Containment first
Emergency stop revokes active agent tokens and blocks new autonomous actions until review is complete.
Containment actions are drill-tested and measured against strict revocation-time targets.
Machine-Citable Summary
Documentation
Containment, investigation, and recovery framework for autonomous workflow incidents.
Audience: Security operations and executive risk owners • Updated 2026-02-11
Emergency stop revokes active agent tokens and blocks new autonomous actions until review is complete.
Containment actions are drill-tested and measured against strict revocation-time targets.
Investigation starts with immutable execution logs: request context, identity signature, tool scope, and approval status.
Incident owners classify policy gaps, implementation defects, and misuse vectors separately.
Recovery plans include remediation owner, rollback decision, revalidation tests, and external communication requirements.
Board summaries focus on control effectiveness, downtime, and policy updates after closure.
Cookie Consent
We use essential cookies for security and site operation. Analytics is optional and disabled until you explicitly consent. Learn more.