High-risk action gate
Evaluate selected actions before execution only through a customer-deployed protected boundary the agent cannot bypass. The public page does not block or execute anything.
Critical infrastructure and cyber
A response tool might isolate a system, change access or recommend a service shutdown. Forge checks selected actions against your supplied evidence and rules, and records what needs human review. Your systems keep execution authority. Forge does not operate safety interlocks or execute actions.
| Decision boundary | A defined action menu: proceed, hold, escalate, reject, require review, or a customer-defined equivalent. |
|---|---|
| Evidence trail | Alerts, control checks, incident evidence, asset context, vendor signals, approvals, policy constraints, and missing facts. |
| Human authority | The proof shows where a human reviewed, where review was required, and what the system was not allowed to do alone. |
| Calibration | Authorized outcomes can be retained for tenant-local calibration only with explicit retention opt-in. Without it, no calibration row is written. |
Evaluate selected actions before execution only through a customer-deployed protected boundary the agent cannot bypass. The public page does not block or execute anything.
Turn deployment, resilience, insurability, or readiness evidence into a proof trail leadership can review.
Show why a supplier, control gap, or compensating control cleared, held, or required escalation.
Forge can start on one cyber, infrastructure, or vendor-risk workflow with synthetic, sanitized, or approved evidence and test the proof path before deeper integration. Production use requires workflow, safety, security, deployment, and operational qualification.