str, row: dict[str, Any]) -> dict[str, Any]:
+    if dataset == "calendar":
+        is_open = row.get("is_open")
+        return {
+            "exchange": str(row.get("exchange") or "SSE"),
+            "cal_date": yyyymmdd(row.get("cal_date")),
+            "is_open": 1 if is_open in (True, 1, "1", "Y", "y") else 0,
+            "pretrade_date": yyyymmdd(row.get("pretrade_date")) or None,
+        }
+    converted = dict(row)
+    if "volume" in converted and "vol" not in converted:
+        converted["vol"] = converted.pop("volume")
+    scales = SCALE_FIELDS.get(dataset) or {}
+    for field, factor in scales.items():
+        if field in converted:
+            converted[field] = _scale(converted.get(field), factor)
+    return converted
+
+
+def row_key(dataset: str, row: dict[str, Any]) -> tuple[str, ...]:
+    if dataset == "calendar":
+        return (yyyymmdd(row.get("cal_date")),)
+    if dataset == "stocks":
+        return (str(row.get("ts_code") or "").upper(),)
+    if dataset == "status":
+        return (str(row.get("dataset") or ""), yyyymmdd(row.get("trade_date")))
+    return (str(row.get("ts_code") or "").upper(), yyyymmdd(row.get("trade_date")))
+
+
+def project_fields(rows: list[dict[str, Any]], fields: str) -> list[dict[str, Any]]:
+    keys = [item.strip() for item in str(fields or "").split(",") if item.strip()]
+    if not keys:
+        return rows
+    return [{key: row.get(key) for key in keys} for row in rows]
+
+
+def filter_stock_rows(rows: list[dict[str, Any]], params: dict[str, Any] | None) -> list[dict[str, Any]]:
+    payload = params or {}
+    ts_code = str(payload.get("ts_code") or "").strip().upper()
+    status = str(payload.get("list_status") or "").strip()
+    name = str(payload.get("name") or "").strip()
+    filtered = rows
+    if ts_code:
+        filtered = [row for row in filtered if str(row.get("ts_code") or "").upper() == ts_code]
+    if status:
+        filtered = [row for row in filtered if str(row.get("list_status") or status) == status]
+    if name:
+        filtered = [row for row in filtered if name.casefold() in str(row.get("name") or "").casefold()]
+    return filtered
+
+
+def filter_calendar_rows(rows: list[dict[str, Any]], params: dict[str, Any] | None) -> list[dict[str, Any]]:
+    payload = params or {}
+    if payload.get("is_open") in (1, "1", True):
+        return [row for row in rows if int(row.get("is_open") or 0) == 1]
+    if payload.get("is_open") in (0, "0", False):
+        return [row for row in rows if int(row.get("is_open") or 0) == 0]
+    return rows
+
+
+def _scale(value: Any, factor: float) -> float | None:
+    number = _optional_number(value)
+    if number is None:
+        return None
+    return number * factor
+
+
+def _unscale(value: Any, factor: float) -> float | None:
+    number = _optional_number(value)
+    if number is None or factor == 0:
+        return None
+    return number / factor
+
+
+def _optional_number(value: Any) -> float | None:
+    if value in (None, ""):
+        return None
+    number = finite_number(value, default=float("nan"))
+    if number != number:
+        return None
+    return number
diff --git a/backend/data/datahub/redact.py b/backend/data/datahub/redact.py
new file mode 100644
index 0000000..2b1d4ed
--- /dev/null
+++ b/backend/data/datahub/redact.py
@@ -0,0 +1,39 @@
+from __future__ import annotations
+
+from typing import Any
+
+
+SECRET_HINTS = (
+    "token",
+    "password",
+    "secret",
+    "key",
+    "authorization",
+    "credential",
+    "cookie",
+)
+
+
+def redact_value(value: Any, key: str = "", secrets: tuple[str, ...] = ()) -> Any:
+    lowered = key.lower()
+    if any(part in lowered for part in SECRET_HINTS):
+        return "***"
+    if isinstance(value, dict):
+        return {
+            str(item_key): redact_value(item_value, str(item_key), secrets)
+            for item_key, item_value in value.items()
+        }
+    if isinstance(value, list):
+        return [redact_value(item, key, secrets) for item in value]
+    text = str(value) if value is not None and not isinstance(value, (int, float, bool)) else value
+    if isinstance(text, str):
+        return redact_text(text, secrets)
+    return value
+
+
+def redact_text(text: str, secrets: tuple[str, ...] = ()) -> str:
+    redacted = text
+    for secret in secrets:
+        if secret:
+            redacted = redacted.replace(secret, "***")
+    return redacted
diff --git a/backend/data/datahub/settings.py b/backend/data/datahub/settings.py
new file mode 100644
index 0000000..419b3c7
--- /dev/null
+++ b/backend/data/datahub/settings.py
@@ -0,0 +1,120 @@
+from __future__ import annotations
+
