diff --git a/examples/code_execution_file_download.py b/examples/code_execution_file_download.py
new file mode 100755
index 00000000..d2220a75
--- /dev/null
+++ b/examples/code_execution_file_download.py
@@ -0,0 +1,79 @@
+#!/usr/bin/env -S uv run python
+from __future__ import annotations
+
+from anthropic import Anthropic
+
+client = Anthropic()
+
+
+def main() -> None:
+    # Ask Claude to generate a CSV file using the code execution server tool.
+    message = client.messages.create(
+        model="claude-opus-4-6",
+        max_tokens=4096,
+        messages=[
+            {
+                "role": "user",
+                "content": (
+                    "Use Python to create a CSV file at /output/report.csv. "
+                    "Include a header row (name,age,city) and 3 sample rows of "
+                    "data, then print 'Done' when finished."
+                ),
+            }
+        ],
+        tools=[{"type": "code_execution_20250825", "name": "code_execution"}],
+    )
+
+    # Show every block type that came back for reference.
+    print("Response block types:")
+    for block in message.content:
+        print(f"  {block.type}")
+
+    # Extract the file_id from the code execution result.
+    #
+    # NOTE — the following pattern shown in some documentation is INCORRECT
+    # and will never match when using server-side tools like code_execution:
+    #
+    #   if block.type == "tool_use" and block.name == "code_execution":
+    #       for result_block in block.content:
+    #           file_id = result_block.file_id          # never reached
+    #
+    # Server-side tools (code_execution, web_search, etc.) produce a
+    # "server_tool_use" block for the invocation, not "tool_use".
+    # The generated file arrives in a separate block:
+    #
+    #   block                            type = "bash_code_execution_tool_result"
+    #     .content                       BashCodeExecutionResultBlock
+    #       .content[i]                  BashCodeExecutionOutputBlock
+    #         .type == "bash_code_execution_output"
+    #         .file_id                   <-- the ID to pass to files.download()
+
+    file_id: str | None = None
+
+    for block in message.content:
+        if block.type == "bash_code_execution_tool_result":
+            result = block.content  # BashCodeExecutionResultBlock | BashCodeExecutionToolResultError
+            if result.type == "bash_code_execution_result":
+                for output in result.content:  # List[BashCodeExecutionOutputBlock]
+                    if output.type == "bash_code_execution_output" and output.file_id:
+                        file_id = output.file_id
+                        break
+            break
+
+    if file_id is None:
+        print("\nNo file output found in the response.")
+        return
+
+    print(f"\nFound file_id: {file_id}")
+
+    # Download the generated file and save it locally.
+    file_data = client.beta.files.download(file_id)
+    output_path = "report.csv"
+    with open(output_path, "wb") as f:
+        f.write(file_data.read())
+
+    print(f"Saved to '{output_path}'.")
+
+
+if __name__ == "__main__":
+    main()