Recent findings reveal that over 20% of managed cloud platform access policies are either broken or missing. This creates audit silos where actions taken by automation agents cannot be reliably traced back to the actual requesters or approvers, raising concerns for cloud reliability and compliance.
- 20%+ of access policies are misconfigured or missing across reviewed MCP environments
- Agent actions logged under tokens or service accounts without direct user attribution
- Integrating agent activity with identity provider records increases traceability and security
Infrastructure signal
A significant portion of managed cloud platform infrastructures operate with incomplete or flawed access policies. Agents often act under personal tokens or shared service accounts, which are not consistently linked to individual users in audit logs. This pattern means audit trails only show which token performed the workânot who requested or approved it, creating gaps in visibility across cloud infrastructure operations.
Such gaps increase cloud environment risk by allowing untracked credential inheritance and emergent role assumptions by automation agents. They obscure the provenance of actions in event logs, which impacts incident investigation, regulatory compliance, and cloud security posture management.
Developer impact
Without tight integration between identity providers and agent permissions, developers may also face difficulties enforcing least privilege principles. They risk inadvertently granting agents broader access than necessary, which challenges secure coding practices and developer workflow hygiene.