+import json
+import os
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Any, Mapping
+
+from backend.bootstrap.config import APP_DIR
+
+
+DATASETS = (
+    "calendar",
+    "stocks",
+    "daily",
+    "index_daily",
+    "valuation",
+    "moneyflow",
+    "auction",
+    "status",
+)
+
+ENV_DATASET = {
+    "calendar": "CALENDAR",
+    "stocks": "STOCKS",
+    "daily": "DAILY",
+    "index_daily": "INDEX_DAILY",
+    "valuation": "VALUATION",
+    "moneyflow": "MONEYFLOW",
+    "auction": "AUCTION",
+    "status": "STATUS",
+}
+
+DEFAULT_CONFIG_PATH = APP_DIR / "config" / "datahub.config.json"
+
+
+def _truthy(value: Any) -> bool:
+    return str(value or "").strip().lower() in {"1", "true", "yes", "on"}
+
+
+def _int(value: Any, default: int) -> int:
+    try:
+        return int(value)
+    except (TypeError, ValueError):
+        return default
+
+
+@dataclass(frozen=True)
+class DatasetFlags:
+    name: str
+    read: bool = False
+    shadow: bool = False
+
+
+@dataclass(frozen=True)
+class DatahubSettings:
+    base_url: str
+    token: str
+    timeout_seconds: int = 8
+    retries: int = 1
+    page_limit: int = 5000
+    stale_seconds_max: int = 86400
+    datasets: dict[str, DatasetFlags] | None = None
+
+    def flags(self, dataset: str) -> DatasetFlags:
+        mapped = self.datasets or {}
+        return mapped.get(dataset) or DatasetFlags(dataset)
+
+    def any_enabled(self) -> bool:
+        return any(item.read or item.shadow for item in (self.datasets or {}).values())
+
+    def secrets(self) -> tuple[str, ...]:
+        return tuple(item for item in (self.token,) if item)
+
+    @classmethod
+    def load(
+        cls,
+        path: Path | None = None,
+        environ: Mapping[str, str] | None = None,
+        credentials: Mapping[str, object] | None = None,
+    ) -> "DatahubSettings":
+        config_path = path or DEFAULT_CONFIG_PATH
+        payload: dict[str, Any] = {}
+        if config_path.is_file():
+            payload = json.loads(config_path.read_text(encoding="utf-8"))
+        env = dict(os.environ if environ is None else environ)
+        creds = dict(credentials or {})
+        dataset_flags: dict[str, DatasetFlags] = {}
+        raw_datasets = payload.get("datasets") or {}
+        for name in DATASETS:
+            item = raw_datasets.get(name) or {}
+            env_key = ENV_DATASET[name]
+            read = _truthy(env.get(f"DATAHUB_READ_{env_key}")) if f"DATAHUB_READ_{env_key}" in env else bool(item.get("read"))
+            shadow = (
+                _truthy(env.get(f"DATAHUB_SHADOW_{env_key}"))
+                if f"DATAHUB_SHADOW_{env_key}" in env
+                else bool(item.get("shadow"))
+            )
+            dataset_flags[name] = DatasetFlags(name, read=read, shadow=shadow)
+        token = str(
+            env.get("DATAHUB_TOKEN")
+            or creds.get("datahub_token")
+            or payload.get("token")
+            or ""
+        ).strip()
+        base_url = str(
+            env.get("DATAHUB_BASE_URL")
+            or creds.get("datahub_base_url")
+            or payload.get("base_url")
+            or "http://127.0.0.1:8766"
+        ).strip().rstrip("/")
+        return cls(
+            base_url=base_url,
+            token=token,
+            timeout_seconds=_int(env.get("DATAHUB_TIMEOUT") or payload.get("timeout_seconds"), 8),
+            retries=max(0, _int(env.get("DATAHUB_RETRIES") or payload.get("retries"), 1)),
+            page_limit=max(1, _int(payload.get("page_limit"), 5000)),
+            stale_seconds_max=max(0, _int(payload.get("stale_seconds_max"), 86400)),
+            datasets=dataset_flags,
+        )
diff --git a/backend/data/gateway.py b/backend/data/gateway.py
index fcd436c..ac577a2 100644
--- a/backend/data/gateway.py
+++ b/backend/data/gateway.py
@@ -3,8 +3,10 @@ from __future__ import annotations
 from collections.abc import Callable
 from dataclasses import dataclass
 from datetime import datetime
+from typing import Any
 
 from backend.data.contracts import DataUsage
+from backend.data.datahub import DatahubAwareTushareClient, DatahubBridge, DatahubClient, DatahubSettings
 from backend.data.policy import DataSourcePolicy
 from backend.data.providers import IfindProvider, TushareProvider
 from backend.data.quality import DataQualityGate, QualityEvidence, QualityReport
@@ -22,6 +24,7 @@ class DataGateway:
     ifind_provider: IfindProvider
     chart_data: MarketChartClient
     realtime_observer: WebRealtimeAggregator
+    datahub: DatahubBridge
 
     @property
     def ifind(self) -> IfindHttpClient:
@@ -34,7 +37,13 @@ class DataGateway:
     ) -> TushareClient:
         if dataset_id:
