The request log redacts before it writes

Sensitive headers and body fields are redacted by default at capture time, before the entry hits disk. Reads through the container's logs URL apply the redaction again regardless of config. The store lives on the server that runs the container, bounded by row, age, and size caps, and every read of it is itself recorded.

The old stack was a logger, a shipper, a SaaS keeping raw bodies in another region, and a scrubbing job you meant to write. One HTTP gateway means one capture point and one redaction policy.

The log is on your server, it's scrubbed at capture, and reading it leaves a record of its own.