Oso

Oso is a security platform for operating AI coding agents and agentic systems safely. It provides visibility, controls, alerting, auditing, and authorization for agents such as Claude Code, Cursor, Codex, and other assistants used by engineering teams. The platform observes agent execution in real time by recording each prompt, tool invocation, command, and response. Activity is aggregated into monitoring dashboards that show frequency, severity, and threshold-based alerts so teams can evaluate risk across users and agents. Oso enforces behavioral boundaries through configurable rules. Organizations can block actions such as sensitive data exfiltration, infrastructure modification, or violations of custom policy. Built-in content detection identifies regulated data types including phone numbers, email addresses, and more. All activity is written to an immutable audit trail containing every action, tool call, and model response, enabling compliance and forensic review. In addition, Oso performs authorization for agentic workflows. Fine-grained, least-privilege access checks are evaluated inside API calls to determine what an agent is allowed to do based on user permissions and data location, and the system works across deployment environments.