             self.policy.assert_allowed(dataset_id, "tushare", usage)
-        return self.tushare_provider.client()
+        return DatahubAwareTushareClient(self.tushare_provider.client(), self.datahub)
+
+    def dataset_status(self, trade_date: str) -> list[dict[str, Any]] | None:
+        return self.datahub.dataset_status(trade_date)
+
+    def batches(self, trade_date: str, dataset: str = "") -> list[dict[str, Any]] | None:
+        return self.datahub.batches(trade_date, dataset)
 
     def assert_source(self, dataset_id: str, provider_id: str, usage: DataUsage) -> None:
         self.policy.assert_allowed(dataset_id, provider_id, usage)
@@ -64,6 +73,7 @@ class DataGateway:
 def build_data_gateway(
     credentials: dict[str, object],
     tushare_token_supplier: Callable[[], str] | None = None,
+    datahub_settings: DatahubSettings | None = None,
 ) -> DataGateway:
     ifind = IfindHttpClient(
         str(credentials.get("ifind_refresh_token") or ""),
@@ -73,6 +83,8 @@ def build_data_gateway(
         lambda: str(credentials.get("tushare_token") or "")
     )
     policy = DataSourcePolicy.load()
+    settings = datahub_settings or DatahubSettings.load(credentials=credentials)
+    datahub_client = DatahubClient(settings)
     return DataGateway(
         policy=policy,
         quality=DataQualityGate.load(policy),
@@ -80,4 +92,5 @@ def build_data_gateway(
         ifind_provider=IfindProvider(ifind),
         chart_data=MarketChartClient(ifind, EastmoneyChartClient()),
         realtime_observer=WebRealtimeAggregator(),
+        datahub=DatahubBridge(settings, datahub_client),
     )
diff --git a/config/README.md b/config/README.md
index 5b352f6..9b9d834 100644
--- a/config/README.md
+++ b/config/README.md
@@ -12,6 +12,9 @@ These registries describe the approved product surface of the standalone applica
   providers, model entry points, CSS layers, and remaining code hotspots.
 - `data-fields.config.json`: canonical data products, provider eligibility, intended use, and
   known blocked datasets.
+- `datahub.config.json`: optional read-only client for `xiaobai-datahub`. Each dataset has its
+  own `read` / `shadow` flag, all default off. Environment variables `DATAHUB_READ_*` and
+  `DATAHUB_SHADOW_*` can override a single dataset without a master switch.
 - `data-quality.config.json`: freshness, coverage, units, adjustment, point-in-time, and
   fail-closed rules for every canonical data product.
 - `jobs.config.json`: background schedules, dependencies, lock keys, retry policy, timeouts,
diff --git a/config/architecture-inventory.json b/config/architecture-inventory.json
index 8f33db1..107c544 100644
--- a/config/architecture-inventory.json
+++ b/config/architecture-inventory.json
@@ -204,6 +204,11 @@
       "path": "backend/data/providers/tushare_client.py",
       "runtime_role": "stable client facade for primary deterministic market data"
     },
+    {
+      "provider": "datahub",
+      "path": "backend/data/datahub/client.py",
+      "runtime_role": "optional official EOD read path behind per-dataset flags"
+    },
     {
       "provider": "ifind",
       "path": "backend/data/providers/ifind_client.py",
@@ -278,6 +283,18 @@
       "owner": "backend/data/providers/tushare.py",
       "compatibility_fallback": "backend/features/market/service.py"
     },
+    {
+      "client": "DatahubClient",
+      "owner": "backend/data/gateway.py"
+    },
+    {
+      "client": "DatahubBridge",
+      "owner": "backend/data/gateway.py"
+    },
+    {
+      "client": "DatahubAwareTushareClient",
+      "owner": "backend/data/gateway.py"
+    },
     {
       "client": "IfindHttpClient",
       "owner": "backend/data/gateway.py"
diff --git a/config/data-fields.config.json b/config/data-fields.config.json
index 408f6f9..0c1bb1f 100644
--- a/config/data-fields.config.json
+++ b/config/data-fields.config.json
@@ -2,6 +2,7 @@
   "schema_version": 1,
   "providers": {
     "tushare": {"class": "licensed", "calculation_allowed": true},
+    "datahub": {"class": "licensed", "calculation_allowed": true},
     "ifind": {"class": "licensed", "calculation_allowed": true},
     "eastmoney": {"class": "public_web", "calculation_allowed": false},
     "tencent": {"class": "public_web", "calculation_allowed": false},
diff --git a/config/datahub.config.json b/config/datahub.config.json
new file mode 100644
index 0000000..1877fbc
--- /dev/null
+++ b/config/datahub.config.json
@@ -0,0 +1,18 @@
+{
+  "schema_version": 1,
+  "base_url": "http://127.0.0.1:8766",
+  "timeout_seconds": 8,
+  "retries": 1,
+  "page_limit": 5000,
+  "stale_seconds_max": 86400,
+  "datasets": {
+    "calendar": { "read": false, "shadow": false },
