PLATFORM
Runtime Guardrails.
Turn evaluation into control.
A runtime guardrail is useful when it changes what the system is allowed to do, not merely what a dashboard reports.
Decision modes
PASS
Allow
Action meets the configured evidence and policy threshold.
WARN
Proceed with signal
Action can proceed while preserving a warning or remediation.
ESCALATE
Require review
Route to human or higher-authority policy when confidence is insufficient.
BLOCK
Stop
Do not invoke the external tool until the reason is resolved.
Place guardrails at the tool boundary
The strongest runtime location is immediately before a consequential tool call because the trust layer can inspect the exact action and arguments that would reach the outside world.
Bounded claims
Guardrails reduce defined classes of risk; they do not guarantee perfect correctness, compliance or safety. The implemented checks and policy contract should always be inspectable.