Audit Trail

HelpDesk activity is captured in a pluggable audit log. This is a backend, ops-configured pipeline — there's no in-app screen for browsing audit records; instead, they land in whichever sink you configure and get reviewed there (e.g. an S3 bucket queried through Athena, CloudWatch, or wherever your log sink already ships to).

What's captured — every write request (POST/PUT/PATCH/DELETE) and every SignalR hub call, recording the method, path, status code, elapsed time, query string, action name, and a masked copy of the request payload, along with who made the call (effective email, authenticated email, workspace, admin flag, and any "on behalf of" identity for bot/Slack-driven actions). Auth failures and routing errors (401, 403, 404, 405) are captured too, even when they never reach an action — and if an action throws partway through, the event for it is still logged.

Sensitive fields are masked automatically. Any field whose name contains a keyword like secret, key, password, token, credentials, or similar is replaced with *** in the logged payload — no manual annotation needed for the common cases.

Configuring sinks — a logger-based sink is always on and flows through the platform's existing logging configuration. Shipping audit records to cloud storage instead (or as well) is opt-in, configured under an Audit section in the service's configuration:

• Audit:S3:Enabled, Audit:S3:BucketName, Audit:S3:Prefix, Audit:S3:Region

• Audit:AzureBlob:Enabled, Audit:AzureBlob:ConnectionString, Audit:AzureBlob:ContainerName, Audit:AzureBlob:Prefix

• Audit:Gcs:Enabled, Audit:Gcs:BucketName, Audit:Gcs:Prefix

All three cloud sinks are disabled by default. For S3, records are written as newline-delimited JSON, batched and partitioned by date (.../year=YYYY/month=MM/day=DD/...) so they can be queried directly with Athena without any repair step. Enabling the S3 sink requires the service's AWS credentials to have s3:PutObject on the target bucket/prefix.

This audit pipeline covers the AI Studio/HelpDesk API surface. It's a separate mechanism from the general DuploCloud infrastructure audit trail available elsewhere in the platform.