+    "stocks": { "read": false, "shadow": false },
+    "daily": { "read": false, "shadow": false },
+    "index_daily": { "read": false, "shadow": false },
+    "valuation": { "read": false, "shadow": false },
+    "moneyflow": { "read": false, "shadow": false },
+    "auction": { "read": false, "shadow": false },
+    "status": { "read": false, "shadow": false }
+  }
+}
diff --git a/tests/test_data_gateway.py b/tests/test_data_gateway.py
index 536a0b5..f3fcbb0 100644
--- a/tests/test_data_gateway.py
+++ b/tests/test_data_gateway.py
@@ -67,6 +67,9 @@ class DataGatewayTests(unittest.TestCase):
             "TushareClient": {"backend/features/market/service.py"},
             "TushareProvider": {"backend/data/gateway.py"},
             "WebRealtimeAggregator": {"backend/data/gateway.py"},
+            "DatahubClient": {"backend/data/gateway.py"},
+            "DatahubAwareTushareClient": {"backend/data/gateway.py"},
+            "DatahubBridge": {"backend/data/gateway.py"},
         }
         found = {name: set() for name in owners}
         for path in (root / "backend").rglob("*.py"):
diff --git a/tests/test_datahub_bridge.py b/tests/test_datahub_bridge.py
new file mode 100644
index 0000000..56f1f75
--- /dev/null
+++ b/tests/test_datahub_bridge.py
@@ -0,0 +1,254 @@
+from __future__ import annotations
+
+import ast
+import json
+import unittest
+from pathlib import Path
+from typing import Any
+
+from backend.data import build_data_gateway
+from backend.data.datahub.bridge import DatahubAwareTushareClient, DatahubBridge, looks_like_heaven
+from backend.data.datahub.client import DatahubClient, DatahubResponse
+from backend.data.datahub.compare import compare_rows
+from backend.data.datahub.errors import DatahubError
+from backend.data.datahub.native import to_canonical_row, to_native_row
+from backend.data.datahub.settings import DATASETS, DatahubSettings, DatasetFlags
+
+ROOT = Path(__file__).resolve().parents[1]
+TOKEN = "super-secret-datahub-token"
+
+LEGACY_DAILY = {
+    "ts_code": "600000.SH",
+    "trade_date": "20240902",
+    "open": 10.11,
+    "high": 10.25,
+    "low": 10.01,
+    "close": 10.20,
+    "pct_chg": 1.2345,
+    "vol": 1000.0,
+    "amount": 2000.0,
+}
+HUB_DAILY = {
+    "ts_code": "600000.SH",
+    "trade_date": "20240902",
+    "open": 10.11,
+    "high": 10.25,
+    "low": 10.01,
+    "close": 10.20,
+    "pct_chg": 1.2345,
+    "volume": 100000.0,
+    "amount": 2000000.0,
+}
+
+
+class FakeLegacy:
+    def __init__(self, rows: list[dict[str, Any]] | Exception | None = None) -> None:
+        self.token = "legacy-token"
+        self.timeout = 30
+        self.rows = [] if rows is None else rows
+        self.calls: list[tuple[str, dict[str, Any] | None, str]] = []
+
+    def query(self, api_name: str, params: dict[str, Any] | None = None, fields: str = "") -> list[dict[str, Any]]:
+        self.calls.append((api_name, params, fields))
+        if isinstance(self.rows, Exception):
+            raise self.rows
+        return [dict(row) for row in self.rows]
+
+
+class FakeClient(DatahubClient):
+    def __init__(self, error: DatahubError | None = None, response: DatahubResponse | None = None) -> None:
+        super().__init__(DatahubSettings(base_url="http://127.0.0.1:9", token=TOKEN))
+        self.error = error
+        self.response = response or DatahubResponse(
+            data=[dict(HUB_DAILY)],
+            meta={"tier": "official", "trade_date": "20240902", "stale": False, "staleness_seconds": 0},
+        )
+        self.paths: list[str] = []
+
+    def get(self, path: str, params: dict[str, Any] | None = None) -> DatahubResponse:
+        self.paths.append(path)
+        if TOKEN in json.dumps(params or {}) or TOKEN in path:
+            raise AssertionError("token leaked into url")
+        if self.error:
+            raise self.error
+        return self.response
+
+
+def flags(**enabled: tuple[bool, bool]) -> DatahubSettings:
+    datasets = {name: DatasetFlags(name) for name in DATASETS}
+    for name, pair in enabled.items():
+        datasets[name] = DatasetFlags(name, read=pair[0], shadow=pair[1])
+    return DatahubSettings(base_url="http://127.0.0.1:9", token=TOKEN, datasets=datasets)
+
+
+class DatahubBridgeTests(unittest.TestCase):
+    def test_default_config_keeps_legacy_and_does_not_call_datahub(self) -> None:
+        settings = DatahubSettings.load(environ={}, credentials={})
+        self.assertFalse(settings.any_enabled())
+        self.assertTrue(all(not settings.flags(name).read and not settings.flags(name).shadow for name in DATASETS))
+        client = FakeClient(error=DatahubError("INTERNAL", "should not be called"))
+        legacy = FakeLegacy([LEGACY_DAILY])
