Hi
One area where I think the platform could be strengthened is around the lifecycle management and governance of security-sensitive administrative decisions.
Today, in most stacks, certain capabilities such as exports, creating third-party application registries, egress policies that allow integration with external systems are managed by admins and used based on an organization’s risk appetite. However, what seems to be missing is ongoing checking after those decisions have been made. Here’s a few examples that come to my mind:
-
An export is a risk and may have been approved based on a specific assessment and business justification. Over time, the data connection configuration (agents, or egress proxy configuration) could change - or even the external source system configuration - changing the risk profile and allowing the original approval to be used in ways that were never part of the initial assessment.
-
Exports with marking validations in REST API data connections must be enabled for functions that need to call external systems. This is currently a platform limitation that we need to continuously make sure it’s not misused.
-
Egress policies may be granted for a legitimate integration, but months later the integration may no longer be in use, traffic may have stopped entirely, or the team changed. Eventually, the trend is for the accumulation of egress policies.
-
Third-party integrations or other privileged configurations may remain enabled indefinitely without any visibility into whether they are still required, actively used, or should be revalidated. The trend is for the accumulation of test TPAs that in the case of unscoped users have no idea whether it’s being used.
The platform provides audit and egress policy observability logs (recent) that partially address these questions, but they don’t surface the information to both devs and admins that enable these components and need to monitor the lifecycle of these decisions, identify unused privileged capabilities dangling around, or potentially prompt periodic reviews. Use cases want these doors to be opened and reduce time-to-market, but as deployments grow in size and complexity, the risks accumulate. There is no platform-embedded counterforce thats surfaces which doors can be potentially closed.
Is it something that is being considered for future roadmap work?