diff --git a/rollout_data_analysis/src/rollout_data_analysis/cli.py b/rollout_data_analysis/src/rollout_data_analysis/cli.py
new file mode 100644
index 0000000..9ed3d09
--- /dev/null
+++ b/rollout_data_analysis/src/rollout_data_analysis/cli.py
@@ -0,0 +1,80 @@
+import argparse
+import json
+from pathlib import Path
+from typing import Any
+import pandas as pd
+
+from rollout_data_analysis import core
+
+
+class CLIError(Exception):
+    """Custom exception for CLI errors."""
+    pass
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(
+        description="Analysewerkzeug zur statistischen Auswertung von Rollout-Daten."
+    )
+    parser.add_argument(
+        "--input", required=True, type=str, help="Pfad zur CSV-Datei mit Rollout-Daten"
+    )
+    parser.add_argument(
+        "--whitelist", required=False, type=str, help="Optionaler Pfad zur Whitelist-Datei im JSON-Format"
+    )
+    parser.add_argument(
+        "--output", required=True, type=str, help="Pfad zur Ausgabe der Analyseergebnisse im JSON-Format"
+    )
+    return parser.parse_args()
+
+
+def load_rollout_data(csv_path: Path) -> list[dict[str, Any]]:
+    if not csv_path.exists():
+        raise CLIError(f"Eingabedatei nicht gefunden: {csv_path}")
+
+    df = pd.read_csv(csv_path)
+    required_columns = {"run_id", "unknown_rate", "warn_rate", "policy_hash", "pinned"}
+    missing = required_columns - set(df.columns)
+    if missing:
+        raise CLIError(f"Fehlende Spalten in CSV: {', '.join(missing)}")
+
+    data = df.to_dict(orient="records")
+    assert isinstance(data, list) and all(isinstance(row, dict) for row in data)
+    return data
+
+
+def save_results(output_path: Path, results: Any) -> None:
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+    with output_path.open("w", encoding="utf-8") as f:
+        json.dump(results.__dict__, f, indent=2, ensure_ascii=False)
+
+
+def main() -> None:
+    args = parse_args()
+
+    csv_path = Path(args.input)
+    output_path = Path(args.output)
+
+    rollout_data = load_rollout_data(csv_path)
+
+    # analyze_data expected input: list[dict]
+    analysis_results = core.analyze_data(rollout_data)
+
+    save_results(output_path, analysis_results)
+    print(f"Analyse abgeschlossen. Ergebnisse gespeichert unter: {output_path}")
+
+
+if __name__ == "__main__":
+    try:
+        main()
+    except CLIError as err:
+        import sys
+
+        print(f"[Fehler] {err}", file=sys.stderr)
+        sys.exit(1)
+    except Exception as exc:
+        import sys
+        import traceback
+
+        traceback.print_exc(file=sys.stderr)
+        sys.exit(2)