+        wrapped = DatahubAwareTushareClient(legacy, DatahubBridge(settings, client))
+        rows = wrapped.query("daily", {"trade_date": "20240902"}, "ts_code,close,vol,amount")
+        self.assertEqual(rows[0]["amount"], 2000.0)
+        self.assertEqual(client.paths, [])
+        self.assertEqual(len(legacy.calls), 1)
+
+    def test_each_dataset_has_independent_read_flag(self) -> None:
+        settings = flags(daily=(True, False), auction=(False, False))
+        self.assertTrue(settings.flags("daily").read)
+        self.assertFalse(settings.flags("auction").read)
+        self.assertFalse(any(settings.flags(name).read for name in DATASETS if name != "daily"))
+        source = (ROOT / "config" / "datahub.config.json").read_text(encoding="utf-8")
+        self.assertNotIn("master", source)
+        self.assertNotIn("DATAHUB_READ_ALL", source)
+
+    def test_read_flag_replaces_only_that_dataset_and_converts_units(self) -> None:
+        shadows: list[dict[str, Any]] = []
+        client = FakeClient()
+        legacy = FakeLegacy([LEGACY_DAILY])
+        wrapped = DatahubAwareTushareClient(
+            legacy,
+            DatahubBridge(flags(daily=(True, False)), client, shadow_sink=shadows.append),
+        )
+        rows = wrapped.query("daily", {"trade_date": "20240902"}, "ts_code,vol,amount")
+        self.assertEqual(rows[0]["vol"], 1000.0)
+        self.assertEqual(rows[0]["amount"], 2000.0)
+        self.assertEqual(legacy.calls, [])
+        self.assertEqual(client.paths, ["/v1/bars/daily"])
+        calendar_legacy = FakeLegacy([{"cal_date": "20240902", "is_open": 1}])
+        calendar_client = FakeClient(error=DatahubError("INTERNAL", "nope"))
+        calendar_wrapped = DatahubAwareTushareClient(
+            calendar_legacy,
+            DatahubBridge(flags(daily=(True, False)), calendar_client),
+        )
+        calendar = calendar_wrapped.query("trade_cal", {"start_date": "20240902", "end_date": "20240902"}, "")
+        self.assertEqual(calendar[0]["is_open"], 1)
+        self.assertEqual(calendar_client.paths, [])
+
+    def test_fallback_on_down_401_timeout_empty_unpublished_and_stale(self) -> None:
+        cases = [
+            DatahubError("UNAVAILABLE", "down"),
+            DatahubError("UNAUTHORIZED", "401"),
+            DatahubError("TIMEOUT", "late"),
+            DatahubError("EMPTY", "no rows"),
+            DatahubError("DATASET_NOT_PUBLISHED", "not ready"),
+            DatahubError("STALE", "old"),
+        ]
+        for error in cases:
+            with self.subTest(error=error.code):
+                if error.code == "EMPTY":
+                    client = FakeClient(response=DatahubResponse(data=[], meta={"stale": False, "staleness_seconds": 0}))
+                elif error.code == "STALE":
+                    client = FakeClient(response=DatahubResponse(
+                        data=[dict(HUB_DAILY)],
+                        meta={"stale": True, "staleness_seconds": 999999},
+                    ))
+                else:
+                    client = FakeClient(error=error)
+                legacy = FakeLegacy([LEGACY_DAILY])
+                wrapped = DatahubAwareTushareClient(legacy, DatahubBridge(flags(daily=(True, False)), client))
+                rows = wrapped.query("daily", {"trade_date": "20240902"}, "ts_code,amount")
+                self.assertEqual(rows[0]["amount"], 2000.0)
+                self.assertEqual(len(legacy.calls), 1)
+
+    def test_shadow_compares_without_replacing_and_survives_hub_failure(self) -> None:
+        reports: list[dict[str, Any]] = []
+        client = FakeClient()
+        legacy = FakeLegacy([LEGACY_DAILY])
+        wrapped = DatahubAwareTushareClient(
+            legacy,
+            DatahubBridge(flags(daily=(False, True)), client, shadow_sink=reports.append),
+        )
+        rows = wrapped.query("daily", {"trade_date": "20240902"}, "ts_code,amount,vol")
+        self.assertEqual(rows[0]["amount"], 2000.0)
+        self.assertEqual(len(legacy.calls), 1)
+        self.assertEqual(reports[0]["equal"], True)
+        self.assertEqual(reports[0]["matched"], 1)
+
+        failed = FakeClient(error=DatahubError("UNAVAILABLE", TOKEN))
+        fail_reports: list[dict[str, Any]] = []
+        fail_legacy = FakeLegacy([LEGACY_DAILY])
+        fail_wrapped = DatahubAwareTushareClient(
+            fail_legacy,
+            DatahubBridge(flags(daily=(False, True)), failed, shadow_sink=fail_reports.append),
+        )
+        again = fail_wrapped.query("daily", {"trade_date": "20240902"}, "amount")
+        self.assertEqual(again[0]["amount"], 2000.0)
+        self.assertTrue(fail_reports[0]["hub_error"])
