diff --git a/urls_automated_cdx_processor.py b/urls_automated_cdx_processor.py
index b92a35c..e89a165 100644
--- a/urls_automated_cdx_processor.py
+++ b/urls_automated_cdx_processor.py
@@ -35,6 +35,19 @@ def run_cdxsummary(file_path, json_filepath):
         # Parse the JSON output
         json_output = json.loads(result.stdout)
 
+        # Add "cdxcount" entry with value 1
+        json_output["cdxcount"] = 1
+
+        # Add "cdxsize" entry with the size of the cdx.gz file in bytes
+        cdx_size_bytes = os.path.getsize(file_path)
+        json_output["cdxsize"] = cdx_size_bytes
+
+        # Remove "pathquery" and "samples" keys
+        if "pathquery" in json_output:
+            del json_output["pathquery"]
+        if "samples" in json_output:
+            del json_output["samples"]
+
         # Process the tophosts section
         if "tophosts" in json_output:
             processed_tophosts = process_tophosts(json_output["tophosts"])