The UI cannot recover a complete response here because the terminating chunk never arrived. You can still keep every body byte mitmproxy received before that error by forcing this one response into streaming mode and copying the chunks to a file:

from pathlib import Path

writers = {}

def responseheaders(flow):
    if "/the/specific/path" not in flow.request.path:
        return

    Path("partial-bodies").mkdir(exist_ok=True)
    writer = open(f"partial-bodies/{flow.id}.body", "wb", buffering=0)
    writers[flow.id] = writer

    def capture(chunk):
        writer.write(chunk)
        return chunk  # keep forwarding it to the client

    flow.response.stream = capture

def close_writer(flow):
    writer = writers.pop(flow.id, None)
    if writer:
        writer.close()

response = close_writer
error = close_writer

Run it with:

mitmdump -s capture_partial.py

Replace the path check with something that uniquely identifies the request. In v12.2.3, the HTTP layer calls a streaming callback for each ResponseData chunk and forwards the bytes it returns. The error hook closes the file when the incomplete chunked read eventually occurs. This is the same basic mechanism used by the repository’s save_streamed_data.py example.

The saved file is the de-chunked response body received so far. It may still be gzip/Brotli encoded according to Content-Encoding; it is not the original wire framing, and it cannot contain the missing tail. Wireshark remains the right tool if you need the exact on-the-wire bytes.