+        self.assertNotIn(TOKEN, json.dumps(fail_reports[0]))
+
+    def test_compare_classifies_unit_conversion_missing_row_and_value_diff(self) -> None:
+        equal = compare_rows("daily", [LEGACY_DAILY], [HUB_DAILY], {"stale": False, "staleness_seconds": 0})
+        self.assertTrue(equal["equal"])
+        unit = compare_rows("daily", [LEGACY_DAILY], [{**HUB_DAILY, "amount": 2000.0, "volume": 1000.0}])
+        self.assertGreater(unit["unit_conversion_count"], 0)
+        missing = compare_rows("daily", [LEGACY_DAILY], [])
+        self.assertEqual(missing["missing_hub_count"], 1)
+        value = compare_rows("daily", [LEGACY_DAILY], [{**HUB_DAILY, "close": 99.0}])
+        self.assertEqual(value["value_diff_count"], 1)
+        skew = compare_rows("daily", [LEGACY_DAILY], [HUB_DAILY], {"stale": False, "staleness_seconds": 12})
+        self.assertTrue(skew["time_skew"])
+
+    def test_native_roundtrip_matches_known_scales(self) -> None:
+        native = to_native_row("daily", HUB_DAILY)
+        self.assertEqual(native["vol"], 1000.0)
+        self.assertEqual(native["amount"], 2000.0)
+        canonical = to_canonical_row("daily", native)
+        self.assertEqual(canonical["vol"], 100000.0)
+        self.assertEqual(canonical["amount"], 2000000.0)
+
+    def test_heaven_keeps_legacy_even_when_read_flag_is_on(self) -> None:
+        self.assertTrue(looks_like_heaven("backend.features.heaven.market_context", "backend/features/heaven/market_context.py"))
+        self.assertFalse(looks_like_heaven("backend.features.market.service", "backend/features/market/service.py"))
+        client = FakeClient()
+        legacy = FakeLegacy([LEGACY_DAILY])
+        wrapped = DatahubAwareTushareClient(
+            legacy,
+            DatahubBridge(flags(daily=(True, False)), client, heaven_guard=lambda: True),
+        )
+        rows = wrapped.query("daily", {"trade_date": "20240902"}, "amount")
+        self.assertEqual(rows[0]["amount"], 2000.0)
+        self.assertEqual(client.paths, [])
+
+    def test_status_flag_does_not_run_when_off_and_falls_back_when_on(self) -> None:
+        off = DatahubBridge(flags(), FakeClient(error=DatahubError("UNAVAILABLE", "down")))
+        self.assertIsNone(off.dataset_status("20240902"))
+        reports: list[dict[str, Any]] = []
+        failed = DatahubBridge(
+            flags(status=(True, True)),
+            FakeClient(error=DatahubError("UNAUTHORIZED", "nope")),
+            shadow_sink=reports.append,
+        )
+        self.assertIsNone(failed.dataset_status("20240902"))
+        self.assertTrue(reports[0]["hub_error"])
+        ok = DatahubBridge(
+            flags(status=(True, False)),
+            FakeClient(response=DatahubResponse(data=[{"dataset": "daily", "state": "published"}], meta={"stale": False, "staleness_seconds": 0})),
+        )
+        self.assertEqual(ok.dataset_status("20240902")[0]["state"], "published")
+
+    def test_default_gateway_wraps_tushare_without_calling_datahub(self) -> None:
+        gateway = build_data_gateway({}, datahub_settings=flags())
+        client = gateway.tushare()
+        self.assertIsInstance(client, DatahubAwareTushareClient)
+        self.assertFalse(gateway.datahub.settings.any_enabled())
+
+    def test_features_do_not_import_datahub_client(self) -> None:
+        violations = []
+        for path in (ROOT / "backend" / "features").rglob("*.py"):
+            tree = ast.parse(path.read_text(encoding="utf-8"), filename=str(path))
+            for node in ast.walk(tree):
+                names = []
+                if isinstance(node, ast.Import):
+                    names = [alias.name for alias in node.names]
+                elif isinstance(node, ast.ImportFrom) and node.module:
+                    names = [node.module]
+                for name in names:
+                    if "datahub" in name.split("."):
+                        violations.append(f"{path.relative_to(ROOT)} -> {name}")
+        self.assertEqual(violations, [])
+
+
+if __name__ == "__main__":
+    unittest.main()
diff --git a/tests/test_datahub_client.py b/tests/test_datahub_client.py
new file mode 100644
index 0000000..06493e3
--- /dev/null
+++ b/tests/test_datahub_client.py
@@ -0,0 +1,185 @@
+from __future__ import annotations
+
+import io
+import json
+import logging
+import threading
+import unittest
+from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
+from urllib.parse import parse_qs, urlparse
+
+from backend.data.datahub.client import DatahubClient
+from backend.data.datahub.errors import DatahubError
+from backend.data.datahub.redact import redact_text
+from backend.data.datahub.settings import DatahubSettings
+
+
