What high-level problem are you trying to solve?
When using S3 compatible storage to save attachments, it might be useful at times to directly serve files from the s3 bucket, for instance when publishing results on a third-party platform. It is currently possible to retrieve S3 objects keys corresponding to the attachment filenames only by directly querying the ODK central database.
Any ideas on how ODK could help you solve it?
The simplest would be to use the existing export/sync functionality and, for every media/attachment field, simply return the object keys as an extra column along with filenames both in the API responses or csv downloads. S3 permissions management would be left entirely to the user; this feature would only provide the objects keys.
Upload any helpful links, sketches, and videos.