We have several workflows which download files from AWS S3 for processing by n8n. When viewing the output of the workflow, n8n gives the ability to download the processed file directly from S3 to your workstation via a ‘download’ button in the n8n UI (see screenshot). For security reasons, this is undesirable for us since there are people in our organisation who have access to n8n who should not have access to S3 buckets, therefore the n8n UI is circumventing our security policy. Is there anyway to prevent the AWS node in n8n from displaying this ‘download’ button?
Hi @David_Andrews i do not think so we could do that, as it is more of a feature, i think what you could do to is to either dont let any execution getting saved with N8N_EXECUTIONS_DATA_SAVE=none and i think that would not be viable as it would stop all the flows from not getting logged, so i think this should be a feature as if someone wants to disable files being shown and taking persistent storage to just turn that off.
The N8N_EXECUTIONS_DATA_SAVE=none workaround disables all audit logs, which isn’t practical. Have you tried restricting S3 access via IAM instead? Scope your n8n execution role’s policy to only the specific buckets or prefixes your workflows actually need—this keeps logging intact while preventing unrestricted UI downloads.
Hi, thanks for the reply. The IAM role used by n8n to access S3 is restricted to just a single bucket, however the data within this bucket is sensitive. N8n requires access to all files within the bucket in order to process the files, however some of the people who have access to this n8n project should not have the ability to view the files in the S3 bucket. If there’s no way to restrict access to this ‘download’ button in the UI without disabling all audit logs then we’ll have to rethink how we pass the data from S3 to n8n to meet this security requirement. Appreciate your suggestions though.
hi @David_Andrews
I think this can be addressed from a process design perspective by splitting it into two workflows: one workflow handles S3 access and processing, while the other consumes the result without exposing binary data. Another option would be to process the file internally and remove the binary data before the workflow finishes.