+TOKEN = "super-secret-datahub-token"
+
+
+class FakeHubState:
+    def __init__(self) -> None:
+        self.mode = "ok"
+        self.hits = 0
+        self.paths: list[str] = []
+
+
+STATE = FakeHubState()
+
+
+class FakeHubHandler(BaseHTTPRequestHandler):
+    def log_message(self, format: str, *args: object) -> None:
+        return
+
+    def do_GET(self) -> None:  # noqa: N802
+        STATE.hits += 1
+        parsed = urlparse(self.path)
+        STATE.paths.append(parsed.path)
+        token = self.headers.get("X-Datahub-Token", "")
+        if STATE.mode == "timeout":
+            raise TimeoutError("simulated timeout")
+        if token != TOKEN:
+            self._json(401, {"error": {"code": "UNAUTHORIZED", "message": "missing or invalid X-Datahub-Token"}})
+            return
+        if STATE.mode == "unpublished":
+            self._json(404, {"error": {"code": "DATASET_NOT_PUBLISHED", "message": "daily 19990101 尚未发布", "expected_at": "15:05+08:00"}})
+            return
+        if STATE.mode == "empty":
+            self._json(200, {"schema_version": 1, "data": [], "meta": {"tier": "official", "stale": False, "staleness_seconds": 0}})
+            return
+        if STATE.mode == "stale":
+            self._json(200, {"schema_version": 1, "data": [{"ts_code": "600000.SH", "trade_date": "20240902", "close": 10.2, "volume": 100000, "amount": 2000000}], "meta": {"tier": "official", "stale": True, "staleness_seconds": 999999}})
+            return
+        if STATE.mode == "invalid":
+            self.send_response(200)
+            self.send_header("Content-Type", "application/json")
+            self.end_headers()
+            self.wfile.write(b"not-json")
+            return
+        if parsed.path == "/v1/health":
+            self._json(200, {"schema_version": 1, "data": {"status": "ok"}, "meta": {"tier": "official", "source": "datahub", "stale": False, "staleness_seconds": 0}})
+            return
+        if parsed.path == "/v1/calendar":
+            self._json(200, {"schema_version": 1, "data": [{"cal_date": "20240902", "is_open": True, "pretrade_date": "20240830"}], "meta": {"tier": "official", "trade_date": "20240902", "stale": False, "staleness_seconds": 0}})
+            return
+        if parsed.path == "/v1/bars/daily":
+            query = {key: values[-1] for key, values in parse_qs(parsed.query).items()}
+            self._json(200, {
+                "schema_version": 1,
+                "data": [{
+                    "ts_code": "600000.SH",
+                    "trade_date": query.get("date") or "20240902",
+                    "open": 10.11, "high": 10.25, "low": 10.01, "close": 10.20,
+                    "pct_chg": 1.2345, "volume": 100000.0, "amount": 2000000.0, "adj_factor": 1.1,
+                }],
+                "meta": {"tier": "official", "trade_date": "20240902", "stale": False, "staleness_seconds": 0, "source": "tushare:daily"},
+            })
+            return
+        if parsed.path == "/v1/datasets/status":
+            self._json(200, {"schema_version": 1, "data": [{"dataset": "daily", "state": "published", "trade_date": "20240902"}], "meta": {"tier": "official", "stale": False, "staleness_seconds": 0}})
+            return
+        self._json(400, {"error": {"code": "INVALID_ARGUMENT", "message": f"unknown endpoint: {parsed.path}"}})
+
+    def _json(self, status: int, payload: dict) -> None:
+        body = json.dumps(payload).encode("utf-8")
+        self.send_response(status)
+        self.send_header("Content-Type", "application/json; charset=utf-8")
+        self.send_header("Content-Length", str(len(body)))
+        self.end_headers()
+        self.wfile.write(body)
+
+
+class DatahubClientTests(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls) -> None:
+        cls.server = ThreadingHTTPServer(("127.0.0.1", 0), FakeHubHandler)
+        cls.thread = threading.Thread(target=cls.server.serve_forever, daemon=True)
+        cls.thread.start()
+        cls.base = f"http://127.0.0.1:{cls.server.server_address[1]}"
+
+    @classmethod
+    def tearDownClass(cls) -> None:
+        cls.server.shutdown()
+        cls.server.server_close()
+
+    def setUp(self) -> None:
+        STATE.mode = "ok"
+        STATE.hits = 0
+        STATE.paths = []
+        self.client = DatahubClient(DatahubSettings(base_url=self.base, token=TOKEN, retries=1, timeout_seconds=2))
+
+    def test_health_envelope(self) -> None:
+        response = self.client.health()
+        self.assertEqual(response.schema_version, 1)
+        self.assertEqual(response.data["status"], "ok")
+        self.assertIn("stale", response.meta)
+
+    def test_missing_and_bad_token_401(self) -> None:
