Securing Claude Code: The New Compliance API, Local Visibility, and Identity Governance

Anthropic has introduced new Compliance API endpoints that provide security teams with enhanced visibility into the activities of Claude Code, a local AI agent running on developers' machines. These endpoints help monitor local session transcripts but reveal that activity logs alone cannot fully determine the legitimacy of agent access. Claude Code operates as a harness executing commands locally, requiring new security models to manage endpoint agents effectively.