RobbclanSafe Harbor / Agent observability & audit
Understand agent activity.
Review observed agent activity, follow execution traces, and account for model usage in one place.
From activity to clarity.
Give organization administrators a place to investigate the agent activity the platform observes, understand its context, and review usage.
Observe
See recorded agent connections and platform tool activity, attributed to the account or credential involved.
Investigate
Follow collected execution traces and audit events. Review source references, outcomes, and recorded transfer metadata.
Account for usage
Review reported model usage and estimated costs using configured rates. Compare activity across collected sources.
Know what is covered.
Agent activity and audit
Review observed Robbclan activity and ingested events in a tenant-scoped workspace. Filter records to investigate an agent or execution.
Transfer metadata
Inspect recorded document references, link domains, and prepared request or response sizes. Metadata provides context without recording full content by default.
Cross-suite event collection
Use authenticated event ingestion to collect compatible external events. ServiceNow, Salesforce, and SAP demonstrations currently use simulators; live connectors are not included.
Tenant model configuration
Tenant administrators can configure cloud or private DMZ model endpoints across Robbclan. Private security-analysis routing can be configured; automated security analysis is not yet available.
Coverage is limited to instrumented activity and submitted events. Safe Harbor does not discover every external agent or inspect downstream forwarding. Cost estimates are not provider invoices.
Make agent activity easier to understand.
Available to authorized organization administrators.