+        missing = DatahubClient(DatahubSettings(base_url=self.base, token=""))
+        with self.assertRaises(DatahubError) as raised:
+            missing.health()
+        self.assertEqual(raised.exception.code, "NOT_CONFIGURED")
+        bad = DatahubClient(DatahubSettings(base_url=self.base, token="wrong"))
+        with self.assertRaises(DatahubError) as raised:
+            bad.health()
+        self.assertEqual(raised.exception.code, "UNAUTHORIZED")
+        self.assertNotIn(TOKEN, str(raised.exception))
+
+    def test_unpublished_and_empty_and_stale_codes(self) -> None:
+        STATE.mode = "unpublished"
+        with self.assertRaises(DatahubError) as raised:
+            self.client.daily_bars(date="19990101")
+        self.assertEqual(raised.exception.code, "DATASET_NOT_PUBLISHED")
+        STATE.mode = "empty"
+        response = self.client.daily_bars(date="20240902")
+        self.assertEqual(response.data, [])
+        STATE.mode = "stale"
+        stale = self.client.daily_bars(date="20240902")
+        self.assertTrue(stale.meta["stale"])
+
+    def test_invalid_json_maps_to_internal(self) -> None:
+        STATE.mode = "invalid"
+        with self.assertRaises(DatahubError) as raised:
+            self.client.health()
+        self.assertEqual(raised.exception.code, "INTERNAL")
+
+    def test_timeout_maps_and_retries(self) -> None:
+        hits = {"n": 0}
+
+        def boom(_request, timeout=None):
+            hits["n"] += 1
+            raise TimeoutError("late")
+
+        client = DatahubClient(
+            DatahubSettings(base_url=self.base, token=TOKEN, retries=1, timeout_seconds=1),
+            urlopen=boom,
+        )
+        with self.assertRaises(DatahubError) as raised:
+            client.health()
+        self.assertEqual(raised.exception.code, "TIMEOUT")
+        self.assertEqual(hits["n"], 2)
+
+    def test_token_never_appears_in_error_text_or_logs(self) -> None:
+        stream = io.StringIO()
+        logger = logging.getLogger("xiaobai.datahub")
+        handler = logging.StreamHandler(stream)
+        logger.addHandler(handler)
+        logger.setLevel(logging.DEBUG)
+        try:
+            with self.assertRaises(DatahubError):
+                DatahubClient(DatahubSettings(base_url=self.base, token="wrong")).health()
+            blob = stream.getvalue() + redact_text("header " + TOKEN, (TOKEN,))
+            self.assertNotIn(TOKEN, blob)
+            self.assertIn("***", redact_text(TOKEN, (TOKEN,)))
+        finally:
+            logger.removeHandler(handler)
+
+    def test_calendar_and_status_contract(self) -> None:
+        calendar = self.client.calendar("20240901", "20240902")
+        self.assertEqual(calendar.data[0]["cal_date"], "20240902")
+        status = self.client.dataset_status("20240902")
+        self.assertEqual(status.data[0]["dataset"], "daily")
+
+
+if __name__ == "__main__":
+    unittest.main()
diff --git a/tools/build_architecture_inventory.py b/tools/build_architecture_inventory.py
index 2bd5c75..32c4bee 100644
--- a/tools/build_architecture_inventory.py
+++ b/tools/build_architecture_inventory.py
@@ -217,6 +217,7 @@ def build() -> dict[str, Any]:
         ),
         "external_data_adapters": [
             {"provider": "tushare", "path": "backend/data/providers/tushare_client.py", "runtime_role": "stable client facade for primary deterministic market data"},
+            {"provider": "datahub", "path": "backend/data/datahub/client.py", "runtime_role": "optional official EOD read path behind per-dataset flags"},
             {"provider": "ifind", "path": "backend/data/providers/ifind_client.py", "runtime_role": "realtime, charts, snapshots, enrichment"},
             {"provider": "eastmoney", "path": "backend/features/market/charts.py", "runtime_role": "display chart fallback"},
             {"provider": "eastmoney", "path": "backend/data/realtime.py", "runtime_role": "isolated realtime observation"},
@@ -235,6 +236,9 @@ def build() -> dict[str, Any]:
         ],
         "provider_construction": [
             {"client": "TushareClient", "owner": "backend/data/providers/tushare.py", "compatibility_fallback": "backend/features/market/service.py"},
+            {"client": "DatahubClient", "owner": "backend/data/gateway.py"},
+            {"client": "DatahubBridge", "owner": "backend/data/gateway.py"},
+            {"client": "DatahubAwareTushareClient", "owner": "backend/data/gateway.py"},
             {"client": "IfindHttpClient", "owner": "backend/data/gateway.py"},
             {"client": "MarketChartClient", "owner": "backend/data/gateway.py"},
             {"client": "WebRealtimeAggregator", "owner": "backend/data/gateway.py"},