# AgentVeil Safe Mode

AgentVeil Safe Mode routes high-impact AI coding-agent actions through one
policy boundary before they write, push, publish, deploy, install, or run.

## What It Controls

- Configured routed actions from supported agent workflows
- Approval, block, and redirect decisions before execution
- Evidence records for controlled decisions
- Operator-visible boundaries for risky action categories

## Boundary

Configured-route control does not claim host-wide shell
interception, comprehensive sandboxing, or prevention across other
agent action paths.

## Canonical URLs

- Human page: https://agentveil.dev/safe-mode
- Home: https://agentveil.dev/
- Console: https://agentveil.dev/console/auth