diff --git a/README.md b/README.md
index 3ce886b..2b0af00 100644
--- a/README.md
+++ b/README.md
@@ -1110,6 +1110,11 @@ Legacy MinIO object layout (relative key):
 `joblib`. The flavor is configuration, never discovered from the artifact, and there is
 no fallback between flavors: a failing loader propagates.
 
+`model_config` also carries ad-hoc behavioral flags, unrelated to flavor, e.g.
+`skip_predict_timestamp` (default `False`): when `True`, `request_predict` does not reinsert the
+`timestamp` column derived from the index into the transformed data before calling predict. Full
+list: [docs/model-flavors.md § Other `model_config` flags](docs/model-flavors.md#other-model_config-flags).
+
 `joblib` reads the artifact directly — download, `<artifact>/model.pkl`, `joblib.load` — and writes on
 retrain (`model.pkl` + `code/`, no `MLmodel`, so only this runtime reads it back).
 
diff --git a/docs/model-flavors.md b/docs/model-flavors.md
index 9efd64c..6b70ca4 100644
--- a/docs/model-flavors.md
+++ b/docs/model-flavors.md
@@ -18,6 +18,16 @@ reads the artifact's `MLmodel` manifest to guess it.
 `predict_flavor` drives `load_predict_model` (artifact `prediction_model`); `transform_flavor` drives
 `load_transform_model` (artifact `data_model`). Both default to `sklearn` when the key is absent.
 
+## Other `model_config` flags
+
+Ad-hoc behavioral overrides, read directly with `.get(key, default)` — not part of the model
+document written by import (`model-import.md` § 12), and not validated against a schema:
+
+| Key | Default | Effect |
+|---|---|---|
+| `skip_transform` | `False` | `fit_models` (retrain) skips calling `transform` before fitting. |
+| `skip_predict_timestamp` | `False` | `request_predict` skips reinserting the `timestamp` column (from the index) into the transformed data before calling predict. |
+
 ## Accepted values
 
 | Flavor | Read | Write (retrain) | How it loads |
diff --git a/laborious/activities/mlflow.py b/laborious/activities/mlflow.py
index f4d42a3..d075508 100644
--- a/laborious/activities/mlflow.py
+++ b/laborious/activities/mlflow.py
@@ -237,6 +237,9 @@ class MLFlow(MinioManager):
                     - data (dict): Transformed data for prediction
                     - model_name (str): Name of the MLFlow model to use
                     - model_retention (int): Model retention period in minutes
+                Optional keys (in model_config):
+                    - skip_predict_timestamp (bool): If True, don't reinsert the
+                      timestamp column from the index before predicting. Default False.
 
         Returns:
             dict: Prediction results from MLFlow model
@@ -258,10 +261,11 @@ class MLFlow(MinioManager):
         # Convert numpy.nan to None for model compatibility
         data.replace(np.nan, None, inplace=True)
 
-        data['timestamp'] = data.index
-        data['timestamp'] = to_datetime(
-            data['timestamp'], format=DATETIME_FORMAT_WITH_TZ
-        ).dt.strftime(DATETIME_FORMAT)
+        if not model_config.get('skip_predict_timestamp', False):
+            data['timestamp'] = data.index
+            data['timestamp'] = to_datetime(
+                data['timestamp'], format=DATETIME_FORMAT_WITH_TZ
+            ).dt.strftime(DATETIME_FORMAT)
 
         # Request prediction from MLFlow model
         response_data = await self.model_monitoring_repository.predict(